Start with prompts, or you're optimising blind
The first thing we do on a GEO engagement is boring and it changes the whole conversation: we write down 40–100 prompts the way the client's buyers actually ask, run them across every engine, and paste the answers into a sheet. Almost every time, the result surprises the client. Brands that rank well on Google turn out to be absent from shortlist prompts. Brands that thought they were invisible are being named — and described with a service line they dropped two years ago. You cannot fix any of that from a keyword report, because none of these are keywords.
Retrieval doesn't read pages, it reads passages
A generative engine assembles an answer from chunks. That means a page can be excellent and still be unusable: if the paragraph that contains your answer opens with “this approach” and depends on the two paragraphs above it, it can't be lifted cleanly. We rewrite target sections so each one names its subject, makes a specific claim, and carries the number, date or unit that makes it worth quoting. It is the same discipline as writing for a featured snippet, applied across the whole page rather than one answer block.
Entity ambiguity is the most common reason a brand isn't cited
This is the fix that most often produces a fast, visible change. If your site says one thing about your company name and services, your LinkedIn says another, your Google Business Profile has an old address and three directories list a phone number you stopped using, an engine has no confident version of you to attribute anything to — so it names a competitor whose story is consistent everywhere. Cleaning that up feels like admin work rather than marketing, and it moves citation rates more reliably than another 2,000-word blog post. The same principle underpins our answer engine optimisation work, where verifiable entities decide who gets quoted in the AI Overview.
A large part of your AI visibility isn't on your website
When we log the cited sources across a client's prompt set, their own domain is usually a minority of the citations. The rest are comparison articles, industry roundups, review platforms and forum threads — often written by people who have never spoken to the client. If the top three sources an engine consults for “best X in India” don't list you, no amount of on-site polish fixes the recommendation. So part of GEO is unglamorous outreach: getting into the right roundups accurately, correcting outdated descriptions, and earning genuine review presence. White-hat only — the link-package habits that got agencies through the last decade are exactly what makes a source untrustworthy to a model.
Check what your robots file is doing before anything else
We've now found several sites that had quietly blocked GPTBot, PerplexityBot or Google-Extended — in every case a default from a plugin or a security setting nobody revisited. They were running an AI visibility problem they had created themselves. It's a legitimate decision to block AI crawlers if you don't want your content used, but it should be a decision, made deliberately, engine by engine. Make it once, write it down, and stop wondering why you're not being cited.
The honest limits of GEO measurement
We won't pretend this is as clean as rank tracking. LLM outputs shift between sessions and model releases, personalisation muddies results, and two phrasings of the same question can return different vendors. So GEO measurement is sampling: the same prompt set, the same engines, the same schedule, tracked as share of voice over time rather than a position number. That is enough to tell a real trend from noise — and it's far more honest than a screenshot of one lucky ChatGPT answer. If someone shows you a guaranteed AI ranking, ask how they verified it.
GEO, AEO and SEO are one programme, not three invoices
AEO wins the answer inside a search result. GEO wins the recommendation where there is no search result at all. Both sit on the foundation that SEO builds: crawlable, fast, well-structured pages with genuine authority behind them. We've never seen a case where GEO worked on a weak foundation, which is why we run it inside the wider SEO services programme. If you want the split explained properly, read GEO vs SEO vs AEO and our practical walkthrough of how to get cited in ChatGPT, Perplexity and AI Overviews.
Want the strategy without a full retainer? Work directly with our SEO consultant in India and your team executes. Running a business in our home city? See our SEO in Prayagraj. When you're ready, book a free audit and we'll send you the prompt-level baseline — what every major AI engine says about your brand today, and who it recommends instead.