citability.devTalk to usWhen a buyer asks ChatGPT or Perplexity for a company like yours, you are either named or you are not. We measure which, name the competitors cited in your place, build the fixes, and re-test with the identical questions to show whether it worked. No guaranteed rankings, because nobody can honestly offer those.
We test your real buyer questions across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews, then hand you the per-engine citation rate with the actual cited URLs and the competitors named in your place. Every number is a real API test, never a composite score we invented.
Off-site entity authority, answer-ready content structure, a schema graph for entity disambiguation, and a clean AI crawler allowlist including llms.txt. Then we re-test with the identical question set to show whether the citation rate actually moved.
Daily AI-readiness infrastructure checks plus an identical monthly citation re-test. Critical regressions alert within 24 hours. Citation changes wait for a complete comparable manifest, so an engine outage never gets reported to you as a decline.
Full scope and inclusions live on the services page.
Including us. This category is new enough that the gap between rigorous and decorative work is hard to see from the outside, so here are the questions that separate them.
A citation rate with no URLs behind it cannot be verified or acted on. Insist on seeing which pages an engine named, and which competitor pages it named instead.
Several vendors report a composite visibility score derived from proxies rather than from querying the engines. Ask directly which engines were queried, how many times, and on what date.
A before-and-after that changes the questions between runs proves nothing. The only honest lift measurement re-runs the same manifest.
If a provider times out and the agency counts those as zero citations, your report shows a decline that never happened. Ask how incomplete runs are handled.
Nobody can. Answer engines do not publish their source-selection rules and change them without notice. An agency that guarantees a citation number is guessing, and you should treat that as disqualifying.
We cannot guarantee a citation rate, and we cannot attribute revenue to a citation. What we can do is measure a baseline you can check, ship a specific set of fixes, and re-run the identical test so you can see whether the number moved. If it did not move, we will tell you that, because a report you cannot trust when the news is bad is worthless when the news is good.
citability.dev is built and run by Chudi Nnorukam, an AI-Visible Web Architect and the creator of the open-source AVR (AI Visibility Readiness) framework that scores how often AI engines cite a site. The methodology has been featured on freeCodeCamp. You are hiring the person who built the measurement, not a reseller of someone else's dashboard.
A generative engine optimization (GEO) agency works to get your brand named and cited inside AI-generated answers, on engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews. It overlaps with SEO but optimizes for a different outcome: being selected as a source inside an answer, rather than ranking as a link on a results page.
SEO optimizes for position on a results page. GEO optimizes for being extracted and cited inside a generated answer. The inputs differ too. Rankings lean on links and relevance, while citation selection leans on entity clarity, extractable answer structure, schema, and whether AI crawlers are even allowed to read your site.
Ours is $497 for a one-time audit, $1,997 for a four-week implementation sprint, and $39 per month for ongoing monitoring. Retainer-based GEO agencies commonly price in the four-figure-per-month range. Be sceptical of any engagement that cannot tell you what it will measure.
No, and an agency that offers a guarantee is telling you something useful about itself. Answer engines do not publish their source-selection criteria and revise them continuously. What an agency can commit to is a measured baseline, a specific set of fixes, and an identical re-test that shows whether the number moved.
Crawler-access fixes such as robots.txt and llms.txt can change what an engine is able to read within days. Entity authority and content restructuring take longer, because engines have to re-crawl and re-select sources. We re-test at the end of a four-week sprint, which is long enough to see movement but not long enough to call a trend.
If you have someone in-house who can act on the findings, a monitoring tool is usually enough, and we compare six of them including ours. An agency makes sense when the fixes need building rather than only tracking.
The free scan takes about 60 seconds and tells you whether you have a problem worth paying anyone to fix. Start there.