The parts behind each matching score, how much each part counts, and what happens when information is missing
How much the brand is likely to want this person, built from five parts:
| Part | Weight | What it looks at |
|---|---|---|
| Brief fit | 40% | How well the talent meets the brand's criteria. A "must" that is missed blocks the pair. |
| Closeness | 20% | How similar the talent's profile is to the brief, or to talent the brand already liked |
| Price | 20% | The brand's budget against what the talent accepts |
| Quality | 10% | Engagement for their audience size, and other quality signals |
| Speed | 10% | How fast the talent usually answers in past deals |
Two small adjustments apply on top:
How much the talent is likely to want this brand, built from four parts:
| Part | Weight | What it looks at |
|---|---|---|
| Brands they like | 30% | Brands they use, love, want to work with, or have on their allow list. Naming a rival never counts for this brand. |
| Category fit | 20% | Whether their content sits in the brand's category |
| Price | 35% | The budget against their minimum. A budget well above their minimum scores highest. |
| Brand speed | 15% | How fast the brand usually answers in past deals |
Adjustments:
The match is the two scores combined. It is held back by the weaker side. A pair with 95 and 20 gets 33, so one side that does not want the deal pulls the whole pair down.
Based on the time between messages in past deals. A side needs at least two past replies before its speed counts.
Missing facts never count as zero. They are left out of the score, and the confidence drops instead.
These block a pair outright, however high the other scores are:
About one slot in ten in each shortlist goes to a talent ranked roughly 6th to 50th who still scores at least 40. This keeps newer or thinly profiled talent visible, and stops the results from only reinforcing the current ranking.