AI search optimization: how to get AI systems to name your business

The short answer: what to actually do

1. Put your text in the HTML, not behind a JavaScript render.
2. Answer the subheading’s question in the first sentence below it.
3. Replace every vague claim with a number, a date, a place, or a source.
4. Use structured data for your business details, services, and prices, and expect no separate AI advantage from it.
5. Get other sites to name your business, because AI systems compose answers from several sources.
6. Measure monthly with a fixed list of questions and track referral traffic from AI tools separately.

Step 1: make sure AI crawlers can read the page at all

Start here. The five steps after this one do nothing for a page that never gets read. Run two checks.

Check 1: is your text in the source? Open the page, press Ctrl+U (Cmd+Option+U on a Mac), and search for a literal sentence from your copy. If it is missing, your content is not in the HTML. From a terminal it is faster: curl -s https://yoursite.com/page | grep "a sentence from your copy". No output means the same problem.

Check 2: does robots.txt block AI crawlers? Request https://yoursite.com/robots.txt and look for GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, and Google-Extended. A Disallow: / there is a decision someone made once and you can reverse deliberately. One misconception worth clearing up: blocking Google-Extended does not keep you out of AI Overviews or AI Mode. Those follow ordinary Googlebot rules. Google-Extended governs the Gemini app and model training.

Server-side rendering, and why a pure JavaScript site stalls here

Googlebot executes JavaScript, though in a second pass and with a delay. The crawlers run by OpenAI, Perplexity, and Anthropic generally take only the HTML the server returns. A site assembled in the browser with React, Vue, or Angular hands those crawlers a near-empty page. Google’s AI search optimization guide points to the standard JavaScript SEO guidance here. Outside Google, the bar is stricter.

The fix is server-side rendering or pre-rendering: the server sends complete HTML, the browser makes it interactive afterward. Our own site ran on a browser-assembled build until July 2026 and was rebuilt with server-side rendering. That same round surfaced a second problem few people check: language detection put browser language ahead of the HTML language attribute, so bots with an English browser locale kept receiving the English text instead of the Dutch. Check not only whether a crawler gets text, but which language version it gets.

Step 2: write answer-first instead of building up

Put the answer in the first sentence under the subheading, in forty words or fewer, with no back-references to the sentence before. AI systems lift a sentence like that out whole. A paragraph that reaches the point in sentence four cannot be quoted on its own and drops out.

Two versions of the same paragraph:

• Building up: “The timeline for an AI project depends on many factors. Below you will find out which ones.”
• Answer-first: “An AI project of ours goes live within a week on average, with a working demo inside 48 hours.”

Phrase subheadings as the question your customer asks rather than as a label. “What does a chatbot cost for a small business?” beats “Pricing.” This page follows the same pattern. Every subheading asks a question or makes a promise, and the sentence below settles it.

Step 3: make facts verifiable

AI systems favor sources that confirm a fact quickly and checkably. A claim without a number, date, or name is dead weight. Walk your main pages and swap:

Add a date when a figure can go stale, and name the source when the figure is not your own. The reverse carries equal weight: a number you cannot back up does not belong on the page. We removed claims like “500+” and “10x ROI” from our own pages in July 2026 because we could not prove them. An AI system weighing your page against other sources punishes an unsupportable figure harder than a modest one that holds.

Step 4: use structured data where it earns its place

This is where a lot of advice overreaches. Google is clear in its own guide: structured data is not required for generative AI features in Search, and no special schema.org markup exists for them. The same document says you do not need an llms.txt file, you do not need to chop content into blocks, and you do not need a separate writing style for AI.

Where structured data does pay: establishing your business as a recognizable entity, and qualifying for rich results in the classic search listings. Four types are worth the effort:

• Organization with name, address, phone number, and sameAs links to your profiles.
• Service or Product with the price that also appears visibly on the page.
• BreadcrumbList, so the page’s place in your site is unambiguous.
• FAQPage on a genuine FAQ block. It guarantees no rich result, so write the questions for the reader.

Write your business name, address, and phone number identically everywhere: on the site, in the structured data, in your Google Business Profile, and in directories. One inconsistent spelling makes your business harder to resolve as a single entity. Google names Merchant Center feeds and the Business Profile in the same guide as routes for products and services to surface in AI answers.

Step 5: build mentions off your own site

An AI answer is assembled from several sources. Most of them are not yours, and your own page at best confirms what already appears elsewhere. This outweighs a classic backlink profile: one competitor we measured in July 2026 held zero organic positions and one referring domain, and still came up repeatedly in ChatGPT.

Building a targeted list takes about half an hour:

1. Ask ChatGPT and Perplexity the buying question your customer would ask, such as “which agency can reactivate my dormant customer database.”
2. Note which sources the answer cites. They are almost always comparison lists, industry directories, and trade press.
3. Get listed on exactly those pages: apply to the directory, approach the list’s editor, offer a case to the trade publication.
4. Where possible, have clients publish a case on their own site with your business name spelled out.

Use the same spelling of your business name across all of them. Four variants produce four weak signals instead of one strong one.

Step 6: measure whether it works, without a rank tracker

Position does not exist here, so three other things get measured instead.

Referral traffic. Open the Acquisition report in GA4 and filter on session source. Look for chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com. The volume is small, though these visitors tend to arrive further along in the buying process than search traffic does.

Crawler visits. Search your server logs or CDN statistics for the user agents from step 1. If GPTBot never appears, the problem is still technical rather than editorial.

Mentions in answers. Write ten questions your customers would ask and repeat them every month. Record the platform, the model, the date, the exact question, and which companies get named. Use a clean session with no chat history. Expect noise: when we repeated the same buying question three days apart in July 2026, eight of the twelve companies named had changed. One measurement says nothing. A trend across four months does. Google points to Search Console for measuring its own AI features.

“AI search optimization” or “AI for SEO”?

These two get mixed up constantly, more so in English than in other languages. This page means the first throughout: getting your business to appear inside the answers AI systems produce. “AI for SEO” is the reverse, using AI tools to do SEO work faster, such as keyword research, draft copy, or clustering pages. Those tools save time. They do not make you visible in AI answers. The six steps above do that.

What this means for your site

Steps 1 and 2 pay off fastest and you can run them yourself or with your developer. Steps 5 and 6 take months and need a standing routine. To find out where your site stands now, we measure crawlability, answer structure, and current AI mentions in one pass.

Frequently asked questions