Home/Resources/How Reneka measures AI visibility

Reference · Updated September 2026

How Reneka measures AI visibility.

Every Reneka case study and client report that quotes a citation number uses the method on this page. It is published so the numbers can be checked, compared across clients, and challenged.

The panel

A fixed set of prompts, run the same way every month.

Each engagement starts with a panel of 50 category prompts written the way the client's buyers ask questions: vendor recommendations, comparisons, "how do I" questions, and questions that name the client and its competitors. The panel is fixed for the engagement so month-to-month numbers compare like with like. Prompts are added only when a new service or market is added, and additions are dated.

The panel runs on four engines: ChatGPT, Perplexity, Gemini, and Claude. Each engine is reported on its own. Engines are never averaged into a single score, because they behave differently enough that an average hides the result.

What is counted

Two numbers per prompt, per engine.

The four things recorded for every prompt on every engine.
MeasureDefinitionWhy it is separate
CitationThe engine linked to a page on the client's domain in its answer or source listShows the site was retrieved and used
Brand mentionThe engine named the client in the answer text, with or without a linkShows the buyer saw the name
Competitor setWhich other brands were named in the same answerGives share of voice its denominator
AccuracyWhether what the engine said about the client was correctWrong facts are logged and traced to their source page

Citations and mentions are kept apart because most citations do not name the brand: the Semrush and Growth Memo study of 3,981 domain appearances found 61.7% were ghost citations. Citation share of voice is the share of prompts on which the client was cited; mention share of voice is the share on which it was named. Both are reported against the competitor set.

Limits

What this method cannot tell you.

  • Answers vary between runs. The same prompt can return different sources on different days. Reneka reports month-over-month movement on the full panel, not the result of any single run, and treats small changes as noise.
  • Location matters. The panel runs from Houston, Texas, with a United States location setting. Results in other markets will differ.
  • No engine publishes its internal metrics. Google states that no third party has access to its internal Search data, and the other vendors publish none. Everything here is observed from the outside.
  • Traffic is undercounted. Referrals from assistants are only partly visible in analytics, so citation and mention share are the primary measures and referral traffic is a secondary one.
Related

See also.

The AI search statistics page collects the third-party numbers behind this method. The platform page describes Citation Intelligence, the reporting layer that uses it. The editorial standards cover how Reneka sources published claims.

FAQ

Common questions.

How many prompts does Reneka track?

A fixed panel of 50 category prompts per engagement, run monthly on ChatGPT, Perplexity, Gemini, and Claude, and reported per engine.

What is citation share of voice?

The share of prompts in the panel on which the engine cited a page from the client's domain, reported per engine and against a named competitor set. Mention share of voice, the share on which the client was named in the answer text, is reported separately.

Get cited.
Not just ranked.

A 30-minute call with a Houston-based GEO strategist. We will run a live AI citation audit on your brand and walk you through what ChatGPT, Perplexity, Gemini, and Claude are saying about you today. No pressure, no pitch deck.