How matching works
The pipeline from a request to a ranked shortlist, step by step.
When you run a match on a request, four steps happen in order. Knowing them makes surprising results diagnosable rather than mysterious.
1. Expand the requested skills
The skills on the request are expanded through the ontology. A request for “ESG reporting” also considers people with closely related skills such as sustainability disclosure, with a distance penalty - the further the relationship, the smaller the contribution.
2. Filter the candidate pool
Hard constraints are applied: availability in the request window at the required commitment, grade range, location or mobility, and any conflict rules your firm has configured. People who fail a hard constraint are excluded, not down-ranked. This is why a strong-skills candidate can be absent from a list entirely - check their availability first.
3. Score everyone who remains
Each remaining candidate gets a match score built from skill fit (weighted by evidence source and recency), availability fit, and any configured factors such as continuity with the client or alignment with development goals.
4. Rank and present
Candidates are presented in score order, with the contributing factors shown per person so you can see why each one is there.
What matching does not do
It does not read free-text descriptions, only structured skills. It does not know about conversations outside the platform - a verbal promise to a partner is invisible until it becomes a booking. And it does not make the decision; it produces a defensible shortlist for a human to take into a conversation.