2026 മെയ് മാസത്തിൽ Google പുറത്തിറക്കിയ ആദ്യ AI സെർച്ച് ഗൈഡ് പറയുന്നത്, AEO-യും GEO-യും SEO-യിൽ നിന്ന് വേറിട്ട വിഭാഗങ്ങളല്ല എന്നാണ്. ഇന്ത്യൻ ബിസിനസുകൾ എന്ത് മാറ്റണം, എന്തിന് പണം നൽകുന്നത് നിർത്തണം എന്ന് ഈ ലേഖനം വിശദീകരിക്കുന്നു.
For two years the industry sold AEO and GEO as new disciplines with new budgets. Then, on 15 May 2026, Google published Optimizing your website for generative AI features on Google Search — its first official word on the subject — and said plainly that optimising for AI search is still SEO. If you are a business owner in Kerala who has been quoted a separate fee for "AI search optimisation", this is the article to read before you renew.
What Google Actually Said
The document appeared in Google Search Central on 15 May 2026. Two things about it are notable. First, it is the first time Google has addressed the acronyms AEO and GEO by name in official documentation. Second, its conclusion is deflationary: optimising for generative AI features is "optimizing for the search experience, and thus still SEO."
That single sentence undercuts a large amount of what has been sold in India over the last eighteen months. It does not mean AI search is unimportant — Google spent the rest of the document explaining how its AI surfaces choose sources. It means the mechanism of earning that visibility is the one you already knew: useful content, clear structure, and demonstrable expertise.
The practical translation for a business owner: you should not be buying "SEO" and "AEO" as two separate line items that duplicate the same underlying work. You should be buying one search programme that is measured against both classic rankings and AI citations.
Four Things Google Says You Do Not Need
The guide is unusually direct about tactics that have been marketed heavily and that Google says do nothing for its generative features. If a proposal you are reviewing leans on these, ask hard questions.
- A dedicated
llms.txtfile. Google states it does not usellms.txtfor its generative AI features. It costs nothing to publish and other tools may read it, but it is not a ranking mechanism and should never be the centrepiece of a paid engagement. - "Content chunking" for AI. Splitting pages into machine-sized fragments is not a Google requirement. Good heading structure already does this job for both humans and machines.
- AI-specific rewriting. There is no separate register of prose that AI prefers. Writing that is clear, specific, and answers the question early works for readers and for extraction. Rewriting a page "for AI" and making it worse for humans is a net loss.
- Special AI schema. There is no AI-only structured-data type. The standard vocabulary — Organization, LocalBusiness, Product, Article, FAQPage — is what matters, and it matters because it disambiguates facts, not because it flatters a model.
Note what is not on that list: structured data itself, factual consistency, page speed, and clear information architecture. Those still count. The guide is dismissing AI-branded novelties, not fundamentals.
What Still Decides Whether AI Cites You
Google's position is that the signals driving AI Overview and AI Mode citations are the signals driving organic rankings. In practice, five things do most of the work.
1. A direct answer near the top of each section
If a heading implies a question, the first sentence beneath it should answer that question in plain language. This is the single highest-leverage habit in AI-era content, and it is also simply good writing. Bury the answer in the fourth paragraph and both a hurried reader and an extraction model will miss it.
2. Specificity that cannot be paraphrased away
Generic advice is infinitely substitutable, so a model has no reason to cite you over anyone else. Concrete numbers, named tools, real constraints, and local detail — GST treatment, UPI settlement behaviour, Kerala-specific timelines — give a model something only your page supplies.
3. Demonstrable expertise attached to a real person
E-E-A-T is not a score you can buy. It is the accumulation of author identity, consistent professional presence, and content that shows first-hand experience. Anonymous content competes badly in AI answers because there is nothing to trust.
4. Facts that agree with themselves everywhere
Your phone number, address, hours, and pricing should be identical on your website, your Google Business Profile, and every directory that lists you. Contradictions are a cheap negative signal, and they are the most common technical fault I find on Kerala business sites.
5. Content that is actually reachable
If critical text only renders after a click or a slow client-side fetch, it may not be seen. Keep what matters in the initial HTML. This is old advice; AI surfaces have simply raised the penalty for ignoring it.
Why This Matters More in India Than Almost Anywhere
India is one of the fastest-moving AI-search markets in the world, which makes the question of where to spend unusually consequential. Reported figures for early 2026 put ChatGPT at roughly 180 million monthly users in India and Gemini at around 118 million, with Gemini growing fast on the back of default placement on Android — and Android holds well over 90% of the Indian smartphone market.
That distribution matters strategically. For most Kerala businesses, the largest machine audience reading your website is Gemini inside Chrome on an Android phone, not a specialist answer engine. Optimising narrowly for one chatbot while neglecting the Google surfaces that Gemini draws on is backwards — and Google's guide is essentially confirming that the same work serves both.
The other India-specific factor is language. A meaningful share of local queries are Malayalam or Manglish, and AI surfaces handle them by synthesising from whatever authoritative content exists. There is far less competition for well-structured Malayalam-adjacent content than for English, which is a real and shrinking opportunity.
A Sensible Plan for the Next Quarter
If you want to act on this without rebuilding everything, work in this order.
- Consolidate your budget. Stop paying separately for SEO and AEO if the deliverables overlap. One programme, measured on rankings and citations.
- Audit your top 20 pages for the answer-first pattern. For each H2, check whether the first sentence answers the implied question. This is usually a day of editing with outsized returns.
- Fix factual inconsistencies. Reconcile name, address, phone, hours, and prices across your site, Google Business Profile, and major directories.
- Validate your structured data. Not exotic types — just correct Organization or LocalBusiness, Article on posts, and FAQPage where you genuinely answer questions. Remove any ratings you cannot substantiate.
- Start measuring AI visibility. Track whether your brand appears in AI Overviews and chatbot answers for your core queries. You cannot manage what you never look at.
- Keep publishing depth, not volume. One genuinely expert page beats ten thin ones, and thin content is exactly what AI summarisation makes worthless.
None of this is exotic. That is the point of Google's guide, and it is good news for businesses that were worried they had missed a train.
The Honest Caveat
Google describes how Google works. It does not speak for ChatGPT, Perplexity, or Claude, and those systems retrieve differently — some lean on their own indexes, some on Bing, some on live fetching. So "AEO is just SEO" is precisely true for Google's surfaces and only approximately true elsewhere.
The sensible reading is not that AI visibility is a myth. It is that the work is largely shared, so a separate AI-only invoice is hard to justify, while separate measurement across engines is entirely justified. Traffic patterns have genuinely changed — a large share of searches now end without a click — and that changes what you optimise for. It does not change who gets chosen.
Frequently Asked Questions
Did Google say AEO and GEO are fake?
No. Google said they are not separate disciplines from SEO. Its May 2026 guide describes optimising for generative AI features as "optimizing for the search experience, and thus still SEO." The goal is real and worth pursuing; the claim being rejected is that it requires a distinct methodology and a distinct budget.
Should I delete my llms.txt file?
There is no need to delete it. Google has said it does not use llms.txt for its generative AI features, so it should not be treated as a ranking lever or sold as one. It is cheap to maintain and some non-Google tools may read it, so keeping an accurate one is harmless. Just do not let it substitute for real content and technical work.
Does structured data still help with AI search?
Yes, but not because of a special AI schema. Standard Schema.org markup — Organization or LocalBusiness, Article, Product, FAQPage — makes your facts unambiguous, which helps any system state them correctly. What Google dismissed was the idea of AI-specific schema types, not structured data as a practice.
If AI search is just SEO, why is my traffic falling?
Because visibility and clicks have decoupled. Independent 2026 analyses report that a large majority of Google searches now end without a click to any website, and that pages ranking well alongside an AI Overview see materially lower click-through than before. You can hold or improve your visibility while losing sessions. That is why brand mentions, direct enquiries, and citation tracking now belong in your reporting alongside organic sessions.
What should a Kerala small business do first?
Fix factual consistency across your website and Google Business Profile, then rewrite the top of each important page so the first sentence answers the question the heading asks. Those two steps cost very little, help human visitors immediately, and address the majority of what actually determines whether an AI answer can use you as a source.