The short answer
AI engines recommend the brands they can most easily verify. When a competitor is named instead of you, it is rarely because their product is better. It is usually because their business is described more consistently across the web, their pages answer buyer questions more directly, and independent sources mention them more often. Those are fixable, observable problems, and this note shows how to check them yourself before spending anything.
What AI is actually doing when it names a competitor
ChatGPT, Perplexity, Gemini and Claude build answers from a mix of what the model learned in training and, for many prompts, sources retrieved live from the web. Either way, the engine is making a confidence judgement: which businesses clearly exist, clearly do the thing asked about, in the place asked about, and are described that way by more than one source. A brand with a clean entity footprint gives the engine very little to guess. A brand with five versions of its name, three addresses and a homepage that never states its category gives it every reason to pick someone else.
| Reason a competitor gets named | What it looks like | First fix |
|---|---|---|
| Clearer entity signals | Their name, category and location are identical everywhere | Align website, Google Business Profile, directories and socials |
| More quotable pages | Pages that answer one buyer question directly, with specifics | Rewrite key service pages answer-first, with proof and numbers |
| More independent mentions | Reviews, local press, partners and directories describe them | Earn reviews and citations where buyers actually check |
| Fresher proof | Recent reviews, dated content, current pricing or availability | Publish dated updates; ask for reviews continuously, not in batches |
| Better question match | Their content matches how buyers phrase the ask | Build pages around real prompt phrasing, not internal jargon |
How to check in under an hour
- Write 20 buyer prompts. Use the words customers use, with your city or service area in half of them. Example pattern: "best [service] in [city]", "top rated [service] near [suburb]", "[service] for [specific need]".
- Run each prompt fresh. Use a new chat or logged-out session where the platform allows it, and run the set in at least two engines, ChatGPT and Perplexity behave differently.
- Record four things per prompt: who is named, whether you appear at all, which sources are cited, and how the named brands are described.
- Read the citations. The sources an engine leans on in your category are your real content roadmap. If a directory or review site keeps appearing, being present and accurate there matters more than another blog post.
- Repeat monthly with the same set. AI answers vary by prompt, model, location and time. Treat any single screenshot as an anecdote; a fixed prompt set, re-run on a schedule, is a measurement.
What not to do
Do not buy fake reviews or paid mentions dressed up as editorial. Engines increasingly cross-check sources, platforms remove them, and the reputational cost lands on your brand, not the vendor. Do not chase one magic prompt, either. Buyers ask the same need twenty different ways; visibility across the set, not a single screenshot, is what produces enquiries. And be sceptical of any agency that guarantees a recommendation: paid placements such as ChatGPT Ads are clearly labelled and sit apart from organic answers, and no one outside the platform controls the organic answer.
Where this leaves you
If your hour of testing shows competitors named consistently while you are absent, you have a measurable problem with a known fix list: entity consistency, quotable pages, citations and reviews, then paid support where it is available. That is exactly what the free Dambru audit formalises, 100+ prompts, the citations behind them, and the fixes in priority order, summarised in 5–7 days.