Why Refract never sees your customers' chats
Conversation Intelligence is built from your own campaigns and results, and from open data we processed ourselves. Here's how, and why we chose it.
A question we hear a lot: if Refract maps the conversations people have in ChatGPT, does it read them? No. OpenAI doesn't share chats with advertisers, and we wouldn't want them to.
Where the insight comes from
- Your own campaign plan: the personas and conversation map built from your brief and website, clearly labelled as modelled.
- Your own results: the Ads Manager reports you import or sync. These are totals by day, ad group, ad and country, with nothing about the people who saw the ads.
- Open data, processed once by us: OpenAI's published usage statistics and a research dataset of conversations people chose to share. They ship with the app as small tables of shares, so using them doesn't contact anyone.
Every number says where it came from
In Conversation Intelligence each number is labelled Observed, Modelled, Estimated, Benchmark or Public data, and “Why?” shows the working. The analysis is done by Refract's own code with fixed rules, not by an AI, so the same data always gives the same answer.
Your clients' data stays theirs
Each client's intelligence is worked out from that client's own data only. Benchmarks are the one shared feature, they are off unless a workspace opts in, and they combine rates from at least five workspaces before anyone sees them. The Trust centre lists exactly what we store, who can see it and the few services that handle it.