We ran our product on ourselves. The report told us our competitor is winning Perplexity.

After two posts about how messy AI citations are, we decided to stop theorizing and dogfood our own tool.

We pointed BrandScope at our own site — brandscope.dev — and let it run its full scan: 40 questions across 6 engines, one day.

Everything below is what the product generated automatically. We didn't hand-write a word of it.

Three findings made us sit up:

1. Perplexity is recommending our competitor's pricing page. 34/40 questions answered — and for citations it leaned on 14 third-party articles plus, twice, brandscopeai.com, including its pricing page. Someone asking "what's the best AI mention tracker" gets told to look at our competitor's pricing. That's not noise. That's the ranking our brand currently owns in Perplexity.

2. GPT cited our own site — with our own UTM on it. 38/40 mentions, and when it cited brandscope.dev the link carried the tracking parameter we planted ourselves. However you feel about attribution, this is proof it's measurable: the exact source an engine leans on is visible, and it can be engineered.

3. Gemini had sources — but hid every domain. 34/40 mentions, all grounded — and every citation was a masked redirect. A user can see "there's a source" and never learn whose. Visibility without an audit trail is the hardest problem to fix, because you can't even see what to fix.

Then there's the China story: GLM 0/40, Kimi 2/20, Qwen 17/20 (mentioned us, cited nothing). Three engines, three almost-empty rows — a different market, a different blank slate.

None of this is a pitch. It's just what the report said. The uncomfortable version of "know your brand" is that the data comes with sharp edges — and it's better to see them than to keep guessing.

The tool is free to try — a 30-second scan of your own domain, no credit card:
https://brandscope.dev