When matches look wrong

Sooner or later a match list will surprise you. The obvious candidate is ranked seventh; someone you have never heard of is first. This lesson is a diagnostic routine for that moment, in the order that finds the cause fastest.

Step one: read the request, not the results

Open the request and look at the skills it was raised with. Partners describe work in engagement language (“help with the Meridian carve-out”) and the intake step translates that into ontology skills. If the translation picked “divestiture” but the work is really about “transition service agreements”, the match list is answering the wrong question faithfully. Fixing the request skills fixes the list in seconds.

Step two: check the top candidate’s evidence

Click into the surprising number one. Nine times out of ten their ranking is legitimate: recent verified evidence on exactly the requested skills, plus availability in the request window. The person you expected to see is often lower because they are 80 per cent booked, and availability is weighted into the score. The list is not wrong; it is answering “who fits and is free”, which is the question you actually need answered.

Step three: check the expected candidate’s profile

If your obvious candidate is genuinely missing or buried, open their profile. The usual finding is stale or absent data: the skill the request needs is not on their profile under any related name, or the last evidence is years old. That is a profile fix, and it benefits every future search, not just this one.

When it really is the ontology

Occasionally two skills that practitioners treat as equivalent are distant in the ontology, so near matches are not surfacing. Do not work around this by stuffing extra skills into requests. Raise an ontology query through the platform team; connections added centrally fix the problem for the whole firm. Genuine ontology gaps are the rarest cause on this list, which is why they come last in the routine.