Search in 2026 is not the search of 2020. When your prospect asks "who's a good AI automation agency in India?", they don't scroll a list of ten blue links anymore. They read a paragraph that ChatGPT or Perplexity synthesised from three or four sites, then they click through to one of the cited sources.
If your site isn't structured in a way that lets those engines cite you cleanly, you lose the click before the click was ever offered. That's the problem AEO — Answer Engine Optimization — solves.
This is the playbook we ship for every client on our SEO retainer. It works for Google's AI Overviews, Bing Copilot, Perplexity, Claude, ChatGPT search, and Gemini simultaneously — because they all read the same signals, just weighted differently.
The five signals AI engines read
Classic SEO won on backlinks, keyword density, and Core Web Vitals. AI engines still weigh those, but they layer five newer signals on top — and if you don't feed them, you don't get cited.
1. Structured markup (JSON-LD schema) — because AI engines can parse Organization, FAQ, HowTo, and Article schema deterministically, they trust it more than free text.
2. Quotable prose — short, self-contained paragraphs that stand alone as an answer to a specific question. AI engines snip a paragraph, cite it, and move on.
3. llms.txt — a plain-text summary at your domain root telling AI engines what the site is, what you sell, and where the important pages are. Analogous to robots.txt for LLMs.
4. First-party citations — original data, original research, original opinion. If you're just paraphrasing what everyone else already said, AI engines pick the higher-authority source, not you.
5. Structured entity facts — you're the AI automation agency in Delhi. Say it exactly like that on the About page, in schema, in llms.txt, in your OG description. Consistency across surfaces is what lets the engine link the entity to your site.
Step 1 — Add the schema block to every page
The three JSON-LD types that matter most for a service business in 2026: Organization (once, sitewide), FAQPage (on every page with an FAQ block), Service (on every service detail page).
Organization schema tells engines what your entity is. Include your legal name, alternate names, logo URL, description, address (at least the country), founding date, list of services you provide, list of countries you serve, list of authoritative external references (your LinkedIn, GitHub, X, Product Hunt). AI engines fuse these to build a stable identity graph for your brand.
FAQPage schema turns your FAQs into cite-able snippets. Each Q&A pair becomes an entity engines can lift verbatim and attribute to you. On our SEO retainer we add 4-8 FAQs to every page — pricing pages get common-objection FAQs, service pages get decision-help FAQs, blog posts get topical FAQs.
Service schema on every service detail page tells engines exactly what you sell, in what currencies, in what markets. Include areaServed (country codes), offers (with priceCurrency and price), and hasOfferCatalog if you have tiered packages.
Step 2 — Write for quotability, not fluency
The single biggest rewrite we do on client sites is chopping 200-word paragraphs into 40-word ones. AI engines cite the smallest chunk that answers the query. If the sentence "we deliver n8n workflow automation for Indian startups within a fixed 4-week scope from ₹1,60,000" is buried in the middle of a 200-word block, engines skip it. If it's a paragraph on its own, engines quote it.
The pattern that works: one paragraph = one claim = one sentence that could stand alone as a definition. Start with the specific factual claim, then optionally add one supporting sentence. Save the storytelling for blog intros.
Use question phrasings as H2 headings — literally the way a prospect would ask it. "How much does n8n development cost in India?" beats "Our pricing" every time because engines match the H2 to the query directly.
Step 3 — Ship llms.txt at your domain root
llms.txt is a proposed convention (championed by the AI community since 2024) that gives AI engines a Markdown-formatted digest of your site — what it is, who it serves, where the important pages live, and a summary of pricing anchors.
It's not standardised by any single vendor yet. But every major crawler (GPTBot, Perplexity, ClaudeBot, Google-Extended) either already reads it or is signaling that they will. Cost of shipping: 30 minutes. Cost of missing it: engines summarise your homepage badly.
The exact structure we ship for clients: H1 with the brand name, blockquote summary in one sentence, a "What we do" bullet list, a "Pricing anchors" list with concrete numbers, a "Where we operate" section with cities, an "Important pages" section with absolute URLs and a one-line description each. Under 200 lines total. Plain Markdown, no HTML.
Step 4 — Robots.txt: explicitly welcome the AI crawlers
The default robots.txt on most CMSes still uses generic wildcard rules that were designed for Google circa 2015. In 2026 you have to be explicit about who's welcome.
For AI engines you want to appear in, add explicit Allow directives per user-agent: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, Bingbot, Applebot-Extended, DuckDuckBot, MistralAI-User, cohere-ai, YouBot, Diffbot. If a bot isn't allowed explicitly, some crawlers back off.
One gotcha: Google-Extended is separate from Googlebot. Googlebot indexes your site for classic Google search regardless. Google-Extended controls whether your content trains and appears in Gemini and Google's AI features. If you disallow Google-Extended, you disappear from AI Overviews.
Step 5 — Publish first-party research or opinion
AI engines almost never cite the tenth site paraphrasing a topic. They cite the one that added original signal. So every quarter, publish something only you could — a benchmark, a pricing survey, a teardown of a workflow, a lessons-learned post about a specific implementation.
For our own site the first-party content is the case studies and the technical blog posts (like this one). For client sites we help them create equivalent original-content assets: an ROI calculator that reflects their vertical's math, a pricing benchmark from their client roster, or a post-mortem of a specific project.
One caveat: AI engines can now detect low-effort AI-generated content. Ironic but true. Content that reads as ChatGPT default prose gets down-weighted. Human-written or human-heavily-edited content still wins.
Step 6 — Build a consistent entity graph across surfaces
AI engines fuse your identity from every surface that mentions you — your site, LinkedIn, Product Hunt, GitHub, Wikipedia, X, Substack, Crunchbase, review platforms. If your tagline is "AI automation agency India" on one and "AI development studio" on another, engines can't decide which is canonical, so they hedge — which hurts your citation rate.
The fix is boring: pick one 8-12 word entity description. Use it verbatim on your site's OG description, your LinkedIn company tagline, your Product Hunt bio, your Crunchbase description, your GitHub org description. Same words, same order. Rebuild the graph.
For international presence: register the same entity description in the local language of each target market too. AI engines cross-reference across languages.
Step 7 — Monitor what engines cite
Classic SEO tools (Ahrefs, Semrush) don't tell you if ChatGPT is citing you. In 2026 you need one of the new AEO tracking tools — we use a combination of Rank AI, AthenaHQ, and manual checks in ChatGPT / Perplexity / Claude with a curated list of 30 prospect queries every month.
The three metrics that matter: citation rate (how often you appear in an AI answer for your target queries), citation position (are you cited first or fifth), and click-through from citation (are people actually clicking your URL when they see it). Optimise for citation rate first, then position, then CTR.
Common mistakes we see
Blocking GPTBot in robots.txt because a well-intentioned marketing person thought "AI companies scraping our content" was bad. Result: invisible in ChatGPT search forever.
Publishing a hundred thin AI-written articles thinking volume beats quality. Result: engines detect the pattern and down-weight the whole domain.
Adding JSON-LD schema with a validation error. Result: engines ignore the whole block silently — no error, just gone.
Writing FAQ answers that are 300 words each. Result: engines don't cite them because they're too long to embed in a summary. Keep answers to 50-90 words each.
Not updating schema when pricing or services change. Result: engines cite stale info and prospects arrive expecting last year's price.
The bare minimum AEO checklist
If you do only five things this week, do these. One: add Organization schema sitewide with your entity description exactly matching every external profile. Two: add FAQPage schema to your pricing and service pages, five FAQs each, 60-word answers. Three: ship a plain-text llms.txt at your domain root. Four: update robots.txt to explicitly allow the seven major AI crawlers. Five: rewrite one paragraph on your homepage as a self-contained answer to the exact question your best prospect asks.
That checklist takes a single senior engineer eight hours. It typically 3x-es AI citation rate within 30 days for the clients we've done it for.
Book an AEO audit
We run a free 30-minute AEO diagnostic on our SEO retainer clients. It covers schema audit, llms.txt review, entity-consistency check across your top 5 external surfaces, robots.txt review, and a 30-query citation baseline in ChatGPT + Perplexity + Claude.
If you'd rather run this yourself, use the checklist above. If you want us to ship the whole AEO layer — schema, llms.txt, robots.txt, first-party content structure, monthly citation tracking — that's part of our SEO retainer from $799/month or an equivalent INR quote for Indian clients. Book at valuetechsolution.com/contact.
