Skip to content

How to Make Sure ChatGPT Search Can Find Your Growing Business

You have outgrown the one-line fix. Here is the light process that keeps a real site reachable as it changes.

John Cravey with EleviFounder10 min read

A one-person business can manage ChatGPT search visibility with a single line in a file and never think about it again. A growing business cannot, and not because the rule changed. The rule is the same: let OAI-SearchBot in or stay invisible. What changes is that you now have more pages to keep reachable, more people touching the site, and more ways for access to break quietly. So the fix graduates from a one-time task into a small, dull, reliable process.

The plain-English version

ChatGPT search finds you through OpenAI's crawler, OAI-SearchBot. Allow it and your pages are eligible to be surfaced and cited. Block it, or let something else block it, and OpenAI is explicit that you will not appear in ChatGPT search answers (the crawler docs). For a business with one page, allowing it is the whole story. For a business with forty pages across three locations, the harder question is not "are we allowed" but "can the crawler actually reach and read the pages that matter," which is a different and ongoing problem.

Think of it as two gates. The first gate is permission: robots.txt says yes. The second gate is reachability: every page you care about is linked, in your sitemap, loads its content as real text, and is not hidden behind a login or a script that a crawler will not run. A small site passes the second gate by accident. A growing site fails it in specific, findable places, usually the newest or the most important pages.

The two gates, and where growing businesses fail the second one

Most visibility gaps on a real site are reachability problems, not permission problems.

The technical version: your standing checklist

Run this whenever the site changes materially, and review it quarterly regardless. It is deliberately boring.

  1. Confirm permission. Fetch the production robots.txt and confirm OAI-SearchBot is allowed, with no blanket disallow catching it. Be explicit rather than relying on defaults so a future change is less likely to break it silently.
  2. Confirm the sitemap is current. Every important page, especially new service and location pages, should be in your XML sitemap. This is how crawlers discover pages that are not one click from the homepage.
  3. Hunt for orphans. Any page with no internal links pointing to it is at risk. Link new pages into your navigation, related-content sections, and location hubs so they are discoverable.
  4. Check that content is text, not just pixels. Key facts, hours, prices, service descriptions, and locations must exist as real text in the page source, not only inside images or as text that appears only after a script runs.
  5. Verify at the edge. If you use a CDN or WAF, confirm it is not challenging or blocking OAI-SearchBot. Verify real requests against OpenAI's published ranges at openai.com/searchbot.json before you rate-limit anything.
  6. Give changes a day. OpenAI processes robots.txt updates in roughly 24 hours, so verify the day after a change, not the same afternoon.
User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

Sitemap: https://www.yourbusiness.com/sitemap.xml
A growing site's baseline: explicit access for search, an explicit training decision, and a pointer to your sitemap so pages get discovered.

The Sitemap line is doing quiet work: it is how a crawler finds the pages your homepage does not directly link to, which for a multi-location business is most of them. The GPTBot line is your explicit training decision, taken on purpose rather than by default. We cover that decision for a business your size in the GPTBot piece for growing businesses.

This is close to SEO, but not identical

If you already do SEO, most of this rhymes with work you know: reachable pages, a clean sitemap, real text, no accidental blocks. The difference is the second half of the job. Ranking on Google rewards pages that are relevant and authoritative. Being used in a ChatGPT answer additionally rewards pages that are extractable: they state a clear answer to a specific question in plain words the engine can lift and attribute.

The practical implication for a growing business is to lead your key service and location pages with a direct answer to the question a buyer would ask, before the marketing copy. A page that opens with "We serve the greater metro with same-day HVAC repair, typically within four hours, seven days a week" gives an answer engine something to quote. A page that opens with "Welcome to our family business" does not. This is the same move that gets professional firms named, which we take apart in depth in the answer engine optimization cornerstone.

Illustrative shape, not a measured average: permission is usually fine, reachability usually needs work. Audit yours to get real numbers.

Who owns it, and how often

This does not need a new hire. It needs an owner and a cadence. The owner is whoever controls the website and can get a change deployed, usually a marketing lead paired with a developer or agency. The cadence is: check access after any material site change (a replatform, a new location, a security tool), and review the whole checklist once a quarter. Put the quarterly review on a real calendar, because the failure mode here is not doing it wrong, it is forgetting to look until a competitor is showing up in ChatGPT and you are not.

A worked example: a three-location home-services company

Numbers and steps are clearer with a real shape, so picture a heating-and-cooling company with three locations, about forty pages, and a marketing lead who does this alongside four other jobs. Here is the checklist run on that business, and what it typically turns up.

  1. Permission. The robots.txt allows crawlers broadly and does not name OAI-SearchBot at all. That is technically fine, but fragile, so we add an explicit allow for OAI-SearchBot and an explicit training decision for GPTBot. Now a future edit is less likely to break it by accident.
  2. Sitemap. The sitemap lists the homepage and the main service pages but is missing two of the three location pages, because they were added later through a different template. Those two locations were effectively hard to discover. We get them into the sitemap.
  3. Orphans. One of the newer location pages has no links pointing to it from anywhere in the navigation. It exists, but nothing on the site points at it, so it is easy to miss. We link it into the location menu and the footer.
  4. Text, not pixels. The service-area map on each location page is an image, and the list of neighborhoods served lives only inside that image. We add the neighborhood names as real text under the map so "near me" style questions can match.
  5. The edge. The company uses a common CDN with bot protection turned on. We confirm it is not challenging OAI-SearchBot, verifying real requests against the published ranges before trusting anything.
  6. Wait a day, then verify. We make the changes, wait for the roughly 24-hour processing window, and confirm the crawler is reaching the previously orphaned pages.

Notice that the permission problem, the thing everyone worries about, was the least of it. The real gaps were reachability and text: two locations missing from the sitemap, one orphaned page, and the most important local detail locked inside an image. That is the typical profile of a growing business, and none of it would have shown up in a rankings report.

Make the quarterly review a real ritual

The single highest-value habit is putting a recurring reminder on a real calendar to run the checklist once a quarter, plus a rule that any replatform, new location, or new security tool triggers an immediate re-check. The failure mode here is never doing it wrong. It is forgetting to look until something quietly broke months ago. A boring calendar entry beats a clever tool you stop opening.

  • Quarterly: run the six-step checklist top to bottom on every property.
  • On any site change: confirm OAI-SearchBot access the next day, before you consider the change done.
  • When you add a page: link it in and add it to the sitemap the same day, so it is never an orphan.
  • When you add a location: treat it as a new mini-site and run the whole checklist on it.

When you outgrow this too

If you cross into hundreds of employees, several stakeholders who all touch the site, legal review, and a CDN team you do not directly control, this light process needs to become actual governance: change windows, ownership on paper, and monitoring that alerts. That is a different article, the mid-market governance piece. If instead you are smaller than this and the checklist feels like a lot, the leaner version is the micro-business piece.

It is worth being honest about effort so the work does not sprawl. A growing business does not need to rewrite its whole site or chase every page. The returns concentrate in a small set: the top handful of service pages and the location pages that map to real revenue. Fix those first, confirm they are reachable and readable, and leave the long tail for later. Trying to do everything at once is how this initiative stalls and gets quietly abandoned. Doing the ten pages that matter, well, is how it actually ships and shows up in answers.

Before and after: rewriting one service page

Extractability is abstract until you see it on a real page, so here is the same service page written two ways. Imagine a commercial cleaning company with a page for its floor-care service.

The before version opens the way most service pages do. "Welcome to our floor care division. With decades of combined experience and a commitment to excellence, our dedicated team delivers comprehensive solutions tailored to your unique needs. We pride ourselves on quality and customer satisfaction." A reader learns nothing, and an answer engine has nothing to lift. There is no fact in it. It could describe any company in any industry. Asked "who does commercial floor stripping and waxing downtown," a model finds nothing here to extract or cite.

The after version leads with the answer. "We strip, wax, and buff commercial floors, vinyl, tile, and concrete, for offices, warehouses, and retail spaces across the downtown metro. Most jobs are scheduled after hours so your business never closes, and a typical 5,000 square foot floor is stripped and re-waxed overnight. We service recurring maintenance weekly, monthly, or quarterly." Every sentence contains a fact a buyer cares about and a model can quote: the exact services, the surfaces, the customer types, the area, the scheduling, and a concrete time frame. The marketing can come after that, but the first three sentences do the work.

Nothing about the second version is longer or fancier. It is just specific. It answers the question a buyer would actually ask, in plain words, at the top of the page, before any throat-clearing. That single habit, applied to your top service and location pages, is most of the extractability battle for a growing business, and it helps your Google ranking at the same time. Do it on the handful of pages that actually earn revenue, not on all forty at once.

One structural add-on earns its keep at this size: mark up those key pages with basic structured data, so the facts you wrote in plain text are also labeled in a way machines read unambiguously. It is not a substitute for clear writing, and it will not rescue a vague page, but on the pages that already answer a real question it compounds with the prose and removes any doubt about what your business is and where it operates.

The competitive window, briefly

There is a reason to do this now rather than eventually. Most growing businesses in most categories have not audited their AI search visibility at all. Their robots.txt was set by a builder default, their newest pages are half-orphaned, and their best local detail is trapped in an image, exactly like the worked example above. That means the bar to being the most reachable, most extractable option in your category is currently low, and it will not stay that way. The businesses that treat this as a quarterly hygiene task while it is still uncontested get named in answers their competitors do not even know they are missing. The window is not permanent, but it is open, and the work to walk through it is a checklist, not a budget.

Permission plus reachability plus extractability, checked on a cadence, is the whole discipline for a growing business. Get all three right and every page that matters is eligible to be found and cited by the tool your buyers increasingly ask first. Want us to run the audit and hand you the checklist filled in? Run discovery or see what we ship.

Written by
John Cravey
Founder

Founder of Frontend Horizon. Writes most of the long-form work on the FH blog.

Newer post
How to Govern ChatGPT Search Visibility Across a Mid-Market Brand
Older post
How to Show Up in ChatGPT Search as a Micro Business
Keep reading

More from the blog

AI·13 min

AEO for SMEs: Build a Repeatable Answer-Engine Program

Your buyers now ask an AI who to hire before they ever open a search page. AEO is what gets you named. For an SME, the win is a process you can run every month, not a tool you overpay for.

AI·12 min

AI-Assisted Content for SMEs: A Repeatable Draft-and-Edit Workflow

One person prompting an AI when they remember to is not a content process. Here is the repeatable version a small team can actually run and measure.

AI·9 min

Should You Let AI Train on Your Content? A Small Business Guide

You have real content assets now. The train-or-block decision deserves a real answer, made once, deliberately, and written down.