ICP rubric builder — the weights screen.
The in-product UI for tuning your ICP rubric. For the concept (why scoring matters, how to pick weights), see ICP rubric explained. This page is the UI reference.
TL;DR
Drag sliders to set 4 weights summing to 100. Within each dimension, configure custom criteria (e.g., for Firmographic: which industries, employee bands count). Re-score is automatic. Version history lets you A/B test rubric changes.
01Screen overview
The screen has three regions:
- Top: 4 weight sliders summing to 100, with a live preview of how 10 sample accounts re-rank as you drag
- Middle: per-dimension criteria editor (expandable accordion for each of the 4 dimensions)
- Bottom: version history with restore + A/B compare buttons
02Weight sliders
Drag any slider — the others auto-adjust proportionally so the total stays at 100. Pin a slider to lock it (the others split the remainder).
Live preview shows the top 10 sample accounts re-ranking as you drag. If you weight Trigger signals to 40%, you'll see funded/hiring accounts jump to the top in real time. This is how you build intuition for what each weight does.
03Per-dimension criteria
Each dimension has an expandable editor where you configure what scores high:
| Dimension | Criteria you set |
|---|---|
| Firmographic | Target industries, employee bands, geographies, stages |
| Technographic | Required tech, complementary tech, replacement tech, missing tech (negative) |
| Trigger signals | Signal-type weights, freshness decay curve, combo bonus |
| Persona density | Required roles, role-tier weights, minimum verified-email count |
Defaults are reasonable — you usually only need to set 1-2 criteria per dimension to get a much sharper rubric than the off-the-shelf version.
04Re-scoring
Saving any rubric change triggers a re-score of all saved accounts. Takes ~30 seconds for <1K accounts, up to 2 minutes for >10K. You can keep using Mama while re-scoring runs in the background.
Real-time signal-triggered re-scores happen continuously regardless of weight changes.
05Version history and A/B
Every rubric save is versioned. From the version history:
- Restore: roll back to a prior version (useful if a tuning change regresses brief quality)
- A/B compare (Pro+): run two rubric versions side by side for 30 days, see which produces higher reply rates
- Annotate: add a note to a version ("changed Trigger to 40% after Q2 review")
06Permissions
By default, only workspace admins can edit the rubric. ICs see the current rubric in read-only mode. Pro+ can allow specific roles (e.g., Sales Manager) to propose changes that admins approve.