The answer is the shortlist
An assistant rarely lists ten options. It names two or three, and those are the ones a buyer tries first.
AI visibility audits for B2B software
More software research now starts in ChatGPT, Perplexity and Gemini. Citadence asks those assistants the questions your buyers ask, repeats each one to see past the noise, and shows you who gets recommended, which sources they trust, and what they get wrong about you.
Nothing to install. No access to your accounts. Just your website address.
What’s a good analytics tool for a 10-person SaaS team?
For a small SaaS team, three tools come up most often:
1 g2.com2 reddit.com3 competitor-c.com
Your product was named in 0 of 6 answers to this question.
An assistant rarely lists ten options. It names two or three, and those are the ones a buyer tries first.
Answers change from one ask to the next and differ between assistants. A single screenshot tells you very little.
Assistants lean on review sites, forums and comparison articles. If you’re missing there, you’re often missing from the answer.
Method
Your public pages tell us what you do, who it’s for, and what your plans cost. That becomes the reference we check every answer against.
We write questions the way people type them, not as keyword strings. Discovery questions never include your name, because a new buyer doesn’t know it yet. The set stays fixed, so later audits are directly comparable.
Each question goes to ChatGPT, Perplexity and Gemini through their official APIs, more than once. We keep every answer in full: who was named, in what order, and which sources were cited.
You get the numbers, the three findings that matter most, and the evidence behind each one: the question, the assistant and the exact words it used.
Deliverable
| Assistant | Answers | Mention rate |
|---|---|---|
| ChatGPT | 20 | 15% |
| Perplexity | 20 | 40% |
| Gemini | 20 | 5% |
Finding 1
Perplexity names you, but almost always after Competitor A.
“For most small teams, Competitor A is the safest default…”
Every claim in a report points to the answer it came from. If we can’t show it, we don’t say it.
We only flag a statement as wrong when your own site clearly says otherwise. Anything we can’t verify is labeled that way.
We use the assistants’ public APIs. No scraping of their apps, no fake accounts.
API answers are close to what people see in the apps, not identical. Every report says so.
For founders and marketers at B2B software companies with a good product and a small audience. If assistants already recommend you first, the report will say so, and we won’t try to sell you anything.
How often, and how favorably, AI assistants mention your company when people ask for recommendations in your category. The work around it is also called AEO (answer engine optimization) or GEO.
ChatGPT, Perplexity and Gemini. Google’s AI Overviews are being added.
Closely, but not exactly. The apps personalize answers using history and location. We measure through the official APIs and repeat each question, which shows the general tendency more reliably than any single chat.
Only your website address. There is nothing to install and no access to your analytics or accounts.
You’ll usually have the report within two business days.
Maren is our AI assistant. She drafts our emails and first replies. A person reviews and approves every message before it goes out, and she will always tell you she’s an AI.
Yes. We’re working with a small number of early customers, and we’d like your honest feedback in return.
Send us your website address. We’ll run the audit and email you the report.