Proof

Your customer asked AI. It recommended someone else.

That conversation happened without you, and nothing in your analytics records it. Proof asks the questions your buyers ask, across all five assistants, and shows you exactly who is being named instead.

Sign inAsk about access

Built and run by WC Media LLC. Not open for self-signup.

Perplexity, asked this morningIllustrative example

“What’s the best standing desk for a small home office?”

For a compact room, reviewers keep returning to the Arbor V2 at 42 inches, which holds a stable frame at the shorter width,1 with the Kestrel Bamboo a close second if you want a warmer surface.2 Both are consistently rated above the budget frames for wobble at standing height.3

Sources named

  • 1nytimes.com/wirecutter
  • 2reddit.com/r/homeoffice
  • 3tomsguide.com

northloopdesks.com — not named in this answer

Being absent is a fact, not a feeling

Every question you track is put to ChatGPT, Claude, Perplexity, Gemini and Grok each week, and each answer is scored three ways: cited with a link, named without one, or missing. The comparison only ever runs against the same questions asked last month, so adding a question can never flatter the trend.

Cited with a linkNamed, no linkNot named
Illustrative example
Illustrative example of how one brand stands across five AI assistants for four tracked questions.
Question trackedChatGPTClaudePerplexityGeminiGrok
Best standing desk under $600ChatGPT: Named, no linkClaude: Not namedPerplexity: Cited with a linkGemini: Not namedGrok: Not named
Standing desk for a small home officeChatGPT: Not namedClaude: Not namedPerplexity: Not namedGemini: Named, no linkGrok: Not named
Are cheap standing desks wobblyChatGPT: Cited with a linkClaude: Named, no linkPerplexity: Cited with a linkGemini: Cited with a linkGrok: Named, no link
Standing desk vs sit-stand converterChatGPT: Named, no linkClaude: Cited with a linkPerplexity: Named, no linkGemini: Not namedGrok: Not named

Then it tells you what to write

Two sources, kept apart on purpose. Breakout searches from Google Trends are worth writing this week and worthless next month. Competitive gaps — terms rivals rank for and you do not — will still be sitting there in ninety days.

Proof clusters both into topics rather than keyword lists, and labels which is which, because the two deserve different urgency and most tools blend them into one undated pile.

Illustrative example
  • Rising now

    Standing desks for very small rooms

    Breakout search, up sharply across the last 30 days.

    Worth writing this week. Breakouts do not wait.

  • Competitive gap

    Bamboo vs laminate desktops, explained

    Three competitors rank on page one. This site does not appear.

    Still there in ninety days. Schedule it, do not rush it.

And whether any of it worked

When a change goes live, Proof records what the page was doing beforehand and keeps reading it afterwards. The record keeps the changes that did nothing and the ones that went backwards, because a list of only the wins is not evidence of anything.

It is the least flattering thing the product does and the reason it is called Proof.

What Proof does with your Google data

Connecting an account grants read-only access to Google Analytics and Search Console. Proof never writes to them and never sees another Google service. Nothing is sold, and nothing is shared with anyone for their own purposes.

Said plainly rather than in the small print: when you ask Proof to research or write about a topic, the search terms behind it — which may come from Search Console — go to the AI models doing the research and the writing. That happens only when you start it, and we never use your data to train anything. The privacy policy names which providers, and is specific about what we have confirmed and what we have not.

Read the full privacy policy