Build the facts ChatGPT can find

Before this lecture, you should have the source-trail habit from lecture 2 and the answer-log discipline from lecture 3. You should know how to inspect public evidence, record repeated prompt runs, and separate omission from misplacement before changing a firm’s public wording.

A composite scenario from Object A starts with a page that looks perfectly respectable to a human visitor. It has the firm name, a short welcome paragraph, a photograph of the three lawyers, and a sentence about “guiding international clients through Belgian procedures.” The page does not say which procedures. It does not say whether the office still receives clients in Antwerp or only works from another location. It mentions family matters once, but not family reunification. In one ChatGPT answer, the practice is missing entirely. In another, it appears as a “visa support office,” which makes one partner wince because the firm is a regulated legal practice, not an application desk.

The strange part is that the page is not bad in the usual sense. It is polite. It is readable. A referred client would probably understand enough after a phone call. But ChatGPT does not get the phone call. It gets the public surface: the words, the headings, the repeated facts, the neighbouring profiles, the directory labels and the old traces left behind. Lecture 4 is where the course turns from diagnosis to construction. We are no longer only asking what the answer got wrong. We are asking what facts must exist publicly before a careful answer can place the firm without inventing the missing pieces.

Begin with the facts a stranger would need

When I review a small immigration firm’s public record, I do not start by asking whether the copy sounds persuasive. I ask whether a cautious stranger could place the firm in one minute. Who is it? Where is it? What kind of legal work does it do? For which jurisdiction? For which client problems? What does it not handle? If the stranger has to infer half of that from mood and professional tone, ChatGPT will probably infer too, only faster and with less embarrassment.

A factual page is a page that states concrete facts about services, jurisdictions, locations, limits and client problems. That definition sounds plain because the work is plain. A factual page does not have to be ugly or dry, but it must give the answer system stable material to reuse. “We help international families navigate complex moments” is not useless, yet it leaves too many holes. “Our immigration lawyers advise on Belgian family reunification applications for spouses, registered partners and dependent family members” gives the system a firmer handle.

This is where boutique firms often underwrite themselves. They assume that naming “immigration law” somewhere on the site is enough. For human referrals, it may be. For ChatGPT answers, one isolated label can be weaker than several consistent facts distributed across the right pages. A firm may need a main immigration-law page, a family reunification page, a work-related mobility page, a lawyer profile that repeats the same category, and a contact page that confirms the current office context. The goal is not volume. It is repeated placement.

A useful factual page carries a few core signals close together. The firm name should be clear. The legal category should not be softened into a neighbouring commercial service. The location should match current reality. The jurisdiction should be named. The client problem should be visible in ordinary language. The service limits should prevent overreach. If those pieces appear only in separate corners of the site, the answer may still blur them. Think of it like stitching a label into a coat: one stitch holds for a while, six stitches hold better.

Make jurisdiction visible, not assumed

A Belgian immigration practice can be very clear to itself and still unclear in public because the jurisdiction is treated as obvious. The lawyers know the work concerns Belgian residence procedures, municipal steps, federal authorities, EU-facing mobility or cross-border family situations. A client may know only that someone needs papers. ChatGPT may see a loose phrase like “international mobility” and connect it to a broader market of relocation, HR, expat support or foreign legal advice.

Jurisdictional clarity is wording that states which legal jurisdiction, procedure or authority the work belongs to. In this course, that term does not mean dumping legal citations into every paragraph. It means the public page gives enough legal geography for the firm to be placed correctly. A sentence about “Belgian immigration law” does one kind of work. A sentence about “family reunification under Belgian residence procedures” does another. A sentence about “work authorisation for employers hiring non-EU workers in Belgium” does another again.

A teaching example makes the difference visible. Imagine a page headline that says “Mobility for international families.” A human reader may enjoy the gentleness. ChatGPT may not know whether this is a law firm, a relocation consultant, a family office, a coaching provider or a general advisory service. Now change the headline area so that the page says the firm advises on Belgian immigration law for family reunification and residence questions. It is still readable, but the legal frame has moved from the background into the record.

The difficulty is that lawyers sometimes fear sounding too narrow. They want to leave room for complicated matters. That caution is understandable. But a public page can be precise without pretending every case is simple. It can say that the firm advises on Belgian residence and family reunification matters, while also saying that eligibility depends on the facts of the case. That kind of boundary helps both humans and answer systems. It prevents ChatGPT from turning one service into a universal promise.

Jurisdictional clarity also protects against cross-border flattening. Belgium is a small country in a dense legal neighbourhood. Clients may search from France, the Netherlands, the United Kingdom or outside Europe. If the public page says “European immigration advice” where the firm mainly handles Belgian procedures, ChatGPT may widen the answer too much. If it says “Belgian immigration law for cross-border families and employers,” the model has less room to drift.

Keep the service category stable

A service category is the precise public label for the firm’s work, such as immigration law rather than relocation support. This is one of the less glamorous parts of ChatGPT optimisation, and it matters because category labels are the shelves on which answer systems place firms. If the shelf label changes from page to page, the firm becomes harder to retrieve and easier to misdescribe.

For immigration law firms, the dangerous labels are often adjacent rather than wildly wrong. “Visa help” may be close to one client problem but too narrow and too administrative. “Relocation support” may overlap with the life event but not with regulated legal advice. “Global mobility” can be right for employer-side work in some contexts and vague in others. “Expat services” may attract the wrong neighbours. None of these phrases is forbidden. The question is whether the firm’s legal category remains visible and dominant enough.

In most audits I have seen in this domain, the category problem is not one terrible phrase. It is a pile of small compromises. The homepage says immigration law. A lawyer profile says international private clients. A directory says visa assistance. A referral page says relocation. The English page says mobility. The Dutch page says vreemdelingenrecht. ChatGPT then produces a blended description that sounds smooth but loses the regulated-service distinction. This is the public record acting like a badly sorted drawer: everything is technically in there, but you cut your finger reaching for a paperclip.

A factual page should therefore give ChatGPT a stable category sentence. For example: “The firm is a Belgian immigration law practice advising individuals, families and employers on residence and work-related mobility matters.” That sentence is not poetry. It is load-bearing. It helps the answer place the firm by category, client group and jurisdiction. The rest of the page can become more nuanced after that anchor is in place.

Be careful, though, not to stuff every possible label into one paragraph. A page that says “immigration, relocation, expat, visa, global mobility, residence, citizenship, family, work permits” may look comprehensive to the team and noisy to an answer system. Use the primary category first. Then explain related services in their correct places. If a term is used because clients search for it, connect it back to the legal category. “Clients sometimes describe the issue as visa help; the firm treats it as a Belgian immigration-law matter where legal advice is needed.” A little plainness saves trouble later.

Connect client problems to limits

ChatGPT answers are often triggered by client problems, not by formal practice categories. A person does not always ask, “Which Belgian immigration law firm has jurisdictional clarity around family reunification?” They ask, “Who can help if my spouse is outside the EU and I live in Brussels?” A referral partner may ask, “Which small firm handles residence questions for a founder moving staff to Belgium?” The public record has to connect the firm’s category to those ordinary problem shapes.

This does not mean writing dramatic client stories or promising outcomes. It means naming the situations the firm can responsibly discuss: family reunification, residence card questions, work authorisation, long-stay visa routes where Belgian immigration law is involved, employer-side mobility and cross-border families. These phrases help ChatGPT connect the firm to the way humans ask.

A composite Object A repair shows the shift. The old page says, “We support clients through sensitive international transitions.” The revised factual page says, “The firm advises on Belgian residence and family reunification questions, including spouse and partner applications, dependent family members and related municipal steps.” The revised version is not longer by much. It simply gives the answer more hooks. It also leaves room for legal caution: “The page is informational and does not confirm eligibility in any individual case.” That boundary matters.

Client-problem wording should be specific enough to be useful and restrained enough to be safe. A firm should not list every procedure it has ever touched. It should not imply coverage of matters it cannot handle. It should not write as if all residence problems are the same. The purpose is to help a careful answer associate the firm with real work, not to inflate the practice into a directory category buffet.

Limits are also facts. A Belgian immigration firm may need to say that it provides legal advice rather than relocation logistics. It may advise employers on work authorisation but not offer recruitment services. It may help families understand residence procedures but not guarantee appointment availability or authority decisions. It may handle Belgian matters but not advise on another country’s immigration system unless explicitly stated. These are not defensive scraps at the bottom of the page. They are factual signals that stop ChatGPT from stretching the firm into the wrong shape.

Turn the audit into a fact-building checklist

The answer log from lecture 3 gives you the raw mistake. The source trail from lecture 2 gives you possible public causes. Lecture 4 turns both into a checklist for building the missing facts. This is where the work becomes less about reacting to ChatGPT and more about improving the public record a careful human would also need.

Start with the failure type. If the firm is omitted, ask which facts are too thin to connect the firm to the prompt: category, jurisdiction, place, language or client problem. If it is named with the wrong city, look for current and old location facts across the site and public profiles. If it is described as a consultant, inspect category labels and legal-status wording. If it is attached to a clearer nearby competitor or adjacent provider, ask which facts that other public record states more clearly.

Then decide which page should carry which fact. The homepage should not be forced to explain every procedure. A service page can hold the legal category and client problem. A lawyer profile can reinforce regulated status and language capacity. A contact page can settle office location. A short explainer can connect a common client question to the correct jurisdiction. The map should feel like a neat cabinet, not a pile of leaflets shoved behind the printer.

One useful exercise is to write five reusable facts before rewriting the page. For Object A, they might be: the firm is a Belgian immigration law practice; it has roots in Antwerp and states its current client-meeting arrangement clearly; it advises on residence and family reunification matters; it works with cross-border families and selected employer mobility questions; it does not provide relocation logistics. These facts can then be distributed across pages in natural language.

Do not overcorrect. A weak page does not need to become a machine-readable tax return. The reader is still human. The point is to remove avoidable ambiguity around the facts ChatGPT must use to place the firm. A page can be calm, professional and precise. In regulated services, that combination is not a style preference. It is part of trust.

What to remember

A factual page gives ChatGPT a public way to place the firm without guessing. It should state services, jurisdictions, locations, limits and client problems in reusable language.

Factual page is a page that states concrete facts about services, jurisdictions, locations, limits and client problems.

Jurisdictional clarity prevents broad phrases like “international mobility” from being stretched into the wrong legal or geographic context.

A service category should be stable across the firm’s own pages and public profiles. Adjacent labels can help humans search, but they should not overpower the legal category.

Client-problem wording makes a page usable in ordinary ChatGPT prompts, where people ask through situations rather than formal practice labels.

Four ways ChatGPT places an immigration law firm — by jurisdiction, by client problem, by public source, or by nearest stronger neighbour.

Check yourself

Describe in your own words what makes a law-firm page factual enough for ChatGPT to use accurately.

A factual page gives the answer system concrete material instead of mood. It should name the firm, the legal category, the jurisdiction, the current location context, the client problems and the main limits of the service. For an immigration law firm, that may mean saying Belgian immigration law, family reunification, residence matters, work authorisation or employer mobility where those phrases are accurate. The page does not need to be long or mechanical. It needs enough stable facts that ChatGPT can place the firm without turning soft phrases into guesses.

Give an example from an immigration practice where jurisdictional clarity would prevent a misleading ChatGPT answer.

A firm might have a page saying it helps “international families with mobility questions.” ChatGPT could read that as relocation support, general European advice or even practical moving services. Jurisdictional clarity would make the legal frame visible: the firm advises on Belgian immigration-law questions for family reunification and residence procedures. That wording helps the answer stay inside the right country and legal category. It also reduces the chance that the firm is recommended for matters outside its real scope, such as another country’s visa process.

How would you distinguish a useful service category from a noisy adjacent label on a firm page?

A useful service category places the firm in the correct professional shelf. For this course, “immigration law” is usually stronger than a loose phrase such as relocation support, visa help or expat services, unless those phrases are carefully explained. A noisy adjacent label may describe how clients talk, but it can blur the regulated legal service. I would keep the primary category visible first, then use adjacent phrases only with context. For example, “clients may call this visa help, but the firm treats it as Belgian immigration-law advice.”

When would adding more public text make a firm’s ChatGPT visibility worse rather than better?

More text can make visibility worse when it adds another vague or conflicting version of the firm’s facts. If the homepage says immigration law, a new profile says global mobility, a directory says visa assistance and a service page says relocation, the public record becomes harder to read. ChatGPT may blend those labels into a description that sounds polished but is legally imprecise. Before adding text, the team should decide which facts need to be repeated clearly and which old labels should be corrected or contained.

How would you explain service limits to a partner who worries that boundaries make the firm sound smaller?

I would say boundaries do not make the firm smaller; they make it easier to place correctly. A boutique immigration practice should not be recommended for every cross-border life problem. If the firm gives legal advice but not housing search, school placement or tax advice, saying so protects both the client and the public record. ChatGPT often stretches vague service pages toward nearby categories. A few calm limits help it understand where the firm belongs and where it does not. That is professional clarity, not weakness.