How this site works

How we review

Everywhere else, a week of real use and a skim of a pricing page are both called a review, and a reader cannot tell which one they are reading. So they discount both. Here they are different things, labelled, with the evidence attached.

Where we are up to today

FOBIE has published no reviews. All 37 tools on the site are marked listed, not reviewed. A published review here carries a named editor and the evidence behind it, and that work has not been done yet. Saying so is cheaper than the alternative, which is being found out.

The four things a listing can be

Hands-on review

We used it to do the job, and said how long for.

What it tells you: Whether the thing actually works when a real person points it at a real task, and where it gets annoying.
What it cannot tell you: How it behaves at a scale or in an industry we did not try it in.

Research review

We read the documentation and the pricing. We did not use it.

What it tells you: What the tool claims to do, what it costs, and how those claims compare with the alternatives for this job.
What it cannot tell you: Whether any of it holds up in practice. Nobody here has run the job through it.

Reader report

Someone who used it told us what happened.

What it tells you: One real experience, with the context it happened in.
What it cannot tell you: Whether that experience is typical. It is one account, and unless it is marked as checked, we have not confirmed it.

Listed, not reviewed

It is in the directory. Nobody here has assessed it yet.

What it tells you: That the tool exists and roughly what job it is sold for.
What it cannot tell you: Anything about whether it is any good.

A tool being here is not a criticism of it. It means nobody has done the work, which is a fact about us rather than about the tool.

The rules a review has to follow

The evidence is published, not summarised. A hands-on review states the hours, the actual task, whose account it was on and who paid for it. A research review lists every source and states what a desk method cannot tell you. A reader report carries the reporter’s own context.
A review that cannot show its evidence does not publish. This is enforced by the build, not by good intentions: a hands-on review with no hours recorded, or a research review with no sources, fails validation and the site does not deploy.
A vendor-funded account is disclosed on the review itself. Not in a footer, not on a policy page. On the review.
A reader report stays a reader report. It is one person’s experience, published in their words and labelled unchecked unless an editor has been back to them. It is never presented as our verdict.
Reviews are dated and re-dated. Every review shows when it was published and when it was last checked, because a two-year-old verdict about software is a historical document.
Nothing about this is for sale. A vendor cannot buy a review, buy a better one, or buy the removal of one. Here is how we do make money.

If we get something wrong

Tell us and we will fix it. Corrections are made on their merits, whether they come from a reader or from the vendor, and a correction is never conditional on a commercial relationship. Report a mistake, or write to corrections@fobie.ai.