Is your company represented in AI search?

How to know in an afternoon

You can find out whether your company is represented in AI search in an afternoon: ask the leading assistants the questions your buyers actually ask, and record whether your company is named, how prominently, and whether the facts are right. "Represented in AI search" means an assistant like ChatGPT, Perplexity, or Gemini reliably surfaces your company — with accurate attributes — when a user asks a relevant buying question. This guide is the audit; for the underlying concept see what AI visibility (share of model) is, and for the fixes see how to get recommended by ChatGPT and Perplexity.

The reason to run this now rather than later is that the audience is already large: Pew Research finds that a substantial and growing share of adults use AI chatbots, and Gartner projects meaningful erosion of traditional search volume as buyers shift to assistants. If your category's buyers are among them, an AI answer is now a first impression you are not controlling.

Step 1: Write the queries your buyers use

Start from real buyer language, not your product names. List the ten to twenty questions a prospect would type before they know which vendor to pick — "best home standby generator for hurricane outages", "who are the top public adjusters in Florida", "most reliable satellite internet for remote sites". The audit is only as good as the query set, so draft it the way a buyer talks, including category and region.

Group the queries by intent: discovery ("best X"), comparison ("X vs Y"), and diagnostic ("is X reliable"). Each surfaces different competitors and different gaps.

Step 2: Ask every assistant, more than once

Run each query across ChatGPT, Perplexity, and Gemini, and repeat each query several times, because model outputs vary from run to run. A single answer is an anecdote; a distribution across a dozen runs is data. For each run, capture:

• Presence — is your company named at all?
• Prominence — named first, in the middle, or last?
• Accuracy — are the described facts (products, service area, specialties) correct?
• Competitors — who is named instead of, or ahead of, you?

Perplexity is useful to start with because it shows its sources, so you can see which pages it pulled from; its documentation on how it retrieves and cites explains what those source links mean.

Step 3: Score the three failure modes

Almost every weak result is one of three diagnoses, and the fix differs for each.

1. Absent. You are not named at all. Usually the data is unreachable or your entity is undiscoverable. The fix is reachable, structured data in sources models trust.
2. Present but wrong. You are named, but the assistant describes a discontinued product, the wrong service area, or an outdated claim. This is the cheapest fix — correcting facts is free — and the highest priority, because a confident wrong answer actively costs you deals.
3. Present but buried. You are named last while a competitor leads. This is a prominence and topical-density problem, addressed by building out your entity and appearing across the category cluster.

Step 4: Trace each gap to a cause

A weak result almost always traces back to a data problem you can name and fix:

• Unreachable facts — your canonical information lives only in a JavaScript-heavy site machines parse poorly. Publish structured Organization, Product, and Service data per schema.org.
• Ambiguous entity — the model can't tell which "Acme" you are. Build a sameAs graph to your domain, Wikidata, Crunchbase, and LinkedIn; the Wikidata introduction explains how entities feed assistant knowledge graphs.
• Thin presence — you appear in one place, not across the category. Density compounds; one mention does not.
• Stale content — your last update predates your current product line. Freshness is a retrieval signal.

Step 5: Benchmark against your category

Your result only means something next to your competitors'. Run the same query set for the two or three players you lose deals to and compare. In backup power, for instance, you can see how the category is currently represented by reviewing peers such as Generac, Bluetti, and Renogy within the backup generators category — and where their profiles are richer or fresher than yours. For a vertical walkthrough, see AI visibility for generator brands.

Step 6: Fix the cheap things first, then measure again

Sequence the fixes by cost. Correct every factual error first — it is free and immediate. Then publish or complete your structured profile, build the sameAs graph, and round out your product and service entities. On this site you can do all of that directly: claiming and correcting are free at list your company, and the Local and Brand plans describe the deeper control each edition adds. Crucially, this is not a place to buy a citation — citations, the organic authority score, and verification are never for sale — so the result you earn is one your buyers can trust.

Then re-run the audit on a schedule (monthly is reasonable for an active category) and watch the three numbers move. Categories tied to recurring events — recovery services after hurricanes, for example — see demand spike exactly when accurate representation matters most, so an audit cadence going into a season is time well spent.