Docs Core concepts ICP rubric explained
Core concepts · 02 of 04

The ICP rubric — scored, not qualified.

Most ICP exercises produce a checklist — fit/not-fit binary. Mama's ICP rubric scores accounts 0-100 across 4 weighted dimensions, which lets you rank instead of filter. The rubric is the single most-load-bearing setting in the product.

Time: 7 min·Updated: 2026-05-25·Audience: every Mama user·Deep dive: long-form essay

TL;DR

Four weighted dimensions out of 100: Firmographic fit · Technographic fit · Trigger signals · Persona density. Default weights are 25/25/25/25 (which produces middling scores for everyone — don't stay there). Push one dimension to 35-40% based on your motion: outbound-heavy → Trigger signals; PLG → Persona density; vertical-specific → Firmographic; stack-replacement → Technographic. Tuning is one screen. Re-scoring is automatic.

01Why scoring beats qualifying

Traditional ICP exercises produce a binary: "fit" or "not fit." The problem is that ~70% of your TAM passes the basic filter — industry, employee count, geography — and you're left with 50,000 "fit" accounts that you can't rank. The binary doesn't help you decide what to do this Tuesday.

Scoring solves this. Instead of "fit/not fit," every account gets a 0-100 score. Now you can rank. The top 200 are what you brief this week. The next 800 sit in the watch list. The rest, you ignore until a fresh signal pulls them up.

The shiftQualification is a wall. Scoring is a queue. Walls let through too many or too few. Queues sort.

02The 4 weighted dimensions

Each dimension is scored 0-100 independently, then weighted into the composite. The weights sum to 100.

DimensionWhat it measuresDefault weight
Firmographic fitIndustry, employee count, geography, company stage — the table-stakes filters25%
Technographic fitTech stack overlap — what they use that your product complements, integrates with, or replaces25%
Trigger signalsActive signals — funding, hiring, exec moves, tech changes — weighted by freshness25%
Persona densityNumber of decision-maker-grade humans at the account on LinkedIn, with verified emails25%

03Why the default 25/25/25/25 is wrong for almost everyone

Even weights produce the middling-score problem: most accounts land between 60 and 75. Nothing differentiates "brief today" from "brief in 60 days." The rubric becomes background noise.

One dimension should always be 35-40% based on your motion. The others drop accordingly. The skew is the point — it's what lets Mama tell you which 20 accounts to act on this week.

04Tuning patterns by motion

Four common motions, with the recommended weight skew per pattern.

Outbound-heavy team
Trigger signals → 40%
Firmographic 20 · Technographic 20 · Persona 20.

Recent signals are your edge. You want the rubric to surface fresh-signal accounts to the top of the queue.
PLG / bottoms-up
Persona density → 35%
Firmographic 25 · Technographic 20 · Trigger 20.

You need many threadable champions, not just one VP. The rubric should reward accounts with depth.
Vertical-specific tool
Firmographic → 40%
Technographic 25 · Trigger 20 · Persona 15.

If your tool only fits one industry, industry fit is the gate. Don't let other dimensions over-promote out-of-vertical accounts.
Stack-replacement
Technographic → 40%
Firmographic 20 · Trigger 25 · Persona 15.

You need the right preceding stack to even pitch. Without that, the account isn't qualified — they have nothing to replace.

05A worked example: Notion Labs

Hypothetical scoring for Notion (Series C SaaS, you sell a data analytics tool):

DimensionRaw score (0-100)ReasoningWeighted (outbound motion)
Firmographic92SaaS, 450 employees, Series C — matches ICP precisely92 × 20% = 18.4
Technographic78Uses Snowflake + dbt — your tool integrates well78 × 20% = 15.6
Trigger signals96Just raised + hired 14 data roles in 30 days96 × 40% = 38.4
Persona density714 decision-makers verified — strong71 × 20% = 14.2
Composite86.6 / 100

Composite 86.6 → top decile. Brief this week.

06What re-scoring does (and when)

Re-scoring is automatic. The trigger:

  • Weight change in the rubric: all saved accounts re-score in ~30 seconds
  • New signal detected on an account: trigger dimension re-scores in real time
  • Persona graph updated: persona density re-scores on next overnight run
  • You manually click "Re-score": from the ICP rubric screen

The score history is kept — you can see how an account's score moved over time. Useful for "is this account warming up?" diagnostics.

07Common mistakes

Leaving weights at 25/25/25/25
Middling scores for everyone. The rubric stops differentiating and becomes wallpaper. One dimension always at 35-40%.
Picking weights based on aspiration, not reality
"We want to sell to enterprise" — but you currently only close mid-market. Score against your real motion, not the one you wish you had. You can re-weight when the motion actually changes.
Treating Persona density as headcount
Persona density is verified decision-makers, not total headcount. A 10,000-person company with 0 verified ICP-relevant decision-makers scores low on this dimension. The rubric is correct.
Re-weighting every week
Stable weights let you compare score-over-time. Re-weighting weekly destroys the temporal signal. Tune once, leave for 60-90 days, re-tune based on closed-won data.
Was this page helpful?