A list of recommendations is not the product.

Every SEO tool on the market produces a report. The reports are largely correct. They are also, for most businesses, unusable — because the constraint was never knowing what to do. It was having someone with the time, access and confidence to actually do it, and then finding out afterwards whether it worked.

So a small business pays for a tool, receives two hundred recommendations, actions six of them, and cannot tell whether those six made any difference. The tool renews anyway. That gap between knowing and doing is where the money and the frustration both sit.

The bet

SearchDone is a bet that the valuable half of this work is everything after the recommendation: making the change, proving it landed, measuring whether it moved anything, and feeding that result back into the next decision. That is a harder product to build than a reporting dashboard, which is precisely why it is worth building.

It is also a bet that this only works if it is trustworthy. Software that edits a revenue-bearing website is asking for a lot, and the only sane answer is to make every change reversible, every decision explainable, and every claim checkable. Those constraints are written down as operating principles, along with what each one costs us.

Where it is

Early. The analysis engine works and runs on real Search Console data. The machinery that changes a site — approve, publish, verify, roll back — is being built now, starting with the safest possible change type. Nothing on this website describes a capability as working unless it is, and each page carries its own build status.

There are no case studies here, and no customer logos, because there are no customers with results yet. When there are, they will appear with real numbers and the customer's permission, and not before.

Who builds it

SearchDone is a product of No Fear Tech Ltd, a UK company. It is built by a small team that answers its own support email, which is a deliberate choice for as long as it remains possible.

UK-first is a practical decision rather than a patriotic one. GDPR is a first-class constraint here rather than a compliance layer bolted on later, and being close to the market the product serves makes it easier to be honest about what it does.

What would make this wrong

If automated changes turn out to be reliably worse than a competent human making the same changes, the product does not deserve to exist in this form. The measurement is built in from the start partly so that question can be answered honestly — including if the answer is unflattering.