Comparison pages written under rules a test enforces

A page headed with our name and someone else's can be specifically, checkably wrong about another company's product. That is the fastest way to lose a reader who uses that product and knows better than we do. So the comparison pages are data, and the rules they are written under live in a test that fails the build when one is broken.

Rule one: facts are read, dated and linked

A fact about another product is only allowed if someone opened that company's own pricing page and read it, and each fact carries the day it was read. A page with facts shows a Checked on date; a page without any must not show one, and says plainly that nobody here has verified the other product. Every page links the other company's pricing page and admits it may already be out of date.

A price may appear only inside a dated fact, never loose in the prose. For a while, four of the twelve pages had no facts at all and said so. On 23 October all twelve were read on that company's own page, eight of them for the second time. Only one had been rewritten since its last reading, so its facts now follow the new text, and one page now links the more specific pricing page that was actually read.

Rule two: no claimed limitation

Nothing on these pages may say another product can't, doesn't or lacks something. The test scans for a product's name followed, in the same sentence, by a negative, which is exactly how that mistake looks in copy. The left column of every table describes what the category is built for, at its best; every claim about capability is about EasyDooo and was read out of our own code.

Rule three: no invented proof

No user counts, ratings, testimonials or review-site badges. The test carries a list of the shapes these take, including any number followed by users, customers or downloads. It caught us once: a fact quoting another product's free plan as being for a number of users failed it, and the number had to be written as a word.

Rule four: every page concedes

Each page names at least two situations where the other product is the better buy, as a real section with real scenarios: your day genuinely is a to-do list, your work runs through a dozen other services, the work is shared with a team. A comparison page that never concedes is an advertisement, and readers can tell.

What else the test checks

• The plan named for each EasyDooo capability matches what the code actually allows on that plan.
• A product's name may appear only on the comparison pages; anywhere else on the site it fails a separate guard.
• Every comparison page is reachable from the others and from the hub, not only from the sitemap.