Prospect Engagement

New prospects and how far each has got — who is stuck at invites, who is onboarded but quiet, and who is ramping

snapshot

Signed up

Prospect roster

Every prospect that signed up in the window, newest first · the status says which step they are stuck on · Invites is accepted / sent, with pending in brackets · Seats is people actually in the account · momentum compares the last 14 days against the 14 before

Needs a nudge

Ordered by how early they are stuck — no invites sent, then invites nobody accepted, then onboarded but never uploaded, then going quiet · the earlier the step, the cheaper the fix

Onboarding ladder

Every prospect in the window and how far down the ladder each got · each step is a subset of the one above · the biggest drop is where onboarding is actually failing

Time to first package

How long after signup the first package appears, for prospects in the window · "never" includes prospects who may simply not have got there yet — the benchmark below corrects for that

Activation benchmark all prospects

Share activated within N days of signup across every prospect on record, not the selected window · this is the line a new cohort should be read against · each point only counts prospects who have had the full N days, so the denominator shrinks to the right — that is exclusion, not failure

Ramp benchmark all prospects

Weeks counted from each prospect's own signup date across every prospect on record · still active is the share sending any package that week — the retention curve · mean packages is average volume per prospect that week · only the first weeks are tracked

Methodology. A prospect is a row in tenants_tenant whose created_at falls in the selected window; engagement is rows in packages_loanpackage belonging to that tenant. Days are Eastern (AT TIME ZONE 'America/New_York'). Internal and test tenants are excluded using the same canonical list as Customer Volume. Processed is package status = 2.

Invites and seats reach the tenant through roles_role.tenant_id — invites_invite and members_member carry only a role_id, so every invite figure depends on that join. Invite status A = accepted, P = pending, C = cancelled; cancelled invites are counted in neither accepted nor pending. Seats counts members with a real user record. Last sign-in is the most recent auth_user.last_login across the tenant's members — a date only, never a person.

Status is the earliest step a prospect is stuck on: No invites (nobody invited yet) → Invite pending (invites sent, none accepted) → Onboarded (someone accepted, but no package yet) → then package behaviour: Dormant (nothing in 28 days), Fading / Growing (last 14 days vs the 14 before, ±20%), Steady otherwise. A prospect younger than 28 days is never labelled Fading or Growing — it has not had time to show a trend.

The two benchmark panels deliberately ignore the window selector and cover every prospect on record. At a 7- or 30-day window a cohort is only a handful of accounts, so a cohort-shaped curve drawn from it would be noise; the benchmark gives the historical norm to compare against instead. Ramp weeks are 7-day blocks from each prospect's own signup date and only the first are recorded, so a long-tenured customer's later volume is in its totals but not in the ramp — the two do not reconcile, by design.

This dashboard counts whether and how much, not who — no user-level, contact or borrower data. Company names are business identities. There is no CRM source, so a tenant that was never a real opportunity looks identical to one that churned; pipeline stage would have to come from HubSpot.