Simple pricing, every engine.
Every plan measures ChatGPT, Perplexity, Claude and Gemini. Pay for the depth of sampling you need, not for engines.
Starter
One brand, every major engine.
$99per month
- 1 project
- 50 prompts
- 5 runs per prompt, weekly
- ChatGPT, Perplexity, Claude and Gemini
- Raw answers and sources
- Email and password sign-in
Most teams
Growth
For teams running experiments.
$399per month
- 5 projects
- 300 prompts
- 10 runs per prompt, daily
- App capture where connected
- Experiments with control groups
- Up to 15 seats
Enterprise
SSO, isolation and scale.
Customannual
- Unlimited projects and seats
- SAML and OIDC single sign-on
- Required SSO and two-factor
- Audit log export
- Bring your own connectors
- Dedicated support
Why do runs per prompt matter?
AI engines give different answers to the same question. Ten runs of a prompt tell you how often you're named; one run is a coin toss. More runs mean tighter ranges.
Do you capture the real ChatGPT app or the API?
Both, and every answer says which. Connect a capture provider to collect from the consumer apps; otherwise Witness uses each engine's API with web search.
Can you guarantee we'll appear in AI answers?
No, and nobody honestly can. We measure, show you what drives competitors' visibility, and let you test what works.
Is there a free trial?
New organizations start on a trial with Starter limits. Billing isn't switched on yet.