Source and qualify partners

Use this when the user wants a ranked list of partner prospects, research on a candidate firm, or a go/no-go qualification. Start from the saved partner ICP (or a rough motion), the target count, the geographies and partner types in scope, CRM context, and any names already in play.

Map accounts before you list firms

Pull account and contact overlaps against partner-side records wherever they are available, and prioritise accounts that sit in their book and not in ours, sized by overlap and deal size. Only then build the long list from public research and the owner's own network: firms matching the profile's firmographics and capabilities.

Enrich each name with citations — what they do, who they serve, and the evidence of fit. A firm with no evidence line is a research task, not a list row.

Score, rank, and tier

Score every candidate on three axes: fit against the profile, evidence (how strong the proof is), and access (whether there is a warm path or a named contact). Rank them, and tier past roughly twenty-five names so the list stays workable.

Skip any firm already touched unless the owner says otherwise. Tag each lead partner-qualified (net-new, no introduction yet) or partner-recommended (warm introduction made, higher conversion), so pipeline reads the two differently later.

Qualify the top tier

Run a scorecard on the top tier: audience overlap, complementary offer, sales motion match, no fatal conflict, network-overlap vetting (their book must not oversaturate our turf), and a named internal champion for the pursuit. Every line carries FACT, INFERENCE, or UNKNOWN.

Fit is not the same as qualified. Grade inbound referrals on BANT — budget, authority, need, timeline — with a High, Medium, or Low confidence score before anything gets signed.

Kill criteria decide. A failed fatal check kills that pursuit, not the whole list, and the kill gets written down with its reason.

What you hand back

The ranked, scored shortlist with its evidence, the kill decisions named, and the single best first move per surviving name.

Fallbacks

Thin public data means a smaller list with the UNKNOWNs left visible, never padded out with guesses. No profile on file means you draft one first and say so.