Introduction — It Had Been True

A wrong phone number that had once been right, and the calls nobody could count.

---

The phone call

Kit had been running the directory for six months when the plumber rang.

He was perfectly civil about it, which somehow made it worse. He said he had been told by somebody at the builders' merchant that his listing on Kit's site had the wrong number on it. He had looked. It did.

The number on the listing had belonged to that business. It had been correct for years. Then they had changed telephone provider, taken a new number, updated the two places they thought mattered, and moved on — as everybody does, because nobody keeps a list of every website in the world that mentions them.

Kit had compiled that listing eight months earlier from a public source, cross-checked it against a second source that agreed, and published it. Two independent sources, both agreeing, both wrong, both wrong for the same reason: they had copied it from the same place at roughly the same time, and none of the three of them had looked at it since.

⚠ Kit had done everything a careful person would do. Kit had checked. The check had confirmed something that was no longer true.

---

What it actually cost

Here is the part that stayed with Kit.

Somebody, at some point, had a leak. They searched. They found the listing — which was, in every other respect, a good listing, with the right address and the right categories and a decent description. They rang the number.

And then they rang somebody else.

That is the whole event. It is small, invisible, and it happened an unknown number of times over an unknown number of months. Nobody complained, because the person with the leak had no reason to think anything had gone wrong — they just rang the next plumber. And the plumber did not know, because you cannot notice calls you never received.

⚠ The characteristic failure of this business is not a complaint. It is an absence of a call that nobody can see, on behalf of a person who never agreed to be represented by you, discovered months later by accident, if at all.

Kit could not tell the plumber how many jobs it had cost him. Kit could not tell him how long it had been wrong. Kit could not even tell him with certainty that the number had ever been right, because by then the trail went back through two sources to somewhere neither of them could reach.

What Kit could say — and did, immediately — was: "That's my fault, I've fixed it, and I'm going to tell you exactly where I got it from and when."

---

The thing nobody tells you about directories

The building is not the business.

A directory is genuinely easy to build. The software exists. Categories, maps, search, listing pages, a claim flow — all of it is available, most of it is cheap, and a competent person can have something working in a fortnight.

And that fortnight is the least important part of the whole enterprise.

Because what you have built is not a website. It is a set of assertions about real businesses — this is where they are, this is what they do, this is when they open, this is how you reach them. Several thousand small factual claims, published under your name, about people who mostly did not ask you to.

Every one of those claims was true at some moment. And from that moment it starts decaying:

⚠ Businesses close. Quietly, and nobody tells you.

⚠ Businesses move.

⚠ Numbers change.

⚠ Hours change, seasonally and permanently.

⚠ Owners change, and the new one does something different.

⚠ And the categories a business belongs in change, because they stopped doing bathrooms and started doing kitchens.

⚠ None of that produces any signal on your site. Your listing does not know. It goes on saying the same thing, confidently, with your name at the top of the page.

Which gives this book its central claim:

A directory is not a publishing project with a maintenance overhead. It is a maintenance project with a publishing front end.

---

Why an unmaintained directory is worse than no directory

This is worth being blunt about, because it is the argument that decides whether you should start at all.

A directory with no listings is useless and harmless.

A directory with nine hundred accurate listings is useful.

⚠ A directory with nine hundred listings of which two hundred are wrong is not "mostly useful". It is a thing that sends people to the wrong place, that costs listed businesses work they never knew about, and that does both while looking exactly as authoritative as the accurate version.

⚠ And the user cannot tell the difference. That is the whole problem. There is nothing on a stale listing that looks stale.

Which means the quality of your directory is invisible to everybody — to consumers, to the businesses in it, and, unless you build a mechanism for it, to you.

Every trade has some version of a quality signal. This one has none. Nobody rings to say "your data was correct today." The only feedback you will ever get is the plumber, months late, by accident.

⚠ So accuracy has to be a system rather than an intention. That is what Chapter 19 and Chapter 26 are, and it is why this book treats a verification date as part of a listing rather than as an administrative detail.

---

The second uncomfortable fact

You are publishing about people who did not ask.

This is normal in the industry, it is how essentially every directory has ever been built, and it is not, in most places, unlawful — ⚠ verify that locally; Chapter 14. But it does two things that are worth facing directly at the start rather than at the point when somebody's solicitor writes to you.

One: it makes every error a wrong done to a third party rather than a service failure to a customer. The plumber was not Kit's customer. He had no contract, no relationship, and no way of holding Kit to a standard. He simply had a business that Kit had described inaccurately in public.

⚠ Which means the ordinary commercial pressures that keep businesses honest do not apply here. Nobody withholds payment. Nobody leaves a review. There is no mechanism by which being sloppy costs you anything, in the short term, at all.

Two: it makes the first sales email strange. At some point you will write to somebody, about a listing they did not request, on a site they have never visited, and ask them for money. How you do that decides whether this business is a respectable local publication or a nuisance — and the difference is almost entirely in whether the listing was already good before you asked.

⚠ The rule that falls out of it: never write to a business about a listing you would not be willing to defend, line by line, on the telephone. Section 5.8.

---

The two rules about money

You cannot promise a ranking. Not to an advertiser, not in a proposal, not casually on the phone. Your traffic arrives on terms set by somebody else, those terms change without notice, and directories are structurally exposed to exactly the kind of change that removes them. A promise about position is a refund waiting for a date.

And nobody pays for a listing. They pay for a lead — for the belief that being on your site brings them work. Which is why the first sale is easy and the twelfth month is where directory businesses die: month one is sold on plausibility, and month thirteen is renewed on evidence.

⚠ Which is why the operating number in this book is not traffic and is not listings. It is DEMONSTRABLE LEADS PER PAYING LISTING — the thing you can put in front of a business owner in month eleven, and the only thing that makes the renewal conversation short.

⚠ And its companion is VERIFIED-LISTING PROPORTION: what share of your database has been checked recently enough to be trusted. It leads everything else, it is entirely within your control, and almost nobody measures it — which is why almost every local directory is quietly wrong about a fifth of what it says.

---

Who this book is for

Anyone starting a local directory. And particularly three kinds of person.

People who like building things and are about to underestimate the maintenance. The site will take three weeks. The database will take the rest of your life, or until you sell it.

People who assume a directory is a passive income project. It is not passive. It is a small publishing operation with a permanent factual liability, and the moment it becomes passive it starts misleading people.

And people who intend to fill it quickly using automation. Chapter 16 and Chapter 37 are for you, and the short version is this: ⚠ an AI system will generate a plausible, fluent, confident description of a business it knows nothing about, and you will publish it under your name, about a real company, and nobody will tell you it is wrong.

---

How to read it

Read Chapters 5 to 8 first, in order. They are the four rules.

Then read Chapter 19 — verification — before you publish a single listing, and decide there what your levels are and what you will show for an unverified record.

Then Chapters 21 and 22, on reviews and moderation, before you build a review feature. Reviews are the single most dangerous thing you can add to a directory, and the correct answer for many operators is not to have them at all.

Then build Resource 15, the verification card, and Resource 36, the freshness audit. Between them they are the product.

---

What happened to the plumber's listing

Kit corrected it in about four minutes.

Then Kit did something more useful, which took the following six weeks: went through every listing in the two most active categories, rang each number, and recorded three things against each one — the date, the method, and the result.

⚠ Of about two hundred and sixty listings, nineteen had a number that no longer reached the business. Just over seven per cent. All of them had been "verified" in the sense that Kit had cross-checked them against a second source at compilation, and all of them had been wrong for somewhere between four months and three years.

⚠ Nobody had rung to tell Kit about a single one of them.

That is the number this book exists to prevent, and the mechanism is not care. Kit was careful. The mechanism is a date on every listing, a cycle that comes round, and the discipline to treat a listing without a recent verification date as something you are not entitled to publish.

---

The four rules, once more

1. Every listing is about somebody who did not ask.
2. The data is decaying while you read this.
3. You cannot promise a ranking, to anybody, ever.
4. Nobody pays for a listing; they pay for a lead.

Now go and read Chapter 1.

---

©2026 James Henderson / https://localhandyman.work

Chapter 1 — Why a Local Business Directory Works as a Side Hustle

You are building an index, and it compounds only for as long as it remains true.

1.1 The proposition: you are building an index, not a product

⚠ You are not making anything. You are organising information about things that already exist, and selling access to the attention that organisation attracts.

Which produces four structural advantages:

⚠ No inventory, no premises, no travel. Genuinely location-independent in a way almost nothing else in this series is.

⚠ The work is asynchronous. ⚠ You can do it at seven in the morning or eleven at night, in blocks, around another job. This is the single strongest argument for it as a side hustle.

⚠ Marginal cost per listing approaches zero once the structure exists.

⚠ And it compounds. A listing built well in month two is still working in year three — provided it is still true. Chapter 6.

What you give up in exchange:

⚠ Immediate income. ⚠ There is no version of this where you earn money in week one. Section 1.9.

⚠ Any control over your traffic. Chapter 7.

⚠ And the ability to stop. ⚠ Section 6.3 — an abandoned directory does not become dormant, it becomes misleading, and it does so while still carrying your name.

1.2 What the two audiences are actually looking for

⚠ You have two, they want different things, and confusing them is why most directories serve neither well.

THE CONSUMER wants:

⚠ To find somebody who can do a specific thing, near them, now.

⚠ A working phone number. ⚠ This is genuinely most of it. Chapter 6.

⚠ Enough information to choose between three options without leaving the page.

⚠ And a reason to believe the page.

THE BUSINESS wants:

⚠ Work. ⚠ Not a listing, not a profile, not "visibility" — work. Chapter 8, and everything commercial in this book descends from that sentence.

⚠ To not look worse than their competitors on your page.

⚠ And, quietly, to be able to correct anything wrong about them. Section 5.5.

⚠ Notice what neither of them wants: your site. Nobody has ever wanted a directory. They want the thing on the other side of it, and a directory that forgets this builds features nobody uses.

1.3 Why it is cheap to start and slow to matter

⚠ The entry cost is a domain and some hosting. The barrier is eighteen months of patience, and it is a much higher barrier.

Why it is cheap:

Software to run a directory is widely available and inexpensive. ⚠ This book names none. Section 13.2.

No stock, no equipment, no vehicle.

⚠ The first hundred listings cost only your time.

Why it is slow:

⚠ Search traffic accrues over months, not weeks. And you do not control it. Chapter 7.

⚠ The two-sided problem. ⚠ Businesses will not pay until there is an audience; the audience will not come until there are listings. Section 2.2 — and there is exactly one way round it.

⚠ And the money arrives last. ⚠ A directory earning nothing at month six is not failing. A directory earning nothing at month eighteen is.

⚠ This mismatch — trivial to start, slow to matter — is why so many local directories exist and so few are maintained. Starting is a weekend. Continuing is the whole job.

1.4 The four rules, and where they come from

⚠ Each addresses one structural property of publishing an index of other people's businesses.

⚠ Rule 1 — every listing is about somebody who did not ask. (You are publishing assertions about third parties who have no contract with you.) Chapter 5.

⚠ Rule 2 — the data is decaying while you read this. (Every fact was true at a moment and is ageing from it, with no signal when it stops.) Chapter 6.

⚠ Rule 3 — you cannot promise a ranking, to anybody, ever. (Your traffic is borrowed on terms you did not set.) Chapter 7.

⚠ Rule 4 — nobody pays for a listing; they pay for a lead. (Which is why the first sale is easy and the renewal is not.) Chapter 8.

⚠ Rules 1 and 2 are two halves of the same obligation and are deliberately separated: rule 1 is about consent and rule 2 is about truth. You can satisfy one and fail the other.

1.5 What this business rewards that most side hustles do not

Six things, and none of them are technical:

⚠ Patience across quarters rather than weeks.

⚠ A tolerance for unglamorous repetition. ⚠ Verification is ringing numbers and writing dates down. It is the entire product and it is extremely boring. Chapter 19.

⚠ Local knowledge. Which streets, which trades, which businesses everybody already uses. This is a genuine and underrated advantage.

⚠ The willingness to correct something publicly and quickly. Section 5.5.

⚠ Comfort with being told no by strangers. Chapter 29.

⚠ And the discipline to do maintenance when nothing is wrong — ⚠ which is the only time it can be done, because by the time something is visibly wrong it has been wrong for months.

1.6 What it punishes that others forgive

⚠ Scale without maintenance.

In most businesses, more is better. Here, ⚠ every listing you add is a permanent obligation, and a database you cannot maintain is a liability that grows.

So this business punishes, specifically:

⚠ Bulk imports. Section 16.9.

⚠ Generated content at scale. ⚠ Section 24.5, and Chapter 37 — a fluent description of a business you know nothing about, published under your name.

⚠ Any promise about search position. Chapter 7.

⚠ Reviews added because everybody has them. ⚠ Chapter 21 — the most dangerous feature you can build, and the correct answer for many operators is not to have them.

⚠ And going quiet. ⚠ A directory left alone for a year is not paused. It is wrong, publicly, on behalf of several hundred people who did not consent to it.

1.7 The central inversion of this book

⚠ Most people building a directory think the project is the site.

⚠ The site is three weeks. The database is the business, and it is permanent.

Which means:

⚠ Your quality is invisible. ⚠ Nobody can tell an accurate listing from a stale one by looking at it, including you. Section 6.6.

⚠ There is no feedback loop. ⚠ Nobody rings to say the data was right. The only signal you will ever receive is a complaint, months late, by accident.

⚠ So accuracy cannot be an intention. It has to be a mechanism with a date on it. Chapter 19, Chapter 26, Resource 36.

⚠ And the corollary that decides whether you should start at all: if you are not going to maintain it, do not build it. A directory nobody maintains does active harm to the businesses in it, and it does that harm quietly, for years, with your name on the page.

1.8 Who does well at this, and who struggles

Does well — Resource 1 is the full self-assessment:

⚠ People who will ring a hundred businesses to check a phone number, twice a year, forever.

⚠ People who correct things immediately and without defensiveness.

⚠ People who can hold a long horizon — who are not demoralised by month five.

⚠ People who are genuinely interested in their local area.

⚠ And people who can sell to a stranger without pretending the listing was their idea. Chapter 29.

Struggles:

⚠ People who want to automate the data. ⚠ The automation works and the errors it produces are undetectable. Chapter 37.

⚠ People who measure progress in listing count. ⚠ A thousand listings you cannot maintain is worse than three hundred you can. Section 6.8.

⚠ People who need income this quarter.

⚠ And people who find repetitive maintenance unbearable — which is most people, honestly, and it is worth knowing about yourself before eight months in.

1.9 What the first year actually looks like

⚠ No income figures anywhere in this book. This is a description of shape.

Months 1–3. ⚠ Building, and the first hundred listings by hand — Section 16.6. Slower than expected, and you will be tempted to import. Nobody visits. This is correct and expected.

Months 4–6. ⚠ The first search traffic appears, unevenly, on pages you did not predict. The first claim requests arrive. And the first wrong-detail email — Chapter 33 — which is genuinely useful information arriving in an unpleasant form.

Months 7–9. ⚠ The first sales conversations. Section 29.1 is the awkward one. Some will say yes, and they will say yes on hope rather than evidence, which is a debt you are taking on. Section 8.2.

Months 10–12. ⚠ The first renewals — which is where you find out whether you have a business. And the first full verification cycle comes round, which is where you find out how much of your database has quietly gone wrong.

⚠ The two things that most reliably go wrong in year one: a bulk import that fills the site with plausible errors, and a directory that stops being maintained in month seven because nothing appeared to be happening.

1.10 Chapter summary

⚠ You are building an index, and it compounds — but only for as long as it remains true.

⚠ Consumers want a working phone number; businesses want work. Neither wants your site.

⚠ Trivial to start, slow to matter, and permanent once started.

⚠ The site is three weeks; the database is the business.

⚠ Your quality is invisible to everybody including you, so accuracy must be a dated mechanism rather than an intention.

---

A realistic scenario

Kit's first three months, and the decision that shaped everything after.

⚠ Kit had a working directory in eleven days. Categories, search, maps, listing pages, a claim button. It looked, and worked, better than the two existing directories covering that town.

⚠ It had four listings, all of which were businesses Kit knew personally.

The obvious next step was to fill it.

⚠ Kit found a data source that could have supplied several thousand local business records in an afternoon. It was available, it was affordable, and it would have taken the site from four listings to about four thousand in a single evening.

⚠ Kit spent a week deciding, and then did not do it.

The reasoning, which was mostly borrowed from somebody who had run a directory before:

⚠ "You can't verify four thousand listings. So you'd be publishing four thousand assertions you can't stand behind, about four thousand businesses who haven't asked, and about a fifth of them would be wrong on day one."

⚠ "And you'd never find out which fifth, because nobody will tell you."

⚠ "And when you eventually try to sell to one of them, the first thing they'll do is look at their own listing — and it'll be a scraped, categorised, description-less entry that says almost nothing, and they'll conclude, correctly, that you haven't done anything for them."

So Kit did the first hundred by hand instead.

⚠ It took five weeks of evenings. Roughly twenty-five minutes per listing: find it, check it against two independent sources, ring the number, write an original description of two or three sentences, assign categories properly, and record the date and the source.

⚠ Twenty-five minutes each. A hundred listings. Just over forty hours of unpaid work with nobody visiting the site.

What those five weeks actually produced, which Kit did not anticipate:

⚠ A hundred listings that were genuinely correct on the day they went live — which meant that when the first sales emails went out in month seven, every recipient looked at a page that was accurate, complete and better than what existed anywhere else about them. ⚠ Section 29.2, and it is the whole of why anybody said yes.

⚠ A realistic figure for what a listing costs to build. ⚠ Twenty-five minutes. Which meant Kit could calculate what a thousand listings would cost in hours before committing to them, rather than after. Resource 40.

⚠ **A working knowledge of which data sources agreed with each other and which did not — ⚠ and specifically, the discovery that two sources agreeing does not mean two sources checked, because they had frequently copied from each other. Section 16.4.

⚠ And nine businesses who, during those verification calls, asked what the site was and ended up as the first claimed listings — before there was any traffic at all.

⚠ Kit's note: "The import would have taken an evening and looked like a year's progress. What I'd have had is four thousand pages I couldn't defend and no idea which ones were lying."

---

AI prompts for this chapter

Prompt 1 — The Maintenance Inversion

"I am starting a local business directory website. ⚠ Help me examine a claim: that a directory is not a publishing project with a maintenance overhead, but a maintenance project with a publishing front end — because the site takes a few weeks to build and the database is a permanent factual obligation about real businesses that decays continuously from the moment it is published. ⚠ Test that claim properly and tell me where it is overstated. ⚠ Then work through the consequence I find most difficult: that there is no quality signal in this business at all. Nobody contacts me to say the data was correct, a stale listing looks identical to a fresh one, and the only feedback I will ever receive is a complaint arriving months late by accident. ⚠ Given that, what mechanism could substitute for feedback, and what would it cost me per listing in time?"

Prompt 2 — What the Two Audiences Actually Want

"Help me think clearly about the two audiences of a local business directory and how their wants differ. ⚠ My position: the consumer wants to find somebody who does a specific thing nearby, right now, with a WORKING PHONE NUMBER being most of what they need; and the business does not want a listing, a profile or 'visibility' — they want work, and nothing else. ⚠ Test both. ⚠ Then follow the consequence through: if the business is buying work rather than presence, what does that mean for how I should price, what I must be able to demonstrate, and why the first sale might be easy while the renewal is hard? ⚠ Then help me identify the features directories commonly build that neither audience actually wants, and how I would tell the difference before building them."

Prompt 3 — Suitability Assessment

"Help me assess whether running a local business directory suits me. ⚠ Ask me about specific dispositions rather than general interest: whether I would ring a hundred businesses to check phone numbers twice a year, forever; whether I can work for six months with no income and no visible progress; whether I will correct a public error immediately and without defensiveness; whether I can approach a stranger to sell them a listing they never requested; and whether I can resist bulk-importing data I cannot verify. ⚠ Tell me honestly which of those is most predictive of failure — I believe it is the first, because verification is the actual product and it is extremely repetitive. ⚠ Test that. ⚠ Then describe a realistic first year quarter by quarter in terms of what I will DISCOVER rather than what I will earn, and include no income figures."

Prompt 4 — The Import Decision

"Help me think through a decision I face early in building a local directory: whether to bulk-import several thousand business records or to build the first hundred by hand. ⚠ The case for importing: it is fast, cheap, and produces something that looks like a year of progress in an evening. The case against: I cannot verify what I cannot check, so I would be publishing thousands of assertions about businesses who did not ask, a meaningful proportion would be wrong on day one, I would never learn which ones, and when I later approach one of them to sell, the first thing they will look at is their own thin, scraped listing. ⚠ Work both sides properly rather than agreeing with me. ⚠ Then help me estimate what building by hand actually costs: ask me to time myself on five listings and work out the implication for a hundred, a thousand and five thousand. ⚠ Finally, tell me whether there is a defensible middle position."

---

⚠ AI checkpoint for this chapter

One — did it recommend a specific platform, plugin, host or data source? This book names none, and a recommendation you cannot evaluate is worse than none.

Two — did it treat listing count as a measure of progress? A thousand listings you cannot maintain is worse than three hundred you can.

Three — did it produce traffic or income projections? You control neither, and any figure is an illustration of arithmetic at best.

Four — did it treat a directory as passive income? It is a permanent factual obligation about other people's businesses.

---

Do This Now

1. Write the four rules on one page and keep them where you will see them. Resource 3.

2. Time yourself building one listing properly, end to end. That number decides everything.

3. Look at an existing local directory and check five phone numbers. Ring them.

4. Decide, before you build anything, whether you will still be verifying this in three years.

5. Write down what you will do with the site if you stop. Section 6.3.

---

©2026 James Henderson / https://localhandyman.work