What is AI search optimisation?

By Yamm Labs · Updated September 2026 · 6 min read

A marketing head at a life insurer in Mumbai asks Perplexity which term plan suits a 30-year-old non-smoker earning twelve lakh rupees a year. The answer names three insurers, links to two comparison sites and a rival’s claims page, and never mentions her plan. Her plan page ranks well on Google for the same words. The engine did not ignore her brand on purpose. It simply found nothing on her site it could quote with confidence. Fixing that is AI search optimisation.

What does AI search optimisation cover that SEO does not?

SEO aims for a position in a list of links. AI search optimisation aims for a sentence inside a generated answer. The buyer may never click through, so what the engine says about you is the result. That shifts the unit of work from the whole page to the single passage, and the measure from rank to whether your brand is named, cited and described accurately.

The terms GEO and AEO describe parts of this work. Our GEO and AEO FAQ for financial brands separates them.

How do AI engines decide which pages to use?

Each engine works differently, but the public guidance points in one direction: be crawlable, be clear, be consistent. Three things are documented by the engines themselves.

• Google uses its normal index. Google says a page must be indexed and eligible to be shown in Search with a snippet to appear as a supporting link in AI Overviews or AI Mode, and that no special optimisations, AI text files or special schema are needed beyond its search best practices.
• Google searches more than your query. Google says both AI features may use a query fan-out technique, running several related searches across subtopics to build one response. A page that answers the follow-up questions has more chances to be picked.
• ChatGPT search has its own crawler. OpenAI says sites that block OAI-SearchBot will not be shown in ChatGPT search answers, and that blocking GPTBot is a separate setting about model training. A site that blocks every AI crawler in one rule loses both.

Why is AI search optimisation different for banks, insurers and funds?

An engine reads a regulated page in pieces. It can quote the paragraph that states a benefit and skip the footer that qualifies it. The advertising rules in India attach to your communication, not to the engine that summarises it, and we have not found regulator guidance written for AI summaries. So the design decision sits with the brand: write every quotable passage as if it will be read alone.

SIGNALNewsletter by Yamm Labs

The next piece, straight to your inbox.

New writing on BFSI brands, B2B and influencer marketing, and AI search.

Email *

Subscribe

Subscription form is not available at the moment

One click to unsubscribe. Privacy policy

The rules make that concrete. SEBI requires the standard market-risk warning in mutual fund advertisements, and once lengthened the time it must be shown and voiced in audio-visual ads to five seconds because it was being read too fast to follow. IRDAI’s advertisement rules treat an ad as misleading if it hides or underplays the risks of the policy, or discloses its important exclusions and conditions insufficiently. They also require the insurer’s registered name to appear prominently alongside its trade name or logo, mandatory disclosures that are clear, conspicuous and legible, and no wording that disguises or obscures a product’s terms and conditions.

The engine quotes the paragraph, not the page, so the paragraph has to carry the warning.

Which regulator-driven design decisions change the page?

These are the placement rules we apply when a BFSI page is rewritten for AI search. Each one comes from reading the rules above, and our own compliance practice, against how engines extract text.

• Qualifier in the same sentence. “Benefits are not guaranteed” sits in the sentence that describes the benefit, so any faithful quote carries it.
• One risk line per section. Every section that names a scheme, a return or a payout carries its own warning, even if that repeats on the page.
• Registered name in the answer text. The insurer’s or AMC’s registered name appears in the passage, because an engine does not read your logo.
• Filed product names only. Headings and answers use the product name exactly as filed, not a campaign nickname, so the engine repeats the right name.
• Language matches. A Hindi answer section carries its disclosure in Hindi, not an English footer.
• Facts out of the PDF. Charges, eligibility and exit loads move from the brochure PDF into an HTML table, sourced from the filed documents.

Common AI search optimisation mistakes

Most financial brands that stay invisible in AI answers are not doing anything exotic wrong. They are making a handful of ordinary mistakes, each of which the layers above would catch.

Blocking every AI crawler in one rule. A single robots.txt line that blocks all AI bots feels safe. OpenAI treats OAI-SearchBot, which decides whether you can appear in ChatGPT search answers, separately from GPTBot, which relates to model training. Decide each one on purpose.

Keeping the facts in a PDF. Charges, eligibility and exit loads that live only in a brochure PDF are hard for an engine to quote cleanly. Move them into an HTML table, sourced from the filed documents.

Writing for the scroll. A page that builds slowly to its point gives the engine nothing to lift. Put the answer in the first sentence of each section.

Parking the warning in the footer. The engine quotes the paragraph, not the page. A risk line that sits only in the footer will not travel with the claim.

Using campaign nicknames. If the heading says one name and the filed product has another, the engine may repeat the wrong one. Use the filed product name.

Measuring rank instead of answers. A page can rank well on Google and still be missing from AI answers, as the term plan example at the top shows. Track whether your brand is named, cited and described correctly.

A worked example: the term plan page

Go back to the life insurer in the opening. Her plan page ranks, but Perplexity does not quote it. Here is how AI search optimisation would work through the page, layer by layer, using only the steps in this article.

Access. The web team checks robots.txt for each AI crawler, confirms the page is indexed, and checks that no snippet control stops Google showing text from it.

Answer structure. The page is rebuilt around the questions a buyer asks. Each section opens with a one-sentence answer. Eligibility and premium factors move from the brochure PDF into an HTML table, taken from the filed product documents.

Entity consistency. The team lists every place the plan is described: the insurer’s site, aggregator listings, the app store page. Where the plan name or a key fact differs, it is corrected so every source says the same thing.

Compliance in the passage. Every section that mentions a benefit carries its qualifier in the same sentence. The insurer’s registered name appears in the answer text, not only in the logo. Compliance reviews the sections once, as reusable blocks.

Measurement. The same prompt is run again on each engine at a fixed interval, and the team records whether the plan is named, cited and described correctly. The baseline from before the rewrite is the comparison.

An AI search optimisation checklist for a product page

Use this checklist before a BFSI product page goes live, or when you rewrite an existing one. Each item maps to one of the four layers of AI search optimisation in the table above.

1. The page is indexed and eligible to show a snippet in Google Search.
2. robots.txt rules for OAI-SearchBot, GPTBot and other AI crawlers match a decision someone has made and written down.
3. Every section answers one question, and the answer is in its first sentence.
4. Comparisons and charges sit in HTML tables, not images or PDFs.
5. The product name matches the filed name in every heading and answer.
6. The registered name of the insurer, bank or AMC appears in the answer text.
7. Every section that names a return, benefit or payout carries its own qualifier in the same sentence.
8. Hindi or regional-language sections carry their disclosure in the same language.
9. Any structured data matches the text a reader can see.
10. The facts on the page match your listings and third-party pages.

Regulatory items on this list are general practice drawn from the rules cited above. Confirm the current text with your compliance team before you apply them.

How to brief a team on AI search optimisation

AI search optimisation stalls when it is handed to one team. A useful brief names all three owners and what each must deliver.

• Web team: crawler rules, indexing, snippet controls and moving PDF facts into HTML.
• Content team: one question per section, answer-first writing, filed product names and tables for comparisons.
• Compliance: approved qualifier lines and disclosure blocks that can be reused in every section, in every language the site uses.

Attach the prompt list and the baseline results to the brief, so every team can see which layer is broken. Agree the review order before the first draft: compliance reviewing reusable blocks once is faster than reviewing every page from scratch. If you use an outside partner, ask how they handle the disclaimer inside the passage. A partner who plans to fix it in the footer has not understood the problem.

What should a buyer decide before starting?

Start with a fixed list of prompts your buyers type, run them on each engine, and record whether your brand is named, cited and described correctly. That baseline tells you which layer is broken. If crawlers are blocked, the fix is technical and quick. If pages are long and footer-disclaimed, the rewrite is the main cost, and most of that cost is compliance review time, which falls when the disclaimer is designed into each section from the start.

Then decide who owns it. The work spans web, content and compliance, so it stalls when it belongs to one team alone. We run this for financial brands through our AI visibility programme for BFSI. For shorter answers on measurement, schema and engine-by-engine behaviour, see the 100 AI search optimisation questions this page expands on.

People also ask

Is AI search optimisation the same as GEO?

Mostly, yes. GEO, generative engine optimisation, is the common name for getting a brand named and cited in AI answers. AI search optimisation is the plain-language umbrella that also covers crawler access and answer formatting, often called AEO.

Do we need special schema or an llms.txt file to appear in AI answers?

Not for Google. Google says no new machine-readable files, AI text files or special schema are needed for AI Overviews or AI Mode, only that structured data matches the visible text. Clear, crawlable pages matter more than new files.

Can an AI engine drop our disclaimer when it quotes us?

It can, because you do not control the engine’s summary. You control the source. Put the qualifier in the same sentence as the claim, so a faithful quote of that sentence keeps the warning.

How long does AI search optimisation take to show results?

There is no fixed timeline, and any agency that promises one is guessing. Access fixes such as a robots.txt rule can take effect as soon as the engines crawl the page again. Rewrites take longer, because they need compliance review and the engines need to revisit the page. Measure against your own baseline prompts rather than a promised date.

Which AI engines should a financial brand track?

Start with the engines your buyers use and the ones this article names: ChatGPT, Perplexity, Gemini, and Google’s AI Overviews and AI Mode. Run the same fixed prompt list on each one and record whether your brand is named, cited and described correctly. Add engines when your buyers do.

Does AI search optimisation replace SEO?

No. Google says a page must be indexed and eligible to show a snippet before it can appear as a supporting link in AI Overviews or AI Mode. That is SEO groundwork. AI search optimisation adds answer structure, entity consistency and, for regulated brands, the disclaimer inside each passage.

Who should own AI search optimisation inside a bank or insurer?

One named owner, usually in marketing or digital, with web, content and compliance each responsible for their layer. The work crosses all three, so it stalls when one team holds it alone. The owner runs the baseline prompts, tracks the fixes and reports whether AI answers describe the brand correctly.

What drives the cost of AI search optimisation?

Which layer is broken. If AI crawlers are blocked or pages are not indexed, the fix is technical and quick. If product pages are long, PDF-heavy and disclaimed only in the footer, the rewrite is the main cost, and most of that is compliance review time. Designing the disclaimer into each section from the start brings that review time down. Run the baseline prompts first, so the budget goes to the layer that is actually failing in AI answers.