Terms
Last updated 5 September 2026
What we learn from you, and what we tell others
We learn from every customer’s results, and that learning improves every other customer’s recommendations — including your competitors’.
That is not a side effect. It is the product. A small site cannot measure its own title changes: a handful of impressions per page per month will not settle anything in a useful time. Pooling is how a new customer gets advice worth having on the first day, before they have data of their own. You benefit from everyone who came before you, on the same terms.
If that trade is not acceptable, this is the wrong product, and we would rather you knew now than found out later.
What leaves your account
Patterns, never particulars. The test is whether a fact would still be true if your site did not exist.
| Pools | Never pools |
|---|---|
| “Titles phrased as a question outperform the click-through curve at positions 5–10” | Your URLs, in any form |
| “A title naming a place beats one naming a category, for local searches” | Your titles, current or proposed |
| The shape of a rule, and how strongly it held | Your queries, impressions, clicks or positions |
| Your brand, domain or account identity |
A competitor must not be able to derive anything about you from what we tell them — including by asking repeatedly and comparing the answers. Two consequences we accept because of that rule: a pattern drawn from too few sites is not published, because with two contributors a pattern is a description of both of them; and we will decline to answer a question whose answer would be a single customer’s data wearing a pattern’s clothes, even when that customer asks it of their own account.
What we cannot promise
Withdrawal is not straightforward, and pretending otherwise would be the dishonest part. If your results have contributed to a pattern, removing them does not unlearn it. What we can promise, and do:
- Delete your account’s data — every metric, finding, recommendation and verdict — on request, within 30 days.
- Stop contributing your future results from the moment you ask, without ending your subscription.
- Not re-derive past patterns from a copy of your data after deletion.
We cannot promise that a pattern already learned will be forgotten. We say so here rather than in an appendix afterwards.
What we do not do
- We do not sell your data, in aggregate or otherwise, to anyone.
- We do not use your content to train a general-purpose model. Replacement copy is written by a third-party model under an agreement excluding training on our traffic; the pooled learning above is our own statistics.
- We do not read anything Search Console does not show us. The grant is read-only, and we cannot change anything on your site with it.
See also Privacy and what this does.