Ask four AIs the same question: your first 10-minute AEO audit
You don't need a tool to find out how AI assistants talk about your brand. You need one good question and ten minutes. Here's the exact process.
Before you buy anything — including us — do this. It's free, it takes ten minutes, and it will tell you more about your AI visibility than any deck.
Step 1 — Write the question your buyer actually asks
Not "tell me about [your brand]." That's the vanity version, and the AI will just paraphrase your website.
Write the question a real prospect types when they don't yet know you exist. For a project-management tool that might be:
"What's the best project management app for a small agency that hates Jira?"
Category, context, and pain — no brand names. That's how buyers actually ask, and it's the only prompt that tells you whether the AI reaches for you on its own.
Step 2 — Ask all four, unchanged
Paste the exact same question into:
- ChatGPT
- Claude
- Gemini
- Perplexity
Don't tweak it between them. The whole point is a fair comparison. For each answer, note three things:
- Mentioned? Are you named at all?
- Cited? Did it link a source — and is any of it yours?
- Tone? "The go-to choice" vs. "an option some teams use" vs. total silence.
Step 3 — Read the competitor answers, honestly
Here's where it stings and where the value is. Whoever the AI did name — go read why. Nine times out of ten it's not that they're better. It's that they're more legible to the model: clearer comparison pages, more third-party mentions, review-site presence, a pricing page that states things plainly.
That gap is your roadmap. AI assistants reward content that answers the question directly, in language a model can lift. Most brands lose here not on product, but on being hard to quote.
Step 4 — Decide what to do with it
You'll finish the audit in one of three states:
- Invisible — nobody names you. Priority one is becoming mention-worthy: comparison content, getting cited by sources the models trust.
- Mentioned but losing — you're in the set but not the recommendation. Priority is sentiment and positioning.
- Winning — great. Now the job is not to slip, because these answers change constantly as models update.
The catch, and where a tool comes in
You just took one snapshot. AI answers are non-deterministic and they drift — ask the same question next week and it can change. A single audit tells you where you stand today; it can't tell you whether you're improving.
That's the part we automate. Seenlark runs this exact audit — your prompts, all four platforms, mentions + citations + sentiment + competitor share of voice — on a schedule, and charts the trend so you can tell a real gain from a lucky roll. You bring your own OpenRouter key; the free tier is there to set it up.
But run the manual version first. Ten minutes, one honest question. It's the cheapest market research you'll do this quarter.