The Authority Book

Most buying decisions are made in a room you are not in. A prospect hears your name from a colleague, looks you up on a Tuesday evening, reads whatever you have left lying about on the internet, and forms a view. By the time they book a call, the call is mostly a confirmation. You are not pitching; you are being checked against an impression that has already set.

A book is that reason. Not a book in the sense of a hardback sitting in an airport, though it can become one, but a structured, generous body of your thinking that a buyer can explore before they ever speak to you. A hundred entries, each one useful, each one in your voice, each one quietly demonstrating that you have been here before. The prospect reads three entries over coffee and recognises their own problem in the second. That recognition is the handshake. The meeting merely follows it.

You can test the principle this week without writing a word of the book. Look at the last five clients who signed with you quickly and ask what they had read, watched or heard of yours before the first call. Then look at the five who took months and ask the same question. The pattern is usually blunt. The fast ones arrived pre-introduced. The slow ones had to be introduced in the meeting, which is the most expensive place to do it.

Every firm has a brochure, even if it is now called a capabilities deck or a services page. It lists what you do, who you have done it for, and a few phrases about passion and partnership. It is read once, by someone already half-interested, and then it is filed in the same mental drawer as every other brochure they have ever been sent.

There is a further difference in what each format can carry. A brochure can make claims. It can say we are experts in supplier negotiation . A book can show the expertise: entry forty-one on the three clauses suppliers always try to remove, entry forty-two on what to do when they succeed. The buyer does not need to take your word for anything because the evidence is in front of them, and they have tested it against their own experience while reading.

You will still need a services page. People need to know what to buy and how. But stop expecting it to do the persuading. Its job is to be found after the persuading has happened, as the practical next step for someone who has already decided you know what you are talking about.

Why a hundred? Not ten, which is a blog post, or thirty-seven, which is a number nobody chose on purpose? The honest answer is partly psychological and partly practical, and both parts work in your favour.

Practically, a hundred is the size at which a book stops being read and starts being browsed. Nobody sits down to read a hundred entries in order, and that is fine. They read the contents, find the three that match their week, and dip in. Next month they come back for two more. A book that rewards browsing gets opened far more often than one that demands to be read cover to cover, and each opening is another moment of contact with your thinking.

There are objections. A hundred feels like a lot. It is less than it feels. At four to five hundred words per entry, it is the length of a modest non-fiction book, and most experts have said every word of it already, in meetings and emails and on calls. A hundred can also look like padding if the entries are thin. That is a real risk, and much of this guide is about avoiding it. Every entry must teach one thing a reader could use this week. If you cannot find a hundred such things, your niche may be too narrow, and that is useful information too.

The first fear most experts have about writing a book is that they will give away the secret sauce. If the method is all in print, why would anyone pay for it? It is a reasonable worry and almost entirely unfounded.

In fact, giving the method away tends to increase demand for the implementation. A reader who works through your entry on running a pricing review now understands why pricing reviews matter, what a good one looks like, and how many ways it can go wrong. They are better informed, which makes them a better buyer. They also realise they do not have the time to do it properly. You have, in effect, written their business case for them.

That does not mean you must publish every template, price sheet and client file. Hold back what is genuinely proprietary: specific tools you have built, client data, the commercial details of how you work. But be generous with the thinking. The thinking is what demonstrates that you are worth hiring, and it is the part competitors cannot really copy, because it only works when applied with the judgement that produced it.

The experts who worry most about giving things away are often those with the least to give. If your knowledge would be exhausted by a hundred entries, the book will tell you so. If it would not, the book will tell your buyers. Either way you learn something. Generosity is not charity. It is the cheapest form of proof there is.

Most expertise is invisible, and not because experts are shy. It is because expertise lives in places that cannot be seen from outside: in the head of the person who has it, in hundreds of emails sent to individual clients, in calls nobody recorded, in decisions that were obvious to the expert and therefore never explained.

The result is a curious unfairness. Two consultants may be equally capable, but the one who has written things down will be perceived as more capable, because only that one can be assessed without a meeting. The other is relying on referrals and luck. Referrals are lovely. They are also not a strategy, and they tend to dry up exactly when you need them.

There is a pleasant side effect. Writing the top layer clarifies the lower ones. Experts often discover, halfway through the book, that they hold views they had never quite articulated, or that two of their standard recommendations contradict each other. The book becomes an audit of your own thinking, and you come out of it sharper in meetings as well as more visible outside them.

There is an old sales maxim that people believe what they can touch. It is not quite true, since most of us believe plenty of things we cannot touch, but it points at something real. Evidence that exists as an object carries more weight than evidence that exists as a claim.

The tangibility matters even when the book is digital. A PDF in someone's downloads folder, a web reader they have bookmarked, an app on their phone: these are objects in the buyer's world that persist after you have left the conversation. They can be shown to a colleague. They can be checked again before a decision. A claim made on a call evaporates by the next meeting. A book stays where it was put.

Specificity is the other half. A book that stays at the level of principles is tangible but vague, which puts it in the same box as a nicely printed brochure. The proof is in the detail: the named failure mode, the worked example, the number of steps, the question to ask on day three. Readers trust specificity because it is expensive to fake. Anyone can say they understand procurement. Only someone who has been in the room can describe the moment when the finance director goes quiet.

This week, gather every form of proof you currently offer prospects and place each on those two axes. Notice how empty the specific-and-tangible corner is. That emptiness is your opportunity. Fill it with a hundred things worth keeping, and you will no longer have to ask anyone to take your word for it.

A book borrows authority from the format before a word has been read. That is useful and slightly unfair, and it is worth being clear-eyed about both halves of the sentence.

But the format only lends; it does not give. The borrowed authority buys you attention, and attention is a loan with short terms. Within a few pages, the reader decides whether the content matches the jacket. If the entries are thin, generic or obviously padded, the book format makes things worse, not better, because it raised expectations that were then disappointed. A weak book is more damaging than no book, in the same way a bad suit is worse than a decent jumper.

This is why the rest of this guide spends far more time on entries, structure and voice than on covers and layouts. Design matters, and there is a part on it, but design is in service of content. A beautiful book of empty advice is a beautiful brochure, and we have already discussed what happens to brochures.

A useful check this week: pick any business book on your own shelf that you never finished. Ask what made you stop. It is rarely the cover or the typeface. It is usually that the book promised more than the pages delivered, and you noticed somewhere around chapter three. Your readers will notice at the same point. Make sure that chapter three is one of your best. Formats are rented. Trust is bought outright.

Ask any expert what slows their sales and you will hear a familiar list: too many calls that go nowhere, too much time explaining the basics, prospects who do not understand the value until the third meeting, decisions that drift for months. A book does not fix all of these. It quietly improves most of them.

It shortens the explanation. Much of a first meeting is usually spent establishing shared vocabulary: what you mean by a diagnostic, why you start with interviews, what a good outcome looks like. If the buyer has read entries on each of these, you can skip ahead. Calls start at minute twenty. Proposals need less justification because the reasoning has been published.

None of these effects is dramatic on its own. You will not see a sudden flood of signed contracts the week the book goes live. What you will see, over a few months, is a gradual change in the texture of your sales conversations. People arrive using your language. They reference entries by name. They ask better questions. The ratio of calls to clients improves, and the calendar has fewer meetings that should have been emails.

Then, after the book has been in circulation for a quarter, ask the same three questions. If nothing has changed, something is wrong with the book or with how it is being used, and later parts of this guide will help you find out which. If things have changed, you will know the book is earning its keep. A pipeline is mostly friction. A good book is mostly lubricant.

This guide is written for a fairly specific kind of business, and it is worth being honest about the edges. A hundred-entry book is a serious undertaking, even with good tools, and it pays best where certain conditions hold.

It suits businesses where each client is worth a meaningful amount. If a single engagement keeps you going for weeks or months, then winning even one extra client a quarter because of the book makes it worthwhile many times over. If your typical sale is small and transactional, you will need volume the book cannot easily deliver, and a different kind of marketing will serve you better.

It suits people willing to hold a point of view. A book of carefully neutral advice, offending nobody and committing to nothing, will be ignored. If you have opinions about how your field should work, and you are prepared to put them in writing, you have the raw material. If you would rather not, a book will feel like a chore and read like one.

If you recognise yourself in the first group, keep going. Take ten minutes this week to write down, in one sentence, what clients pay you for that they could not easily get elsewhere. If the sentence comes easily, the book will too. If it does not, that sentence is the first thing to work on, and it would be the first thing to work on even if you never wrote a book. Fit is not a formality. It is the whole bet.

Most marketing evaporates. A post is seen for a day, a talk is remembered for a week, a campaign runs for a month and then stops when the budget does. The effort is real, the return is real, but nothing is left behind once the activity ends. You are paying rent on attention.

It also compounds. Each client the book helps you win becomes a source of new material: a fresh problem, a new example, a refinement of an entry that turned out to be slightly wrong. Feed those back into a refreshed edition, and the book gets better and more specific to your market each year. Better books attract better clients, who generate better material. The loop tightens.

The compounding only happens if you keep the book alive. A book that is written, launched and then never touched will slowly date. References go stale, examples feel old, and readers start to notice that the edition says a year that is no longer this one. The fix is cheap: a short, scheduled refresh, a few entries rewritten, a new edition date, and a reason to tell people about it again. This guide returns to the refresh habit near the end, because it is where the real value of the asset sits.

The most common mistake in planning an authority book is writing it for everyone. It feels safe. A book on better leadership or growing your business has an enormous potential audience, and surely a few of them will become clients. They will not, because a book for everyone is a book nobody recognises as being for them.

Narrowness feels like giving up readers. It is actually a way of gaining the right ones. The size of your market does not matter much when you only need a handful of clients a year. What matters is that the people who could become clients feel, on first sight, that this was written for them. That feeling is the difference between a book that sits in a downloads folder and one that gets read on the train home.

How narrow is narrow enough? A useful test: could you name ten real people who would immediately recognise themselves as the reader? Not types of people, actual names. If you can, the niche is probably right. If you find yourself writing business owners or leaders or anyone who wants to grow , keep narrowing. Add a sector, a size, a role, a stage, or a specific problem. Usually two of those are enough.

This week, write your niche as a single line in this shape: for [role] at [type of organisation] who [have this problem] . Then cut whichever part is vaguest and replace it with something more specific. You can always write a second book for the neighbours. Being chosen requires being recognisable first.

Experts tend to write for other experts. It is natural: those are the people whose respect they want, whose work they read, and whose critique they fear. The result is books full of nuance, jargon and careful qualifications that impress peers and lose clients.

That means a few concrete changes. Define terms the first time you use them, or better, choose words that need no definition. Lead with the consequence the reader cares about before the method you care about. Use examples from their world, not yours: the operations director's shift handover, not the theory of organisational communication. Cut the paragraph where you defend your approach against a rival school of thought that the reader has never heard of.

There is a risk of overcorrecting into simplicity. Buyers are not children. They can handle complexity when it is relevant to them, and they notice when they are being talked down to. The aim is clarity, not dilution. You can say difficult, precise things in plain language, and the effort of doing so tends to sharpen the thinking as well as the prose.

Your peers may, in fact, read the book and find it a bit obvious in places. Let them. Peers rarely notice that obviousness to an expert is revelation to a buyer, and they are not your market. Try this week: take one paragraph you have written for clients recently and show it to someone outside your field. Ask them to underline any word they would not use themselves. Those underlines are the distance between you and your buyer. Close it.

Every authority book needs a promise line: one sentence that says what the reader will get from it. Not a mission statement, not a tagline, not a list of topics. A promise. After reading this, you will be able to do something you could not do before.

A good promise line usually has three parts, though they need not appear in this order. Who it is for, sometimes implied. What problem they have, or what they want. What they will be able to do after reading. Fill in all three, then remove words until it hurts. The subtitle of a book is often the best home for the result, sitting under a shorter title that does the catching.

Beware promises you cannot keep. Double your revenue in ninety days is a promise, but it is the kind that attracts the wrong readers and invites justified scepticism. Buyers in professional markets have been oversold for years and have excellent detectors. A modest, specific, achievable promise is far more persuasive than a grand one. Run your first pricing review in a fortnight will win more trust than transform your profitability .

This week, write ten versions of your promise line. Write them fast and do not judge them while writing. Then choose the one that a buyer would most want to be true, and that you could most confidently make true. That one is your book. The others are useful mainly for showing you what it is not.

Titles are where experts most often indulge themselves. After months of serious thought, it is tempting to reach for something clever: a pun, a metaphor, an allusion to a novel. The trouble is that clever titles work brilliantly on people who already know what the book is about, and not at all on people who do not. Your buyers are the second group.

So prefer clarity. The Shift Handover tells a logistics manager exactly what they are getting. Passing the Baton might be more charming, but it could be about relay running, succession planning or relationships. You can add charm in the subtitle, or in the entries, where the reader has already committed to reading. The title is not the place to make them work.

Run three tests on any shortlist. First, the stranger test: show the title alone, with no subtitle, to someone in your target market and ask what the book is about. If they cannot tell you, it fails. Second, the search test: would your buyer type any of the words in the title into a search box when looking for help? If none of the words are ones they would use, discovery will suffer. Third, the say-it-aloud test: can you introduce the book in conversation without explaining the title? If you find yourself saying it's a play on , start again.

This week, list every title you have been considering and apply the three tests ruthlessly. Most people find that their favourite fails the first. That is normal and slightly painful. Keep the favourite as the name of a part, or an entry, or a workshop. Give the cover to the title that does its job. Clever is a luxury. Clear is a courtesy.

There are two main shapes for a hundred-entry authority book, and the choice affects everything from how you find entries to how readers use the result. Both work. They suit different expertise and different buyers.

The second is the hundred-chapter book: ten parts of ten chapters, written as flowing prose with an argument that builds from start to finish. Each chapter still teaches one thing, but there is a through line. Chapter forty assumes you have met the ideas in chapter twelve. It reads more like a conventional book, rewards sustained attention, and carries more authority with buyers who value depth. It is also more work, because each chapter must be good on its own and fit the whole.

The hybrid is what this guide recommends for most experts, and it is the format this guide itself uses. Ten parts give the structure an arc; ten chapters per part give each part enough room; every chapter is written to stand alone. The same data can then be published as a ranked card deck for phones and as a book for reading, which later parts cover in detail.

This week, take five of your likely entries and write the first paragraph of each twice: once as a ranked item, once as the opening of a chapter. Read both versions and notice which you find easier to write and which a client would find more useful. Those are not always the same answer. When they differ, follow the client.

Every field has things its practitioners know and rarely say in public. The software project that was doomed from the scoping call. The coaching engagement that worked because the client was ready, not because of the coach. The marketing campaign that succeeded for reasons nobody on the team could explain. These unspoken truths are the best raw material for an angle, because they are interesting, credible and almost entirely absent from your competitors' marketing.

Good angles often come from the uncomfortable centre of your field. Look for a hard truth that clients do not want to hear but need to. An old habit everyone follows that you have found to be counterproductive. A hidden cost that nobody puts on the invoice. A topic that is slightly taboo: the politics, the emotions, the cases where the textbook method fails. Any one of these can anchor a book, and the best angles touch several.

You do not need to be contrarian for its own sake. An angle built purely on disagreement wears thin quickly, and readers can smell a manufactured controversy. The aim is to be truthful in a way that happens to be unusual. Your angle should be something you believe, can defend with experience, and would happily argue with a client across a table.

Not all problems are worth writing about equally. Some your buyers will pay almost anything to solve; others they will acknowledge politely and never act on. An authority book should spend most of its pages on the first kind, and a useful tool for sorting them is the problem ladder.

Write the list of problems your clients bring to you, then place each on the ladder. Be honest. Many experts discover that their favourite subjects, the ones they find most intellectually interesting, sit on the lower rungs. Buyers do not care about them nearly as much as the expert does. Meanwhile the problems that actually trigger a purchase are often less glamorous: a missed deadline, a lost client, a board asking awkward questions.

There is a subtlety. Some of the most valuable advice concerns problems that buyers place lower than they should: the costly-but-tolerated issue they have stopped noticing. Part of your authority comes from showing them that a second-rung problem is really a top-rung one in disguise. That is a powerful move, but it works best once you have earned trust by addressing what they already know hurts. Lead with their urgency, then introduce yours.

This week, take your list of client problems and rank them on the ladder. Then look at your website and count how much space goes to each rung. The mismatch is usually striking. The book is a chance to correct it, putting your best words where your buyers' worst days are.

Information is free. Every reasonably well-documented field has more freely available advice than any buyer could read in a lifetime, much of it accurate. If your book merely collects that information, however neatly, it competes with everything else on the internet and with every AI assistant that can summarise the same sources in seconds. It will lose.

Points of view are usually visible in small moments. An entry that says most guides recommend X; I have stopped recommending it, and here is why . A sentence that names a common practice as a waste of time. A preference for one approach over another, stated plainly with reasons. These moments are where readers sit up. They are also where readers decide whether they agree with you, which is exactly the decision you want them to make before they hire you.

A point of view also helps with AI-assisted production, which later parts cover. Agents are very good at producing balanced, comprehensive, opinion-free prose. Left alone, they will write a perfectly reasonable book that sounds like nobody in particular. Your point of view, stated clearly in the brief and enforced in the edit, is what stops this. It is the ingredient the machine cannot supply.

This week, write a list of ten statements about your field that start with I believe . Make at least three of them ones a respectable competitor would disagree with. Keep the list somewhere you will see it while writing. When an entry feels flat, check whether it reflects any of the ten. If it does not, it is probably information, not authority. Information is plentiful. Conviction is scarce, and scarcity is what people pay for.

Before committing to a hundred entries, write ten. This is the cheapest test you can run on an authority book, and it will save you from the most expensive mistake: discovering after months of work that the promise does not land.

What you are looking for is not praise. People are polite and will usually say something nice. Watch for behaviour instead. Do they ask for more? Do they forward it to someone? Do they quote a line back to you? Do they ask whether you could help with the problem in entry six? Those are signs the promise is working. Polite silence, or feedback about the design, usually means it is not.

Expect to adjust. Most first samples reveal that the promise is slightly off, that the niche is slightly too broad, or that one kind of entry consistently outperforms another. That is the point. Changing the promise after ten entries costs an afternoon. Changing it after a hundred costs the book. If the sample flops completely, take that seriously too: it might mean the niche is wrong, or that the angle is not as interesting to buyers as it is to you.

This week, pick your ten and write one. Just one, properly. It will show you how long an entry really takes, how your voice holds up at length, and whether the promise feels real when you are inside it. A hundred is just ten, ten times. The first ten are the only ones that need courage.

Before production starts, everything you have decided about the book should fit on one page. Reader, promise, angle, title, format, and a few notes on voice. If it does not fit on one page, you have not decided yet; you have only accumulated thoughts.

A workable brief has six short sections. The reader: one sentence on who they are, in enough detail that you could name a real person. The promise: the one-line promise from earlier. The angle: the stance that makes this book yours, in a sentence or two. The title and subtitle. The format: Top 100, hundred chapters or hybrid, with the planned parts. And voice: three or four adjectives, a couple of rules, and a list of phrases you never want to see.

The brief is also your defence against drift. Halfway through a project, it is very easy to add an entry because it is interesting, or to change tone because you have been reading someone you admire. Each change seems small. Together they produce a book that started as one thing and ended as another. Checking new ideas against the brief keeps the book coherent, and gives you a polite reason to say no to your own enthusiasm.

This week, write yours. Give yourself one hour and one page, and do not let either stretch. Print it, read it aloud, and ask whether a stranger could use it to produce something close to the book in your head. If they could, you are ready for the next part. If not, the gap you feel is the part of the book you have not yet thought about. Better to find it here than in entry sixty-three.

The most daunting moment in an authority book project is the blank list. A hundred entries sounds like a mountain, and the natural response is to wonder whether you really know a hundred things worth saying. You almost certainly do. The problem is not knowledge; it is retrieval.

So do not start by trying to list everything you know. Start by going to the places where your knowledge has already been triggered and written down. The questions clients ask you, over and over. The mistakes you have seen, made or prevented. The emails where you explained something at length to one person. The calls where you talked a client through a problem. Each of these is a deposit of expertise, laid down at the moment it was needed, and they are much easier to mine than your memory.

There is a mindset shift worth making early. You are not writing a hundred new things. You are recovering ninety-odd things you have already said, and adding a handful you have been meaning to say. That reframing matters because it changes the task from creation to curation. Creation is slow and anxious. Curation is brisk and rather satisfying, like clearing out a cupboard and finding it full of things you are glad to own.

This week, set a timer for twenty minutes and write down every question a client has asked you in the last month. Do not filter. Most people reach twenty or thirty before the timer goes. That list alone is a third of a book. The rest is waiting in places you have not yet looked.

Your sent folder is a draft of your book, written in instalments over years and addressed to one reader at a time. Every long email in which you explained something to a client, advised a colleague, or argued for an approach is a candidate entry. Most experts have hundreds of these, and almost none have ever been read by more than one person.

Then strip identifying details. Client names, project names, figures that could identify a business, anything said in confidence. The book will draw on these experiences, but it must not expose them. Replace specifics with representative ones: a forty-person logistics firm rather than the real name. This is both an ethical requirement and a practical one, since prospects reading your book will wonder how you would treat their information.

An AI agent is useful here, provided you control what it sees. You can give it a folder of anonymised excerpts and ask it to group them by theme, suggest entry titles, and flag duplicates. Do not give it raw client correspondence; anonymise first, then process. The agent is not finding your ideas for you. It is sorting them, which is tedious work you would otherwise do badly at eleven at night.

This week, spend an hour in your sent folder and collect ten explanatory excerpts. Anonymise them as you go. Read them back together and notice how much of the book is already written, in your voice, about real problems. The work from here is mostly assembly. Your past self has done the writing. Your present self only needs to edit it.

Every question a client asks more than once is an entry. That is the simplest rule in this guide and one of the most productive. It works because repeated questions mark exactly the points where buyers get stuck, which are exactly the points where your book can help them.

Within a few weeks the log will show patterns. Certain questions recur in different forms. Others cluster into themes: questions about cost, about timing, about what to do first, about what can go wrong. Each recurring question is an entry. Each cluster might be a part. You are, in effect, letting your market design your table of contents, which is a far better designer than you are.

Do not overlook the questions prospects ask before they become clients. These are especially valuable, because they reveal the doubts and confusions that stop people buying. An entry that answers a pre-sale question honestly, including the parts that might make you look less essential, does two jobs at once: it helps the reader and it removes a barrier to working with you.

If you have been in practice a while, you can reconstruct a question log from memory and records rather than waiting weeks. Go back through meeting notes, discovery call summaries, and the frequently asked questions page you never updated. Ask colleagues what clients ask them. Ask a couple of clients what they wish they had known before starting.

Many experts speak better than they write. On a call, explaining something to a client who is visibly confused, they find the right analogy, the vivid example, the plain sentence. Then they sit down to write and produce something stiff and over-qualified. The fix is to capture the speaking and turn it into writing, and recorded calls make this easy.

What you are looking for in a transcript is the moment when someone asked a hard question and you answered well. These moments have a recognisable shape: a pause, a reformulation, and then a clear explanation, often with an example. Mark them, extract them, and save them as raw material. A single hour-long call can contain three or four entry-sized explanations.

AI agents are particularly good at this conversion. Give one a transcript excerpt and a voice file, and ask it to produce a written entry that preserves your phrasing and examples while removing the verbal clutter. The result is usually a strong first draft that sounds like you, because it is made of you. Compare this to asking the agent to write an entry from a topic alone, which will produce something accurate and anonymous.

Podcasts and talks are especially useful because they were already intended to be public. If you have appeared as a guest anywhere, the transcript is a ready-made seam of entries in your speaking voice, already cleared for publication.

The most valuable thing an expert knows is usually what goes wrong. Success stories are pleasant and often unrepeatable; they depend on timing, luck and circumstances that will not recur. Failures are instructive and highly repeatable. The same mistakes happen again and again, to different people in different firms, and someone who has seen them many times can spot them early and steer around them.

Each mistake is an entry with a natural shape. What went wrong, described concretely enough that the reader recognises it. Why it happened, which is usually more interesting than the mistake itself, because it reveals the incentive or assumption behind it. How to spot it early, before it becomes expensive. And what to do instead. That last layer is the payoff, the reason the reader keeps the book.

Include your own mistakes. It is tempting to present yourself as having always known best, but readers find that implausible, and it makes you less trustworthy, not more. An entry that says I used to recommend this, and here is what happened is more credible than any amount of confident instruction. It shows that your advice has been tested, and that you change your mind when the evidence demands it.

This week, write down ten mistakes you have seen clients make in the last year. For each, write one sentence on why it happened. You will find that the why sentences cluster around three or four root causes. Those root causes are probably the backbone of your book's argument. The mistakes are the evidence. Your job is to connect them.

Every field has its canon: the established frameworks, models and ideas that practitioners learn and reference. You will need to draw on some of them. Readers expect to see them, and leaving them out entirely looks either ignorant or evasive. The question is how to use the canon without turning your book into a summary of other people's work.

The second rule is application. Do not simply explain the framework; that has been done, usually better, by its originator. Show how it applies to your reader's specific situation. A general model of change becomes much more useful when you explain what it looks like inside a family-owned manufacturing firm with a founder who is about to retire. That application is yours, even if the model is not.

Be careful with direct quotation. Short quotes with attribution are fine. Long passages lifted from someone else's book are not, both legally and stylistically. Paraphrase in your own words, credit the source, and move quickly to your application. Never invent quotes or attribute words to real people that they did not say; that is the fastest way to lose a reader's trust entirely, and an easy mistake to make if you let an AI agent fill gaps unsupervised.

This week, list the three frameworks you reference most often with clients. For each, write down one situation where it works well in your niche and one where it does not. The second list is more valuable. It is the part of the canon you have earned, and the part nobody can get from the original source.

Once you have mined your emails, questions, transcripts and mistakes, combine everything into a single long list. Aim for three hundred candidate entries. Yes, three hundred, for a book of a hundred. The surplus is not waste. It is the raw material that makes the final hundred good.

Reaching three hundred feels hard until you stop judging while listing. Write each idea as a short working title or a single sentence. Do not worry about overlaps, gaps, or quality. Include ideas you are not sure about, ideas that seem too obvious, and ideas that might be two entries. The list is a net, not a filter. Filtering comes later.

AI agents are useful at this stage, with a caveat. Ask an agent to read your collected material and suggest additional entries that are implied but missing, such as the step between two entries you already have, or the opposite of a mistake you listed. These suggestions are often good, because they are grounded in your material. Do not ask it to generate entries from nothing, or the list will fill with the generic average of your field, and you will waste time cutting them.

You may find you blow past three hundred. Fine. You may stall at a hundred and fifty. That is fine too, and still enough to make a strong selection. If you stall below a hundred and twenty, take it seriously: it might mean your niche is too narrow, or that you have not looked hard enough in one of the sources. A later chapter deals with what to do when the well runs dry.

Now the editor comes in. You have a long list of somewhere between one hundred and fifty and three hundred candidates, and you need exactly a hundred. Culling is where the book gets its quality, and it is more enjoyable than people expect, provided you have a clear set of tests.

Second, remove trivia. These are entries that are true but too small to deserve five hundred words: a single tip that could be a sentence in another entry, a fact that is interesting but not useful. Fold the best of them into larger entries as supporting detail; let the rest go.

Fourth, and most important, the defence test. For each remaining candidate, ask: would I defend this in front of a sceptical client? Not is it correct , but would you stand behind it, with your name on it, across a table from someone who disagreed? Entries that pass are authority. Entries that fail are filler, and readers can feel the difference even if they cannot name it.

Be wary of keeping entries for sentimental reasons. Every expert has a pet topic that they find fascinating and clients find slightly baffling. If it passed the defence test but fails the usefulness ranking, it probably belongs in a footnote, a talk, or the bin. Likewise, beware of keeping safe entries simply because they are uncontroversial. A book with no entries anyone could disagree with is a book with no point of view.

After the cull, your hundred will have holes. Some stages of the buyer's journey will be overcrowded, with eleven entries about implementation, while others will be thin, with barely one about how to know whether you need help at all. Gaps are normal. Finding them deliberately is better than letting readers find them for you.

Draw this as a grid if it helps. Across the top, the journey stages. Down the side, a rough measure of coverage: how many entries, how deep, how specific. The cells with high stage importance and low coverage are your gaps. They are where readers will feel the book thin out, and often where they would most like your help.

To fill a gap, go back to your sources with a specific question. If the beginning is thin, look in your questions log for things prospects ask before they buy. If the end is thin, ask long-standing clients what they wish they had known in year two. If a middle stage is thin, look at your mistakes inventory: there are probably failure modes you have not yet turned into entries.

Resist the urge to fill gaps with generic material just to make the numbers work. A gap filled with filler is worse than a gap noticed and addressed honestly. If you genuinely have little to say about one stage, it may be better to rebalance the parts so that stage gets fewer entries, and to acknowledge that your expertise is concentrated elsewhere.

Sometimes, despite honest mining and generous listing, you cannot reach a hundred good entries. You cull your long list and land at sixty-five. You fill the gaps and reach seventy-eight. The remaining twenty-two either do not exist or would be filler. This is a useful signal, and it is worth reading correctly before you reach for the padding.

The second cause is an angle that is too thin. Some angles make a great article but not a book. If your book is built around a single insight, however true, you may run out of ways to apply it. The fix here is not to widen the audience but to deepen the frame: change the angle from a single insight to a way of seeing the whole problem, which will generate entries across every stage.

Whatever you do, do not pad. Padding is easy to produce, particularly with AI agents, which will cheerfully generate twenty plausible entries on any subject. Those entries will look fine at a glance and read as hollow on closer inspection, and readers will notice. A book with eighty strong entries and twenty filler ones is weaker than a book of eighty, and the hundred-entry format is not worth that cost.

This week, if you are below a hundred, write down which of the three causes fits best. Be honest; the third is the hardest to admit and the easiest to spot from outside. Then take the matching fix. A smaller true book beats a larger padded one, and a well that ran dry is still a well. It just needs a deeper hole.

A hundred entries in a single flat list is not a book. It is a heap. Readers cannot hold a hundred items in their heads, and they cannot navigate a list without landmarks. The simplest and most robust structure is ten parts of ten entries each, and it is worth understanding why it works before you start moving entries around.

Ten entries per part give each part enough room to develop an idea without sprawling. A part with three entries feels like an aside; a part with twenty feels like its own book. Ten lets a part start with fundamentals, work through the middle, and end with something ambitious, which gives each part a small arc of its own. That arc matters for readers who go in order, and it does no harm to readers who browse.

Resist the urge to break the grid for one special part that deserves fifteen entries. It may deserve fifteen, but the reader's sense of the book's shape is worth more than one part's completeness. Either split the big part into two, or cut it to ten and move the extras to where they also fit. The constraint will improve the writing; it forces you to choose the ten best rather than include the fifteen available.

The grid works across formats too. In a PDF, ten parts become ten sections with clear dividers. In a web reader, they become ten navigation items. In a swipeable app, they become ten categories a reader can filter by. The same structure travels everywhere, which is one reason later chapters can treat editions as outputs of a single source.

If you choose the Top 100 format, you face a question the chapter book avoids: what order do the entries go in, and what does the order mean? A ranked list implies that number one matters more than number fifty. Readers take this seriously. They will skip to the top ten, argue about the order, and judge your expertise partly by whether the ranking makes sense. So rank deliberately.

Avoid ranking by how interesting the entry is to you, how clever it is, or how much you enjoyed writing it. These are author-facing measures, and readers can tell when a list was ordered for the author's pleasure. The ranking should be a service to the reader: start here, then here, then here.

Expect to adjust. Your first ranking will be wrong in places. Read it in order and notice where the sequence feels off: an entry at number twelve that a reader could not use without first reading number forty, or two adjacent entries that are almost the same idea. Fix these by swapping or merging. Then read it again. Three passes usually produce a ranking that feels right.

If ranking feels false for your material, because the entries genuinely have no hierarchy, do not force it. Use categories without numbers, or the chapter book format. A false ranking undermines trust more than no ranking at all.

A hundred-chapter book needs an arc: a reason why part one comes before part two, and why the book ends where it does. Without an arc, the parts are just ten topics in arbitrary order, and readers who go from beginning to end feel no sense of progression. With one, the book builds toward something, and the last part lands with the weight of everything before it.

This shape mirrors how a good consultant runs a first engagement. You start by establishing that the problem is real and worth solving. You explain how it works, so the client understands their own situation. You set out what to do, in an order that makes sense. And you end with a perspective that reframes the whole thing, giving the client a way of seeing their business that lasts beyond the project. The book does the same, at scale, without you in the room.

The thesis part deserves particular care. It is easy to run out of energy by part ten and fill it with leftover entries that did not fit elsewhere. Resist this. Part ten should contain your most considered thinking: what the future holds for your field, what the reader should take away, and the single claim the whole book has been building toward. Its last entry should state that claim in a sentence anyone could quote.

This week, write the ten part titles on a single line, separated by arrows. Read the line aloud. If it sounds like a journey, keep it. If it sounds like a list, keep working. The arc is what turns a hundred useful entries into one useful book, and it is the difference between a reader who browses and a reader who believes.

Every entry in an authority book should teach one thing a reader could use this week. Not three things, not a general overview, not background. One lesson, clearly stated, practically useful, and complete enough that the reader could act on it without reading anything else.

The this week part matters. Readers of authority books are practitioners with problems, not students with time. An entry that teaches a principle with no obvious application is interesting but forgettable. An entry that teaches a principle and then shows what to do with it on Monday morning is useful, and useful is what readers remember and forward. The final paragraph or two of most entries should be concrete: a question to ask, a document to check, a meeting to hold differently.

A good discipline is to write the lesson as a single sentence before drafting the entry. Book the first-week call before the contract is signed, not after. If you cannot write that sentence, the entry is not ready; you are still working out what it is about. Once you have the sentence, everything in the entry should either lead to it, support it, or show how to apply it.

When working with AI agents, put the one-sentence lesson in the brief for each entry. It is the single most effective way to stop an agent from producing a sprawling overview, which is its natural tendency. An agent told what the one lesson is will build around it. An agent told only the topic will cover everything it can find.

A hundred entries need a repeatable shape, both for your sanity as a writer and for the reader's comfort as they move through the book. Not a rigid formula that makes every entry feel the same, but a default structure you depart from deliberately rather than by accident.

Of the four, the third movement is the payoff. It is the part the reader came for, the part they will act on, and the part that demonstrates you can help with more than diagnosis. Give it room. Many experts write a long, elegant why and a short, hurried what, because explanation is more comfortable than instruction. Readers want the reverse.

The template also makes AI-assisted drafting more reliable. An agent given a four-part structure, a one-sentence lesson, and some source material will produce a much more usable draft than one given only a topic. You can write the template into your voice file, with notes on what each movement should do and how long it should be. The agent will follow it, and you can edit from a consistent starting point rather than from a different shape every time.

Inside the template, prose rules. No bullet lists inside entries, no subheadings, no tables. These belong in reference documents. A book is read in sentences and paragraphs, and the absence of bullets forces you to show how ideas connect rather than just listing them.

Readers of authority books are browsers, not captives. They open an entry, read the first sentence, and decide in about a second whether to read the second. If the opening does not earn their attention, they move on, and the entry might as well not exist. Openings matter more in a hundred-entry book than in almost any other kind of writing.

Weak openings tend to share a few habits. They start with background: In today's fast-changing business environment. They start with definitions: Onboarding can be defined as. They start with the author: In my twenty years of experience. They start with a rhetorical question that nobody was asking. All of these delay the moment when the reader finds out whether the entry is for them, and most readers will not wait.

There is a craft to making openings varied across a hundred entries. If every entry begins with a problem statement, the book becomes monotonous. Mix problem, claim, scene, surprise and occasionally a plain direct instruction. Read the first sentences of all ten entries in a part one after another, and listen for repetition. If three start with Many firms , change two.

A test you can apply at once: cover everything but the first sentence of an entry and show it to someone in your niche. Ask whether they would read on. If they hesitate, rewrite it.

The end of an entry is where most writers lose their nerve. Having made the point, they feel compelled to make it again, in case the reader missed it. In summary, the first-week call is critical, and firms that get it right see better results. The reader did not miss it. They read it two paragraphs ago, and now they are reading it again, and they notice.

The alternative is a landing: a final sentence or two that closes the entry with a turn. It might be an aphorism that compresses the lesson into a memorable shape. It might be a dry observation that reframes what came before. It might be a slight twist, an implication the reader had not quite seen. What it should not be is a restatement. A landing moves the reader one small step further, rather than walking them back to where they already were.

A specific failure to watch for is the restating final paragraph. You write five good paragraphs, then a sixth that begins So or Ultimately and goes on to say what the fifth said, slightly more generally. This is the most common weakness in AI-drafted prose, and it is common in human prose too. The fix is almost always to delete the final paragraph and move its best sentence, if it has one, to the end of the previous paragraph.

When editing, read only the last two sentences of each entry in a part, one after another. If they sound repetitive, or if several say the same kind of thing, vary them. If any could be deleted without loss, delete them.

Part titles and category names are the main way readers navigate an authority book, and they are frequently wasted on labels that make sense only to the author. Methodology , Frameworks , Advanced Topics : these describe how the author thinks about the material, not what the reader needs from it. A reader scanning the contents does not know which one contains help with their problem.

This approach also tests your structure. If you cannot name a part after a reader need, the part may be organised around your thinking rather than theirs. That is not always wrong, especially for the opening and closing parts, which often carry the argument rather than a specific need. But most of the middle parts should map onto something a reader would recognise as their situation.

Categories matter even more in digital editions. In a swipeable app or web reader, they become filters and navigation tabs. A reader on a phone might see only the category names and a handful of entries at a time. Clear, need-based names let them jump straight to what they want; abstract names leave them scrolling.

There is a temptation to be clever with part titles, especially when the book title has been made plain. A little personality is fine, but navigation comes first. If a part title is witty, the subtitle must be literal, so the reader is never in doubt about what lies inside.

A hundred entries that each teach one lesson risk feeling like a hundred separate tips. The structure helps, but what really turns a collection into a book is a handful of threads: ideas that recur across parts, connecting entries that might otherwise feel unrelated, and quietly building the book's argument.

Choose three to five threads for your book. They should be ideas central to your point of view: the things you believe most firmly about your field. Then, as you plan entries, note where each thread naturally appears. Not every entry needs a thread, and no entry should be forced to carry one, but most parts should touch at least one, and the thesis part should draw them together.

Threads also give your book a vocabulary. When a phrase recurs, readers start to use it. Clients begin a meeting by saying they think they have a top-rung problem, or that they need a spec sample. That shared language is a powerful sign that the book is working, and it shortens conversations because you are both using words you have already defined.

A practical exercise: take a highlighter, digital or otherwise, and mark every appearance of one thread across a part. If it appears in every entry, it is overused. If it appears in none, either the thread is not really central or the part has drifted from your argument.

Entry one hundred is the most important entry in the book. It is the last thing a reader who goes in order experiences, the entry most likely to be quoted, and the place where the whole argument should come to rest. It should state the book's thesis plainly, in a way that a reader could repeat in a sentence to a colleague.

The thesis should not be a surprise. If it is, the book has not been building toward it. It should feel like the natural conclusion of everything before it, the idea the threads have been converging on, said once with full conviction. A good test is that a reader who has only browsed should still recognise it, because they will have met pieces of it scattered through the parts.

This entry also serves a commercial purpose, though it should never read as a pitch. A reader who finishes the book and agrees with the thesis is well on the way to becoming a client, because they now see their problem the way you do. They do not need to be told to book a call. If they agree, they will. If the final entry ends with a sales message, it undermines the generosity of the hundred entries before it and makes the whole book feel like an elaborate brochure.

Write entry one hundred early. Not first, perhaps, but certainly before you write the middle parts in earnest. Knowing where the book ends helps you build toward it, and you will find yourself planting the seeds of the thesis in earlier entries without forcing it.

For most of publishing history, writing a book meant sitting at a typewriter, or its descendants, and producing words in order until there were enough of them. The bottleneck was the writing itself. A hundred-entry book written that way took months, and many expert books were abandoned somewhere around entry thirty, when the author's enthusiasm met their calendar.

This calls for a different mental model. Not a typewriter, where you produce the words, but a factory, where you design a process with stages and quality checks, and the words come out the other end. The stages are roughly: a brief that sets the rules, an outline that fixes the content, drafting that turns outline into prose, editing that turns prose into your prose, and a QA pass that catches mechanical errors. Each stage has a clear input and output. Each can be improved independently.

Thinking in stages has practical benefits. When something goes wrong, you can find which stage failed. A bland entry usually means a thin brief or a weak outline line, not a bad draft. A factually wrong entry means the source material was missing or the editing was rushed. A badly structured part means the outline was not locked before drafting began. Fixing the stage fixes every entry that passes through it, which is far more efficient than fixing entries one at a time.

The factory model is not cold. A well-run workshop produces handmade goods; the process exists so that the craftsperson spends their time on the parts that need craft. Your book will still be yours, and arguably more so, because you will have had time to think about every entry rather than merely to type it.

The quality of an AI-drafted entry depends almost entirely on what the agent knew before it started. Clever prompting matters far less than people think. What matters is context: who the reader is, what the book promises, what the entry must teach, what the author sounds like, and what source material to draw on. Supply that context well and a plain request produces a good draft. Supply it badly and no amount of prompt engineering will help.

Each layer has a different job. The book layer sets the boundaries and the tone. The part layer gives coherence, so that ten entries in a part feel related rather than random. The entry layer gives precision, so that each entry teaches its one thing rather than drifting into a general overview. Missing any layer produces a recognisable failure: off-tone without the first, disconnected without the second, vague without the third.

Save your layers as files. Do not retype the brief at the start of each session or rely on memory; you will forget things and introduce inconsistencies. Most current agent tools let you attach files or keep project-level instructions that apply to every session. Use them. Your book brief and voice file should be loaded automatically every time.

This week, write the three layers for one part of your book: the book brief you already have, a part brief of a paragraph, and entry prompts for two entries. Draft those two entries with an agent. Compare the results to what you get from a one-line prompt with no context. The difference will persuade you more than anything this chapter can say.

The greatest risk in AI-assisted book production is not error but anonymity. Agents write competent, balanced, slightly over-explained prose that sounds like nobody in particular. A book in that voice may be accurate and useful, but it will not feel like you, and readers deciding whether to hire you want to meet you, not a well-informed stranger.

Sample passages come first because examples teach more than descriptions. Choose five to ten passages of your own writing, or transcribed speech, that sound most like you at your best: an email that explained something well, a paragraph from a talk, a post that people responded to. Include them in full. An agent will absorb rhythm, vocabulary and attitude from examples far more reliably than from any list of adjectives.

Banned phrases are the third layer, and they are surprisingly powerful. Every writer has words and constructions they would never use, and every agent has habits that give its prose away. List both. Overused intensifiers, corporate abstractions, certain transitions, the reflexive it's not just X, it's Y construction, any phrase you wince at when you see it in other people's writing. The agent will avoid them, and the prose will immediately sound less generic.

Test the voice file early. Draft two entries with it and two without, and show all four to someone who knows you well, without saying which is which. If they can identify the ones that sound like you, the file is working. If not, the samples probably need replacing with more distinctive passages.

The temptation, once you have a brief and an outline, is to ask an agent for the whole book at once. It may even appear to work. A hundred entries will arrive, roughly the right length, roughly on topic. Then you start reading and discover that entry seventy-three is a weaker version of entry twelve, that the voice drifted somewhere around part six, and that several entries got truncated mid-paragraph because the output ran out of room.

The rhythm is a loop. Draft a part. Edit the part. Then, before drafting the next, update the voice file with what you learned from editing. This last step is the payoff, and it is the one most people skip. Each part you edit teaches you something about where the agent's drafts diverge from your voice. Capture that lesson in the voice file, and the next part's drafts will be better. By part five, editing should be noticeably faster than it was in part one.

Save each part as its own file. This makes it easy to redo a part without touching the rest, to compare versions, and to assemble the book at the end from verified pieces. It also lets you work out of order if you need to. Writing part ten early, for example, helps you plant its thesis in earlier parts.

Resist editing across parts while drafting. If you notice something in part three that should change part one, note it in the log and move on. Going back mid-flow breaks the rhythm and invites endless revision. There will be a full-book pass at the end, where cross-part issues get fixed together.

Before any prose is drafted, the outline should be locked: all ten parts named, all hundred entries titled, each with its one-sentence lesson and a note on source material. Locked means you do not change it casually during drafting. It is a contract between you and the production process, and like most contracts, it saves arguments later.

With a locked outline, the agent knows exactly what each entry must do, and so do you. Drafting becomes a matter of executing a plan, not discovering one. When a draft comes back, you can check it against its outline line: does it teach the stated lesson? does it use the noted material? does it stay inside its scope? Any deviation is visible immediately, which makes editing faster and more confident.

A good outline line for each entry has four parts. The title. The one-sentence lesson. The source: which emails, transcripts or notes to draw on. And a note on diagram, if the book has them. That is about forty words per entry, four thousand for the book. Writing it takes a day or two. It is the most valuable two days of the project.

Save the outline in a structured format that both you and the agent can read. A simple JSON file or a spreadsheet works. It becomes the spine of the book's data, and later, when you produce multiple editions, the same structure will drive the PDF contents page, the web reader navigation and the app categories.

The author of an AI-assisted book is not a typist and not a passive approver. The role is closer to editor-in-chief: the person who decides what goes in, what comes out, what the book says and how it says it. Agents draft. You decide. Every entry in the book should pass through your judgement before it reaches a reader, and the book should be one you would sign, line by line.

The decision between the two should be quick. Read the draft once, fast, and ask whether the bones are right. If they are, edit. If they are not, do not try to rescue it with line edits; you will spend longer and end up with something worse. Go back to the outline line and the source material, find out what the agent was missing, fix the input, and redraft.

Editing an agent's draft is a particular skill. The prose usually looks polished, which makes it easy to read past problems. Read slowly and ask specific questions. Is this true? Is this what I would say? Is this specific enough that only a practitioner could have written it? Is there a generic sentence I can cut or replace with a real example? Does the ending land or restate? Mark everything that fails, then fix it.

Expect your editing rate to improve. Early parts may need heavy work; later ones less, as the voice file sharpens and you learn where the agent tends to go wrong. If editing does not get faster after a few parts, look upstream: the brief, the outline or the source material is probably thin.

AI-drafted prose has a dangerous quality: it is fluent even when it is wrong. A sentence stating a false statistic reads exactly as smoothly as one stating a true principle. An invented quote attributed to a real person looks just like a real quote. A confident description of a regulation that changed last year sounds authoritative. Fluency is not evidence, and an authority book that contains confident errors damages the authority it was meant to build.

Some categories deserve a blanket rule. No statistics presented as fact unless you have a source you have personally checked. No quotes attributed to real people unless you can point to where they said it. No specific claims about products, prices or features, which change too fast to keep accurate in a book. No legal or regulatory statements beyond the general principle unless verified for your jurisdiction. When in doubt, describe the principle rather than the specific; it is less vivid and much safer.

Build fact-checking into the production line, not as an afterthought. After editing each part, do a separate pass looking only for checkable claims. Highlight every number, name, date and source. For each, either verify it, generalise it, or remove it. This pass is faster than it sounds, because a well-written authority book should not contain many such claims; most of its value is in principles and practice, not data.

Dates and currency matter. A book published this year should be correct as of this year. Check that descriptions of tools, practices and regulations are current, and date the edition clearly so readers know when it was accurate.

A generic sentence is one that anyone could have written about any subject. Communication is key to success. Every business is different. It's important to consider all the options. These sentences are not wrong. They are simply empty, and a book full of them feels empty too, however much useful content sits between them.

The test is simple. For any sentence, ask: could this appear, unchanged, in a book about a completely different subject? If every business is different could appear in a book on marketing, recruitment or catering, it is generic. Either cut it or replace it with something specific. A twelve-person agency and a two-hundred-person consultancy need different onboarding, mostly because only one of them has a person whose job is onboarding says something.

Watch for particular patterns. Abstract nouns doing the work of verbs: the implementation of improvements instead of improving . Hedges stacked on hedges: it may perhaps be worth considering . Throat-clearing transitions: it is worth noting that , in today's world . Fake contrasts: it's not just about X, it's about Y . Lists of three abstract qualities: clarity, consistency and commitment . Add the ones you find to your voice file's banned list.

This edit has a cumulative effect. A single generic sentence does little harm. Twenty in an entry make it feel like a template. Two thousand in a book make readers wonder whether anyone was home. Cutting them is how the book starts to sound like a person.

Editing catches problems of judgement. A QA pass catches problems of mechanics: entries that are too short, parts with the wrong number of entries, diagram labels that will not fit, two endings that say the same thing, a stray bullet list where prose should be. These are tedious for a human to check across a hundred entries and trivial for a script. Let the script do it.

Add checks for the failures you find during editing. If you keep catching restating final paragraphs, add a check that compares the last paragraph of each entry with the one before it and flags high overlap. If certain banned phrases keep slipping through, add a search for them. If entries keep starting with the same three words, flag that. The QA script becomes a record of your book's particular weaknesses, encoded so they cannot recur unnoticed.

QA is not a substitute for editing. A book can pass every mechanical check and still be dull, wrong or unlike you. The script tells you that each entry is the right length, not whether it is worth reading. Treat a passing QA run as a minimum, the floor beneath which nothing ships, not as evidence of quality.

Keep the script with the book. When you produce a new edition, run it again. When you take on a client book, adapt it. Over time it becomes part of your production system, a quiet guarantee that every book you produce meets the same baseline.

AI-assisted production is fast. Once the brief, voice file and outline are in place, drafting a part can take an hour, and a disciplined author can move from blank page to finished book in days rather than months. This is genuinely new, and it changes what is possible for busy experts. It is also easy to misread.

The right way to use speed is to reinvest it. Time saved on drafting is time available for the stages that matter: a sharper brief, a better outline, more thorough editing, more careful fact-checking, a second pass on the weakest part. The best AI-assisted books are not the fastest ones. They are the ones where the author spent the saved time on care. Quality lives where speed and care overlap, and either alone falls short.

Be straightforward about how the book was made. You do not need to narrate your tools in the introduction, but you should never claim or imply that an AI-assisted book was produced the old way. If a reader or client asks, tell them: the knowledge and judgement are yours, the drafting was assisted, and every entry was edited and approved by you. Most people find this reassuring rather than off-putting, because it describes a careful process.

A useful discipline is to set a minimum editing time per part and keep to it, however good the drafts look. It protects quality from your own enthusiasm.

Most business documents are designed like presentations: big headings, bullet points, bold colours, a logo on every page, and lots of white space with not much in it. Experts who have spent years producing decks tend to design their book the same way. It is the wrong instinct. A book that looks like a deck reads like a deck, and decks are skimmed and forgotten.

They also make the book easier to read. Long passages of prose need different design from slides. Line length matters: too long and the eye loses its place; too short and the rhythm breaks. Leading, the space between lines, affects how tiring a page is. Paragraph spacing and indentation help the reader track the argument. None of this is visible when done well, which is precisely why it is worth doing.

This does not mean the book must look old-fashioned. Modern book design is clean, often minimal, and entirely compatible with a contemporary brand. Think of the design of good contemporary non-fiction rather than a textbook: confident typography, a restrained palette, one strong accent colour, and diagrams that feel integrated rather than pasted in. What it avoids is the visual noise of marketing: gradients, stock photography, slogans in large type, and anything that looks like it is trying to sell.

If you do not have a book designer, use a good template. Several publishing tools produce competent book layouts from structured text, and the rest of this part describes how to adapt one to your brand. Competent and calm beats elaborate and noisy every time.

An authority book needs to look like it came from you. That does not mean covering it in your logo. In fact, the more visible your branding, the more the book looks like marketing, and the less readers trust it. The art is to carry your brand in ways that feel native to a book: colour, type, tone and a few consistent details.

Type is the second carrier. If your brand has a typeface, consider using it for headings, where it will be noticed, while choosing a highly readable typeface for body text, which may not be the same thing. Many brand fonts are designed for short marketing copy and become tiring over five hundred words. A book can pair a brand heading face with a classic text face without losing identity.

Tone is part of the brand too. If your firm is known for being calm and precise, the book's design should be calm and precise. If it is known for energy and boldness, a bit more contrast and colour may suit. The design should feel like meeting you, the same way the voice should.

Consistency across editions matters. The PDF, the web reader, the printed copy and the app should share the same accent colour, the same heading type and the same diagram style. A reader who meets the book in several forms should recognise it each time. That is much easier if you define the brand tokens once, as a small set of colours and font choices, and use them everywhere.

If you only have time to get one design decision right, make it the body text. Most of a book is body text. Readers spend almost all their time looking at it. If it is comfortable to read, the book feels good even when other design choices are modest. If it is uncomfortable, no amount of clever layout will save it.

Headings come next. They should be clearly different from body text, through size, weight or typeface, but not so loud that they dominate the page. A good heading hierarchy has only a few levels: book title, part title, chapter title. Most authority books need no headings inside chapters, since entries are short and written as continuous prose.

Beware the decorative temptations. Fancy typefaces for body text, text in light grey to look elegant, very small sizes to fit more on a page, justified text without proper hyphenation that leaves rivers of white space. Each of these makes reading harder. Readers will not consciously notice; they will simply find the book tiring and stop sooner.

A useful shortcut: borrow the typographic settings of a book you find comfortable to read. Measure its line length, estimate its type size and leading, and note its margins. You are not copying its design, just learning from settings that have already been tested on readers.

Most authority books are walls of text, occasionally interrupted by a stock photo. Adding one simple diagram to every entry changes the experience entirely. The book becomes easier to browse, easier to remember, and noticeably more distinctive. It also forces you to understand each entry well enough to draw it, which improves the writing.

Diagrams also help browsing. A reader flicking through a hundred entries can see at a glance what kind of idea each one contains: a process, a choice, a cycle, a comparison. That visual signal helps them find what they need, in the same way headings help in a reference book. In a swipeable app, the diagram often becomes the face of the card, the thing that catches the eye before the title.

One per entry is the right number. More than one per entry turns the book into a slide deck. Fewer than one makes the diagrams feel occasional and decorative rather than structural. One per entry, consistently, makes them part of the book's identity. Readers come to expect them and look for them.

Diagrams can be generated from data. If each entry in your book's data file includes a small diagram specification, such as the kind of diagram, its labels and which element is the focal point, a single drawing engine can render all hundred consistently. This is far more efficient than drawing each one by hand, and it means changing the design of every diagram at once is a small change to the engine.

A small vocabulary of diagram kinds covers almost every idea an authority book needs to express. The skill is matching the kind to the shape of the idea, so that the diagram clarifies rather than decorates. A mismatched diagram is worse than none: it suggests a structure the idea does not have.

A split, a question with two branches, suits decisions: a choice the reader must make, with the better path highlighted. A funnel suits distillation: many inputs narrowing to a few outputs, such as a long list becoming a short one. A two-circle overlap suits intersections: where two things meet to produce a third. A two-by-two grid suits positioning: where something sits relative to two dimensions, with the most interesting cell highlighted.

Vary the kinds across a part. If every entry in a part uses a chain, the diagrams stop helping readers distinguish entries. A simple rule is never to use the same kind more than twice in a row within a part. This occasionally forces you to look at an idea from a different angle, which is often productive: a process might also be a cycle, if you notice that its last step feeds back into its first.

If you are generating diagrams from data, the vocabulary becomes a schema: a fixed set of kinds, each with defined label slots and limits. That constraint keeps the book consistent and makes it easy to check that every diagram fits its kind.

Every diagram in an authority book should have exactly one focal element: the node, box, layer or cell that carries the entry's payoff. It is drawn in the accent colour; everything else is neutral. A reader glancing at the diagram should see the focal element first and understand, before reading a word, where the lesson lands.

The focal element is almost never the first step. In a chain from brief to outline to draft to edit, the payoff is the edit, where the author's judgement turns a draft into a book. In a funnel from a long list to a short one, the payoff is the bottom tier, the final hundred. In a split between two options, it is the better option. Readers' eyes go to colour, so put the colour where you want their attention to end, not where the process begins.

Some diagram kinds fix the focal element by their shape. A hub's centre is focal by definition, since it is what everything else serves. The overlap of two circles is focal, since the intersection is the point. In a grid, the highlighted cell is focal. For other kinds, such as chains, stacks, cycles, funnels and splits, you choose which element to accent, and that choice is where the editorial judgement lives.

When generating diagrams from data, the focal element is a single number in the specification: the index of the element to accent. That makes it easy to review across the whole book. A quick script can list every entry with its focal label, and you can read down the list checking that each one is the payoff and not the input.

Diagram labels have to be short. Not short as a stylistic preference, but short because the boxes are small, the screens are small, and nobody reads a sentence squeezed into a rectangle. Ten characters per label in a four-step chain. A dozen or so for the centre of a hub. A couple of dozen for a wide stack layer. These limits feel brutal at first. They turn out to be one of the most useful constraints in the book.

There is a craft to it. Prefer nouns and short verb phrases. Drop articles. Use the vocabulary of the entry, so the label connects to the text. Avoid abbreviations that readers have to decode; a label that needs a key has failed. Avoid jargon that only experts know. If a concept genuinely cannot be named in the space available, the diagram kind may be wrong; a stack with longer layers might suit better than a chain with short boxes.

Label limits should be checked mechanically. When you write diagram specifications by hand or with an agent, it is easy to exceed a limit by a character or two without noticing. A QA script that checks every label against its kind's limit catches these instantly. Without it, you will find them later as clipped text or overlapping boxes in the finished edition, which is a much worse time to find them.

Test the labels at actual size. A label that looks fine on a desktop monitor may be unreadable on a phone. Render a few diagrams at the smallest size they will appear and check that every label can be read without zooming. If not, shorten further or reconsider the kind.

Most people who ever see your book's cover will see it small. A thumbnail in an email. A tile in a social feed. An icon on a landing page. A small image in a list of resources. The full-size cover, viewed at leisure, is a rare event. Design for the thumbnail first, and the full size will take care of itself.

Judge a cover on two qualities. Legibility: can the title be read at the smallest size the cover will appear? Distinctness: does it look different from other covers in the same space? A cover high on both is strong. A cover that is legible but generic blends into every other business book. A cover that is distinctive but illegible is a nice image that nobody can name.

Keep the cover simple. A plain background in your accent colour or a neutral, the title large, the subtitle smaller, your name, perhaps a single graphic element such as one of the book's diagrams enlarged. That is enough. Stock photography of handshakes, lightbulbs, mountain summits and people pointing at whiteboards should be avoided; they signal generic business content, which is the opposite of what you want.

Test the cover in context. Place a thumbnail of it in a mock email, a mock social post, and a grid of other business book covers. Shrink it to the size of a fingernail. Can you still read the title? Does it stand out from its neighbours? If not, simplify until it does.

An authority book in 2026 will be read on at least three kinds of surface: screens in light mode, screens in dark mode, and paper. Often it will also be read on a phone, which brings its own constraints. Designing for one and hoping the others work is how books end up with invisible diagrams, unreadable accent colours and text that glares at night. Design for all of them from the start.

Light mode is the default most people design for, and the easiest. Dark text on a light background, accent colour as-is. Dark mode needs more thought. The background darkens, text lightens, and your accent colour may need adjusting, since a deep brand colour that looks rich on white can become nearly invisible on dark grey. Diagram strokes and fills need the same attention. Test every diagram kind in dark mode; it is where problems hide.

Phones are the surface most likely to be underestimated. A large share of readers will meet your book on a phone first, perhaps through a link in a message or a post. Text must be readable without zooming. Diagrams must work at small sizes. Navigation must be usable with a thumb. If the book does not work on a phone, a large share of potential readers will never get past the first screen.

This week, take one finished entry with its diagram and view it in light mode, dark mode, printed on paper, and on a phone. Note every problem. Most authors find at least one per surface. Fix them at the token level, not one element at a time, and every entry will improve at once.

The first book is the expensive one. You design the layout, choose the typefaces, define the colour tokens, build or adapt the diagram engine, set up the editions and write the QA script. All of that takes time and thought. The second book should take a fraction of the design effort, because most of it already exists. That only happens if you build the first book as a system rather than a one-off.

The payoff is the top layer. When the system works, a new book is mostly a matter of writing new content. The design, the editions, the diagrams and the QA are all handled by what you built for the first one. You can focus entirely on what the book says, which is where your time is most valuable.

A system also makes refreshes easy. When you update the book next year, you change the data and rebuild. Every edition updates together, every diagram redraws, and the QA script checks the result. Without a system, a refresh means editing each edition separately, which is tedious enough that it often does not happen.

Keep the system simple. It is tempting to build an elaborate publishing platform. Resist this until you have produced several books and know which features you actually need. A folder with a data file, a few templates, a diagram engine and a build script is a perfectly good system. Complexity can be added when the books demand it.

The single most useful technical decision in an authority book project is to keep the whole book in one structured source file and generate every edition from it. One file holds the metadata, the ten parts, the hundred entries and their diagram specifications. From that file, templates produce a PDF, a print-ready layout, a web reader and a swipeable app. Change the file, rebuild, and every edition updates together.

A single source avoids this entirely. It also changes how you think about editions. They stop being separate products to maintain and become views of the same book, each suited to a different situation: the PDF for email, print for meetings, the web reader for links, the app for phones. You can add a new edition later, an email course or a slide deck, by writing one new template, without touching the content.

The source is the book. Treat it accordingly. Keep it under version control so you can see what changed and when. Run the QA script against it before every build. Back it up. When a client or colleague wants to suggest changes, have them suggest changes to the source, or to a view of it, not to one edition.

There is a small upfront cost. Setting up templates for each edition takes time, especially the first time. But the cost is paid once, and the benefits compound with every edit, every refresh and every new book.

Of all the editions of an authority book, the PDF does the most work. It is the edition that gets attached to emails, forwarded between colleagues, downloaded from landing pages, saved to desktops and opened on the train. It is not glamorous. It is reliable, universal and portable, and those qualities matter more than glamour for an asset whose job is to travel.

The PDF's journey follows a pattern. It is attached to an email or downloaded from a page. It is forwarded, often more than once, to people you have never met. It sits in a downloads folder until someone needs it. And eventually, at the moment a problem arises, someone remembers it and opens it again. That last step, recall, is the payoff. The PDF is a time-delayed handshake, waiting in someone's files until they are ready.

Think about the first and last pages. The first page after the cover should tell the reader what the book is and who it is for, in a sentence or two. The last page should tell them who you are and how to get in touch, briefly, without a hard sell. Between those two, the book should be pure value.

Date the PDF. The edition and date should be on the title page or colophon, so readers know how current it is. This matters more for PDFs than other editions, because PDFs persist long after you have published a newer version. A reader who finds an old PDF should be able to tell that it is old.

In a world where almost everything arrives on a screen, a printed book is an event. It is heavier than a business card and lighter than a proposal. It sits on a desk and gets picked up by colleagues. It does not vanish into a downloads folder or get buried under new email. For an authority book, print is less a distribution channel than a calling card: a physical object that represents you in rooms you are not in.

The posted copy is the classic move. You have had a promising first conversation with a prospect. Afterwards, rather than sending a follow-up email that will be skimmed, you post them a copy of the book with a short handwritten note. It arrives a few days later. It sits on their desk. When they next think about the problem you discussed, the book is there, physically, reminding them of you. The call that follows is warmer than any email could have made it.

Design matters more in print than on screen, because physical flaws are harder to ignore. Check margins for binding, check that diagrams print clearly, check that the cover looks right when held. Order a proof copy before ordering several. Read it with a pen in hand; you will find things the screen did not show you.

A small practical point: keep a few copies at your desk and a couple in your bag. A book that is in a box in the cupboard when the moment arrives is not a calling card. The best opportunities to hand someone a copy are often unplanned.

A PDF is a file. A web reader is a place. Publishing your authority book as a set of web pages on your own domain gives it a permanent address, makes every entry linkable, and lets readers search, browse and arrive from anywhere. It is the edition that turns your book from an attachment into a destination.

Linkability also helps discovery. Entries on the web can be found by search engines, by AI assistants that read the web, and by people following links from elsewhere. A buyer searching for help with a specific problem may land directly on the entry that addresses it, without ever having heard of your book. That is the handshake happening without any effort on your part.

Host it on your own domain. Your website, or a subdomain of it, rather than a third-party platform. This keeps the readers' attention on your brand, lets you control the experience, and means the links will keep working as long as your site does. Third-party platforms change their rules, their designs and sometimes their existence; your domain does not, unless you let it lapse.

Keep analytics simple and respectful. Knowing which entries are read most often is useful; it tells you what your market cares about and which entries to improve or promote. Detailed tracking of individual readers is less useful and can feel intrusive. A basic count of visits per entry is usually enough.

Phones are where most reading now happens, and phones are not good at books. Long pages, small text and fiddly navigation make even a well-designed web reader feel like work on a small screen. A swipeable app solves this by presenting each entry as a card: one entry per screen, swipe to move on, tap to read more. It is how people already consume content on their phones, and a hundred-entry book fits the format naturally.

The loop that makes the app valuable is the return. A reader swipes through some cards, saves a few that matter to them, and comes back later to read those properly. Saving is the payoff, because it turns a casual browse into a personal collection. A buyer with five of your entries saved on their phone has, in effect, built their own short version of your book, focused on their problems. That is a strong position to be in when they next need help.

Design for thumbs. Large tap targets, swipe gestures that feel natural, navigation by part that does not require precise taps. Text should be readable without zooming. Diagrams should be legible at card size, which is another reason to keep labels short.

For many readers, the app will be the edition they use most. It is always in their pocket, it is quick to open, and it rewards a minute of attention. That makes it unusually good at keeping you present in a buyer's mind over the weeks or months before they are ready to talk.

A hundred entries make a very long email course. Too long, perhaps, for a single sequence, but the material is ready-made for something more useful than a newsletter that struggles for topics: a structured series of lessons delivered to readers' inboxes, each one a complete entry, each one bringing them a step further into your way of thinking.

The reply invitation is where the value lies. An email arrives in a personal space, and it is natural to reply to it. A reader who finds an entry particularly relevant may hit reply and say so, or ask a question about their situation. That reply is the start of a conversation, which is exactly what an authority book is meant to provoke. The payoff of the email edition is not the subscription or the open rate. It is the conversation.

Keep the emails plain. Rich formatting, heavy images and elaborate templates make emails look like marketing and push them toward the promotions folder. A plain email with good text and perhaps a single diagram image looks like a message from a person, and is more likely to be read and answered.

Link each email to the web reader version of its entry, so readers can share it or find it again. Include a quiet line at the bottom about how to get the whole book. Do not turn every email into a sales message; the entries are the persuasion.

A well-structured authority book is also a library of talks. Each part is a coherent argument about one aspect of your subject, built from ten entries with a beginning, a middle and an ambitious end. That is very nearly the shape of a good talk. With some selection and adaptation, any part can become a keynote, a workshop or a webinar, and the book provides the takeaway.

The diagrams become slides. Each entry already has a simple diagram with a focal element, which is exactly what a good slide should be: one idea, visually clear, with the payoff highlighted. A talk built from three entries might have a dozen slides, most of them diagrams from the book, with a few title slides between sections. The visual consistency between talk and book reinforces the connection.

Speaking and the book reinforce each other. A talk gives you a reason to mention the book, and an audience that has just heard you speak is unusually receptive to reading more. The book gives the talk depth, because you can refer people to it for the detail you had to cut. Offering the book to attendees, in whatever edition suits, is a natural close that does not feel like a pitch.

Record what you can. A talk based on a part, recorded and transcribed, becomes source material for the next edition of the book, often with sharper phrasing than the original entries, because speaking to a live audience forces clarity.

Every entry in your book is a piece of social content waiting to be released. A single entry contains a hook in its first line, a lesson in its middle, a diagram that works as an image, and a landing that works as a quote. Released one at a time, across the channels your buyers use, a hundred entries become months of useful posts that all point back to the same body of work.

The economics are attractive. Most experts struggle with social content because each post requires a new idea. With a book, the ideas exist; the work is adaptation, which is much faster and can be partly automated. An agent can draft a post from an entry in moments, with your voice file ensuring it sounds like you. Card images can be generated from the diagram specifications. Clips need you on camera but no new thinking.

Link each piece back to its entry in the web reader. A reader who finds the post useful can click through to the full entry, and from there browse the book. This is how social content stops evaporating: each post is a doorway to the asset rather than a standalone piece that disappears after a day.

Choose platforms by where your buyers actually are, not where content is fashionable. For most professional services, that means one professional network, perhaps a newsletter, and video only if you are comfortable on camera. Being present in one place consistently beats being present everywhere occasionally.

Every authority book starts ageing on the day it is published. Examples become dated, tools change, regulations move, and your own thinking develops. A book that carries no date gives readers no way to judge how current it is. A book that carries an old date and has never been updated tells them you have moved on. The answer is to date every edition clearly and plan the refresh from the start.

Plan a refresh cycle. For most fields, an annual refresh is about right: frequent enough to keep the book current, infrequent enough to be manageable. Fast-moving fields may need more frequent updates to certain parts. A refresh does not mean rewriting the book. It means reviewing every entry, updating the ones that have dated, replacing the weakest few with better material collected over the year, and changing the edition date.

Keep a change log. Note what changed in each edition and why. This helps you, because you can see how the book and your thinking have evolved. It also helps readers who return to the book, since you can tell them what is new. A short note in the introduction of each new edition, listing the main changes, is a good way to bring back readers who read the earlier one.

Decide what to do with old editions. Usually the right answer is to replace them everywhere you control: the web reader shows the current edition, the PDF download is the current edition, and printed copies are reordered. Copies already in the world will persist, which is fine; the edition date tells readers what they have.

Your book will end up in many places. A PDF in hundreds of downloads folders. Entries quoted in posts and newsletters. Perhaps a listing on a retail platform. Perhaps a version hosted by a partner. That spread is good; it is the book doing its job. But somewhere, there should be one canonical copy that you control completely, and it should be on your own domain.

Picture the options on two axes: how much control you have over the copy, and how much reach it gets. Third-party platforms often offer reach with little control: they decide how your book looks, what sits next to it, and whether readers can contact you. Private files offer control with little reach. Your own domain, well linked, offers the best combination: full control and reach that grows over time as links and search visibility accumulate.

Ownership also matters for data. When readers find your book on your domain, you know which entries are read, which pages lead to enquiries, and which topics attract attention. That information shapes the next edition and your marketing. On a third-party platform, that data belongs to the platform, if it is available at all.

Keep the domain simple and memorable. The book's title as a path on your main site works well. A separate domain for the book can work too, especially if you expect the book to have a life of its own, but it adds something to maintain. Whichever you choose, make sure it is renewed automatically; an expired domain is an embarrassing way to lose an asset.

The instinct, when a book is finished, is to announce it to the world. A post, a press release, an email to the entire list, perhaps a launch event. Most of these produce a brief flurry of attention and then silence. The world, it turns out, is busy, and it has not been waiting for your book.

The reason is that authority books spread through recommendation, not broadcast. A buyer who receives the book from someone they trust, with a note saying you should read this , will read it. A buyer who sees a post about it in a feed will scroll past. Your launch room is the first link in that chain of recommendation. If they find it useful, they will pass it on, and their recommendations will reach people you could never reach directly.

Choosing the room is the important step. Start with everyone you have worked with in the last few years who fits the niche. Add prospects you have spoken to recently. Add peers whose opinion your buyers respect, who might mention the book. Add anyone who has asked you a question that the book answers. You will likely end up with somewhere between fifty and two hundred names. That is your launch list.

Only after the room has had the book for a few weeks should you widen. By then you will have feedback, perhaps some early responses you can mention, and a better book. The wider launch, through social channels, borrowed audiences and partners, will land on people who are more likely to hear about the book from someone in the room first.