What Google's AI Search Guidance Means for Your SaaS Site: AEO, GEO and llms.txt

Quick answer: Google's May 2026 AI search guidance says optimizing for AI Overviews and AI Mode is still SEO. Keep indexing, clear pages, and people-first content. Skip llms.txt and special AI markup for Google Search. They neither help nor hurt rankings there. Measure results in Search Console's Generative AI performance report.

Google published a guide to optimizing for AI features in Search in May 2026, then clarified llms.txt and FAQ rich results. Founders ask the same things: do I need a new "AEO" stack, should I add llms.txt, and is FAQ schema dead?

This piece turns those docs into a keep / stop / ignore checklist for a small SaaS site. It is not an indexing primer. If your pages are not in Google yet, start with why your startup is not showing on Google.

At a glance

What Google actually published

On May 15, 2026, Google Search Central announced the resource. The long-form guidance is the AI optimization guide. Related pages cover AI features appearance and Search updates.

The core claim is simple. Generative AI features in Search are rooted in the same ranking and quality systems as classic results. Retrieval-augmented generation and query fan-out pull from content that already does well for people.

Later clarifications that matter for founders:

• June 15, 2026: Google Search does not use llms.txt for ranking or inclusion in generative AI features
• May 7, 2026: FAQ rich results stopped appearing in Google Search
• June 15, 2026: FAQ rich-result documentation removed

Prefer those primary docs over secondary recaps when you decide what to build.

AEO, GEO, and SEO in plain words

You do not need a separate playbook or "AI SEO" rewrite because the SERP shows an AI Overview or AI Mode answer. Same fundamentals: useful content, crawlable pages, and a site people can trust.

If a vendor sells "AEO packages" that skip indexing or promise guaranteed AI Overview citations, compare the pitch to Google's ignore list below.

Keep / stop / ignore checklist for a small SaaS

Verified against Google's AI optimization guide and AI features docs as of 2026-11-06.

Keep

Stop (or do not start)

Ignore for Google Search (harmless, not helpful there)

Controls that do matter: robots.txt for Googlebot, snippet limits (nosnippet / data-nosnippet / max-snippet / noindex), and Google-Extended for some other systems' training or grounding. Those are separate from llms.txt. See the AI features appearance docs.

llms.txt specifically

llms.txt is a proposed site file meant to help some AI systems find your content. Founders hear about it next to robots.txt and assume Google needs one too.

Google's position, clarified in the AI optimization guide (including the June 15, 2026 note):

• Google Search does not use llms.txt
• Having one does not help your Google rankings
• Having one does not hurt your Google rankings
• You may still publish one if other tools or agents you care about use it

For Google Search alone: optional, not a ranking lever. Do not delay real content or indexing work for an llms.txt file.

FAQ schema after May 2026

In practice:

• You will not get the old FAQ rich-result presentation in Google Search
• Leaving FAQPage markup is fine if it matches visible content; unused structured data is not treated as a problem
• Do not invent FAQ blocks just for schema theater
• Real Q&A on the page can still help humans. That is content quality, not a rich-result trick

If your only FAQ strategy was chasing the rich result, retire that goal. Keep genuine answers on pricing, limits, and setup pages instead.

How to measure AI Overviews and AI Mode in Search Console

Google points site owners to the Generative AI performance report in Search Console. Use it to see how your site appears in generative AI features, rather than guessing from screenshots or third-party tools alone.

Practical habits for a small SaaS:

1. Confirm the property is verified and important URLs are indexed.
2. Open the Generative AI performance report weekly.
3. Pair it with the classic performance report for clicks, queries, and pages outside AI features.
4. Check eligibility cues for Search generative AI features where Search Console surfaces them.
5. Fix crawl, index, and content issues before chasing citations.

Agentic experiences appear in Google's guidance as an emerging area. Treat them as optional until they clearly matter for your category.

Common mistakes and their fixes

Practical next steps for a small SaaS this weekend

1. Pick your 5–10 money pages (home, pricing, docs hub, main use-case pages).
2. Confirm each is indexed and returns a useful snippet for its title or brand + topic.
3. Cut or merge thin duplicates that exist only as keyword variants.
4. Tighten headings and add one concrete example where a page is vague.
5. Leave matching FAQPage markup alone; invent none for vanity. Skip llms.txt unless another system you use asks for it.
6. Bookmark Search Console's Generative AI performance report.
7. Fold leftover hygiene into your SaaS launch checklist. When the product page is ready for directories, submit it on IndieNeed.

Bottom line

Google's AI search guidance is not a call to rebuild your site for machines. Generative AI features sit on top of the same Search systems that already reward clear, crawlable, people-first pages.

Keep indexing and usefulness. Stop scaled thin pages and AI-only rewrites. Ignore llms.txt as a Google ranking tactic. Measure with Search Console, then spend the weekend on the pages that actually sell your product.