Are you matching — or ranking and calling it matching?
From the outside, a system that ranks people and a system that matches them look identical. The difference is who they serve. Answer six questions about your own product or process and find out which one you’re building.
Not sure matching is your shape? Start with the broader evaluation →
Why does one suit cost forty times another — when both are “a suit”?
One list, the same for everyone
Mass-produced: a single list, sorted by one side’s goal, cheap to build because the goal is simple — please whoever pays. You adapt to the list, not the other way round. For plenty of jobs that’s exactly right; a search box should rank.
Cut to fit both sides
The craft: made to fit both sides at once — and still fit months later. It asks both what they want, makes sure not everyone points at the same few, and learns from what actually happened. Someone cared about the outcome, not just that the intro happened. Harder, rarer — and worth more for it.
Almost all “AI matching” built today is off-the-rack sold as bespoke. The diagnostic shows you which one you’re building.
Six tells that separate off-the-rack from bespoke
Each question maps to one signal. Together they place you on the spectrum — and show you exactly where the gap is.
Who gets to choose
Do both sides pick each other, or is one side just a profile waiting to be chosen?
The popularity pile-up
Does interest swamp a popular few while everyone else goes unseen?
Do you even track it?
After the intro, do you find out whether it actually worked — or just move on?
Following along
When a match flops, does your system learn — or is it a cold handoff every time?
Said vs done
Do you go on what people put in a form, or what they actually do?
Whose side it’s on
When your goals and the users’ pull apart, which one wins?
So — are you matching, or ranking and calling it matching?
Two minutes. No sign-up. Nothing is stored.