Being Cited, Not Just Ranked
When someone asks an AI assistant which developer to hire or how to solve a problem, it reads several sources, synthesises an answer, and names a few of them. If you are not among the sources, you are absent from that conversation no matter where you rank on Google.
The First Thing I Check, and Why
Before anything else I check whether the AI crawlers are permitted to read your site at all. This sounds trivial. It is the single most common reason a business is never cited.
Many site templates, SEO plugins and privacy-minded defaults ship with GPTBot, ClaudeBot, CCBot and Google-Extended disallowed in robots.txt. That is a legitimate choice for a publisher protecting content, and precisely the wrong one for a business that wants recommending. You cannot be cited by a system that is not allowed to read you.
Google-Extended is the one almost everyone misses — it controls AI Overviews and Gemini grounding separately from normal Google indexing, so a site can rank perfectly well and still be excluded from every AI answer.
I found exactly this on my own site while building it out. The permissive robots.txt being served was masking a code default that blocked every AI crawler — one file change away from invisibility.
Can They Actually Read the Page?
The second check is whether your content exists in the served HTML. Many modern sites assemble their pages in the browser, so a crawler that does not execute JavaScript receives an empty shell.
The same applies to interface patterns. Accordions that only render their content when clicked are a common trap — the questions ship in the HTML and the answers do not, so the page advertises questions it never answers. That one is invisible in a browser and obvious in the source.
Making Content Worth Quoting
Models lift specifics. "We offer competitive pricing" cannot be quoted. "A focused SaaS MVP costs ₹4–7 lakh and takes 8–10 weeks" can be lifted, attributed, and reused.
So the content work is concrete: put the direct answer in the opening two sentences, replace vague claims with numbers and named thresholds, and structure each section so it makes sense read in isolation, because that is how it will be read.
Entity Consistency
Models weight sources they can identify confidently. That means the same name, the same description and the same URL across every profile you own, tied together with sameAs markup so the connections are explicit rather than inferred.
This is unglamorous and it is most of the work. Fragmented naming across profiles is why a lot of businesses are readable but never recognised.
What I Deliver
A crawler access audit and fix, a generated llms.txt built from your live content so it cannot drift, a server-render audit, rewrites of your highest-value pages for extractability, entity consolidation across your profiles, and a citation baseline you can re-check yourself.
What I Will Not Claim
There is no way to guarantee a citation, no placement to purchase, and no reliable analytics for this yet. Volume from AI referrals is currently a fraction of organic search.
What GEO does offer is a surface where the ranking factors are far less entrenched. Winning a citation depends much more on having answered a specific question clearly than on years of accumulated backlinks — which makes it one of the few channels where a smaller business competes on close to even terms.