I asked AI for my own service and it named a competitor. What do I do?

Because when the engine went looking for evidence, theirs held up and yours did not, and that is the entire story. An engine assembling a recommendation checks what it can verify: clear public answers about who a business serves, a consistent identity across the web, third-party mentions that agree, and signs the business is alive and current. The competitor cleared those checks. You, in the engine's eyes, were either unreadable, unconfirmed, or invisible.

What the answer specifically does not mean:

• Not that their work is better. The engine has no access to work quality, only to documentation of it.
• Not that they paid. No major engine sells placement inside recommendations.
• Not that it is permanent. Answers are reassembled continuously from current evidence.

The hard part to sit with is that the engine's blindness to quality cuts against established businesses hardest, because decades of excellence stored in client memories and referral networks weighs exactly nothing in a system that reads only public text. The fix is not becoming better than the competitor. It is becoming as verifiable as your work already deserves.

Collect evidence, calmly, before changing anything. The instinct is to panic-publish; the professional move is to run the diagnosis while the moment is fresh.

1. Screenshot everything. The full answer, the reasons, any cited sources. This is your baseline; you will want it in three months when the answers move.
2. Re-run the question across engines. At least ChatGPT, Perplexity, and Google's AI answer. Engines each cite their own largely separate reading lists, so one answer is one jury, not the market.
3. Vary the phrasings. Three or four versions of the buyer's question: with location, with niche, with situation. Note where you appear and where you vanish; the pattern locates the gap.
4. Ask each engine about your business directly. 'Tell me about [your business]' reveals what the engine can verify about you: thin, stale, wrong, and empty are four different problems with four different fixes.
5. Extract every because-clause attached to every competitor named. That list is the rubric for the repair work ahead.

One hour, no changes made, and you now hold a better competitive analysis than most owners ever commission.

Diff your public footprint against the named winner's, then repair in the order the engine cares about. The because-clauses you extracted are the checklist; the work sorts into three passes.

Pass one: make yourself readable. For every reason the engine gave for picking them, ask whether the equivalent fact about you exists in plain, public, extractable form. 'They specialize in X for Y' wins because a page says so directly. Most established businesses fail here first: the expertise is real and the website talks around it.

Pass two: make yourself confirmable. Engines weigh mentions on sites you do not control more heavily than your own claims, and the data shows off-site mentions tracking AI visibility more closely than backlinks. Reviews, a podcast appearance, a professional directory, an industry article: each one is a witness the engine can call.

Pass three: make yourself current. Roughly half of what engines cite was updated in the prior three months. A footprint repaired once and abandoned drifts back out of the answers.

Resist the urge to do all of it everywhere at once. The because-clauses tell you which evidence decides your specific category; match that first.

Weeks to a few months for the first movement, not the years SEO trained you to expect, and the difference comes from how answers get made. There is no index position to climb. Engines reassemble recommendations continuously from what they can currently read and verify, several of them favor recent material outright, and the data shows fresh content dominating citations.

The realistic timeline, assuming steady repair work:

• Early weeks: your direct lookup improves first. 'Tell me about [your business]' returns richer, more accurate answers as engines re-read your clarified site.
• One to three months: you start appearing in some buyer-question answers, usually on the more specific phrasings first, where the candidate pool is thinner.
• Beyond three months: presence stabilizes across engines as third-party confirmation accumulates, the slowest and most valuable layer.

Two honest caveats. Wrong information already circulating takes longer to displace than absence takes to fill. And the timeline assumes maintenance, because the competitor who beat you today can go stale tomorrow, and the freshness advantage flips to whoever kept publishing. Re-run your baseline questions monthly and let the screenshots mark the progress.

Then you have met the era's central unfairness, and it is worth staring at directly: engines measure verifiability, not competence, and the two can diverge badly. A mediocre operator with a clear website, active presence, and accumulating reviews will beat a superb operator whose excellence lives entirely offline. The machine is not wrong by its own rules. Its rules just cannot see what you know.

The wrong responses, tempting as they are: disparaging the competitor anywhere public, which reads as noise to engines and desperation to buyers, or dismissing AI answers as broken, which changes nothing about the buyers acting on them, given that users click almost nothing further once a summary answers.

The right response is to weaponize the asymmetry. If you genuinely are better, you hold evidence the mediocre competitor cannot manufacture: harder cases, sharper judgment calls, real outcomes, positions earned from decades of pattern recognition. Documented, that material is unbeatable, because the engine finally gets to see the difference that was always there.

Mapping exactly which evidence gap lets them outrank you, engine by engine, question by question, is what our free AI Visibility Scan is built to do.