How to Audit What Every Major AI Engine Says About Your Brand

Every brand should know, in writing, what each of the five major AI engines says about it for the prompts that matter.

Most don't. The CMO has never run the audit. The communications team is measuring impressions and pickup. Sales is closing on call notes. Nobody is reading the answer.

The audit takes a structured format. Here it is.

The five engines to test

Run every prompt across all five. They don't always agree — and the disagreements are diagnostic.

— ChatGPT (OpenAI) — Claude (Anthropic) — Gemini (Google) — Perplexity — Google AI Overviews

Skip any of them and the audit is incomplete.

The prompt set — five categories

Brand-direct prompts. "Tell me about [company]." "Who is the CEO of [company]?" "Is [company] a good vendor?" These test accuracy and basic completeness.

Category prompts. "Best [category] vendors." "Top [category] companies." "Who are the leaders in [category]?" These test citation share against competitors.

Competitor comparison prompts. "[Company] vs [competitor]." "Is [company] better than [competitor]?" These test how the engine narrates the competitive set.

Risk prompts. "Issues with [company]." "Problems with [company]." "Lawsuits against [company]." These test what surfaces in the worst-case answer.

Buyer-decision prompts. "Should I buy from [company]?" "Is [company] reliable?" These test the answer that closes or loses a deal.

Twenty prompts minimum. Forty is better. Run them all five times to test stability.

What to score

Across all five engines, score each answer on the five dimensions of AI reputation: Accuracy, Sentiment, Completeness, Consistency, Control. [Read: The Five Dimensions of AI Reputation]

A composite score gives the headline. The dimension breakdown gives the repair plan.

This is the same framework 5W's Reputation Index uses to audit brands and public figures across the five engines. The methodology is reproducible — any team with discipline can run it.

What to extract

— Hallucinations. Specific factual errors, by engine. — Sentiment skews. Which engines frame the brand most unfavorably and why. — Completeness gaps. Which strengths are missing from the model's summary. — Citation sources. When the engine cites a source, capture it. The source list becomes the repair map. — Competitor advantages. Which competitors are over-cited relative to the brand and why.

What to do with the audit

Three outputs: a dashboard the CMO and CEO can read in five minutes, a repair plan with specific source-level interventions ranked by impact, and a monitoring cadence — most brands tend to benefit from re-auditing quarterly, high-risk brands monthly.

Without the audit, AI reputation work is guessing. With it, the work becomes engineering.

See also: Signals That Move AI Reputation · AI Reputation Glossary

No communications firm can guarantee specific outputs inside third-party AI systems. The discipline is shaping the inputs the engines retrieve from — not directing the engines themselves.