The short answer
For every other member, the AI asks seven questions about your two profiles, gives each answer a weight, adds them up, and ranks the result. The two biggest questions are whether each of you wants to meet the other’s profession. The next two are whether each of you can actually refer clients to the other. Then shared keywords, location and barter.
The seven factors and their weights
| Factor | Max points | What it checks | Data used |
|---|---|---|---|
| Industry alignment | 25 | Their industry is one you said you want to meet. | Your "who I want to meet" industries vs their industry |
| Reverse alignment | 20 | Your industry is one they said they want to meet. | Their "who I want to meet" industries vs your industry |
| They can refer you | 15 | They listed your industry among the ones they regularly refer clients to. | Their "who I refer" list vs your industry |
| You can refer them | 15 | You listed their industry among the ones you regularly refer clients to. | Your "who I refer" list vs their industry |
| Keyword overlap | 15 | Shared services, specialties and client types, 5 points per overlap up to 15. | Keywords, services offered, ideal-client roles and problems |
| Location | 10 | Same city scores 10, same state 5, otherwise 0. | City and state on both profiles |
| Barter | 10 | Both open to trading services scores 5; a direct offering-to-want match scores 10. | Barter offerings and wants, when both are open to it |
Maximum total: 110. Industry alignment and reverse alignment are deliberately the heaviest because a referral partnership needs both people to want it; a pair that scores only on keywords and location is a coincidence, not a partnership.
Two tiers, so nothing is hidden
The top of your list is “strong” matches: pairs with a real two-sided fit. Below them are “discovery” picks: members who scored lower but might still be worth a conversation, shown with the reason. The network is never hidden behind a strict threshold, which matters in a young network where the perfect partner may not have joined yet.
What the priority boost is, and is not
A member who invites others who join earns a temporary priority boost. It adds to the ordering of that member in other people’s lists. It does not change whether a pair is classified as strong or discovery, because that classification uses the unboosted profile-fit score. Nobody can pay for placement.
How the weights adapt
The points above are the starting point. Once a night we compare the matches members accepted against the ones they declined, and adjust how much each factor counts: a factor that is consistently higher in the accepted ones earns a little more weight, one that looks the same in both earns a little less.
Three limits keep that honest. The weights are relative - they always average out to the same total, so this changes the order of your list and never inflates the match percentages. No single factor can move more than halfway up or down from its starting point. And the adjustment is scaled to how much evidence stands behind it, so a handful of declines nudges the weights rather than swinging them.
It learns from accept and decline decisions across the network, never from who paid for anything: the priority boost below is deliberately excluded from the training, because a system that learned “boosted members are better matches” would be paid placement wearing a different name.
What is excluded
- Members you are already engaged with: connected, messaging, booked, or in your pipeline. They leave your suggestions so the slots go to new people.
- Administrators and staff accounts. They are operators, not match candidates.
- Removed accounts.
How and when it refreshes
The whole network is re-scored every six hours. Opening your dashboard tops up your own list at most once every 15 minutes, or immediately when you press Refresh matches. A new member enters everyone’s candidate pool the moment their profile is saved.
What is never used
Your messages, your call notes, your email and your phone number. Matching reads profile fields only. Your 60-second video is for the people you meet, not for the algorithm.