Help centre
Answers, and an honest note where there is no answer yet.
Short guidance on getting started, connecting data, reading findings and staying in control. Anything not covered here goes to a person.
In build
There is no support ticketing system yet. Support is email, read by the people building the product.
Getting started
How do I get access?
SearchDone is not open to general signup yet. Access is through the early access cohort, which is a small group of UK sites with Search Console history.
What do I need before starting?
A website you control and a Google Search Console property with some history behind it. A year or more of data makes decay detection meaningful; a brand-new property will produce very thin findings.
Do I need to install anything?
Not for analysis. Search Console access is granted through Google's own consent screen. A plugin is only needed later, when you want SearchDone to make changes — and that is not built yet.
Connecting your data
What permissions does Search Console access need?
Read-only. SearchDone reads performance data and does not submit sitemaps, request indexing or change any setting in your property.
How do I disconnect?
From SearchDone, or from your Google account's third-party access settings. Either works immediately and neither requires contacting us.
Why does my data look incomplete?
Search Console itself aggregates and anonymises: queries below a volume threshold are omitted entirely, and totals do not always reconcile between views. SearchDone reports what the API returns rather than inventing figures to fill the gaps.
Understanding findings
What do the four finding types mean?
Striking distance, CTR underperformance, cannibalisation and content decay. The knowledge base documents exactly how each is detected, including the thresholds.
How reliable are the estimated click figures?
They are ordering, not forecasts. The estimate comes from a blended industry click-through curve, so it ranks opportunities against each other consistently. Treat the order as the useful output and the absolute number as indicative.
I disagree with a finding.
Good — that is what the evidence is for. Every finding shows the rows it came from, so you can check the reasoning. If a detector is systematically wrong for your site, that is worth telling us, because it usually means a threshold needs to be configurable.
Why are there so few findings?
Rows below 50 impressions are discarded as noise, and each detector has a floor to stop trivial issues crowding out real ones. A short list of things worth doing is the intended output.
Safety and control
Can SearchDone change my site right now?
No. Nothing in the product writes to any website yet. When it does, it starts at metadata, requires approval by default, and keeps a reversible snapshot of everything it touches.
How do I stop everything?
Pausing the site, pausing an individual agent, and disconnecting an integration are all designed to be one-step actions that need no support conversation.
What if a change breaks something?
Every change stores the exact prior state and can be rolled back individually or in bulk for a time window. Verification also runs after each change, so a change that broke a page should be caught rather than discovered by you.
Still stuck?
Support is an email to the people who wrote the code. There is no ticket queue and no script.