ICP from closed-won
The last forty closed-won accounts enriched and analyzed for what they share, written as nine rules with weights, reviewed in plain language and versioned as ICP v1.
→ A definition the whole team can read.
⟨ USE CASE · AGENTIC AI FOR TARGETING ⟩
Agentic AI for Targeting
ICP definition, fit scoring and segments that decide who is worth the next dollar. Agents turn your best customers into a scored definition, apply it across the CRM and the market, and keep it honest as the data moves.
ICP FROM EVIDENCESCORED ACROSS THE MARKETKEPT HONEST
readable
ICP as rules
The definition written as rules and weights you can read and edit, versioned on every change.
per row
score and reasons
Every account carries its score and the rules that fired, in the CRM field you name.
fit × timing
ranking
Fit scores placed beside hiring, funding, technology and intent signals so the list leads with who fits and who is moving.
1
approval per write
Scores land in CRM fields as a gated batch showing every record and value.
⟨ THE JOB TODAY ⟩
Where the week actually goes.
01 /
The ICP is a slide
Three bullets from a workshop two years ago. Nobody can apply it to a list, and the list gets built on instinct instead.
02 /
Fit and timing get confused
A perfect-fit account with no signal and a mediocre-fit account hiring furiously get the same score, or no score. The rep picks by name recognition.
03 /
The best accounts were never worked
Somewhere in the CRM are hundreds of accounts that fit better than anything in the current pipeline. Nobody has scored them.
⟨ ICP FIT ⟩CRM · 12,400 ACCOUNTS
⟨ WITH ASTROFABRIC ⟩
What changes when agents carry the work.
ICP from your best customers
Closed-won accounts enriched and analyzed for what they share, size, industry, stack, hiring pattern, and turned into a scoring definition you can read and edit.
similar_companies · run_code
Fit scores everywhere
The definition applied across the CRM, a list or the market, with the score and the reasons written to the row.
list_score → CRM field
Fit beside timing
Fit scores placed beside hiring, funding, technology and intent signals, so the list ranks by who fits and who is moving.
fit × signals → rank
Segments that decide
Accounts grouped into segments with a distinct angle each, sized against the market, and pushed as lists, audiences or territories.
segments → lists · audiences
⟨ HOW IT WORKS ⟩
From objective to delivered rows.
- 01 · OBJECTIVE
Seed with the best customers
Closed-won and best-retained accounts named, or the CRM segment that holds them, as the evidence the ICP is built from.
- 02 · ENRICH
Enrich the seed
Firmographics, technology, hiring pattern and funding filled on every seed account so the analysis runs on complete inputs.
- 03 · DEFINE
Derive the definition
What the winners share computed exactly and written as rules and weights you can read, edit and version.
- 04 · SCORE
Score everywhere
The definition applied across the CRM, a list or the market, with the score and the reasons written to each row.
- 05 · SIGNALS
Add timing
Hiring, funding, technology change and intent signals placed beside fit so the ranking shows who is moving now.
- 06 · DELIVER
Segment and deliver
Accounts grouped into segments with an angle each, written to the CRM on approval and pushed as lists, audiences or territories.
⟨ THE DATA ON EVERY ROW ⟩
IndustryHeadcount bandGeographyRevenue rangeTechnology stackFunding stageHiring patternGrowth trajectoryBuying-intent topicsClosed-won similarityFit scoreRules firedTiming scoreSegmentICP versionScore date
⟨ USE CASES ⟩
The jobs this page was written for.
Score the whole CRM
Twelve thousand accounts scored against the ICP, the distribution charted, and the hundreds that fit but were never worked listed for the team.
→ The best accounts already in the CRM, found.
Inbound signups, scored and routed
This week’s signups reverse-enriched, scored and routed, with the ones inside target companies flagged for sales and the rest sent to nurture.
→ Sales time spent on the signups that fit.
Segments with an angle each
High-fit accounts grouped by stack and size into four segments, each sized against the market and given its own angle, pushed as lists and audiences.
→ Targeting that decides who gets the next dollar.
⟨ BEFORE AND AFTER ⟩
Targeting, before and after
| By hand | With AstroFabric | |
|---|---|---|
| The ICP | Three bullets from a workshop | Rules and weights derived from closed-won, versioned |
| Applying it | Instinct, per rep | Scored across the CRM and the market, reasons on each row |
| Fit and timing | Conflated or absent | Fit beside signals, ranked on both |
| Unworked accounts | Invisible | Listed with scores |
| Explaining a score | Nobody can | The rules that fired, under the version in force |
| CRM fields | Empty or stale | Written on approval, refreshed on a schedule |
⟨ MISSIONS YOU WOULD RUN ⟩
Described in plain language. Delivered end to end.
SCORING & VERIFICATION
Score inbound signups against the ICP
This week’s signups reverse-enriched, scored and routed, with the ones inside target companies flagged for sales.
EnrichmentLead ScoringSlack
ONE MISSION · EVIDENCE ATTACHED
REPORTS & DECKS
ICP fit report on the CRM
The whole CRM scored, the distribution charted, and the accounts that fit but were never worked listed for the team.
Lead ScoringHubSpotCharts
ONE MISSION · EVIDENCE ATTACHED
PROSPECTING
Lookalikes of your best customers → CRM
The ICP applied to the market: a hundred companies that look like the best accounts, scored and pushed.
FirmographicsTechnographicsAttio
ONE MISSION · EVIDENCE ATTACHED
Three of a hundred. The full playbook library ships in every workspace, and anything you can describe becomes a mission.
⟨ WHERE THE RESULT LANDS ⟩
⟨ GOVERNED AUTONOMY ⟩
Safe enough to actually turn on.
⟨ CONTROL PLANE ⟩ENFORCED AT RUNTIME · NOT ADVISORY
The definition is readable
The ICP is written down as rules and weights you can edit; the score on every row shows which rules fired.
Versioned
Changes to the ICP are versioned, so a score from March can be explained by the definition in force in March.
Scores write on approval
Writing scores to CRM fields is an approval-gated batch showing every record and value.
FAQ
Questions, answered.
How is the ICP built?
From evidence: your closed-won and best-retained accounts are enriched with firmographics, technology, hiring and funding data and analyzed for what they share, then written as rules and weights you can read and edit. Describe adjustments in plain language and the definition updates, versioned.
How do fit and timing combine?
Fit is scored against the ICP; timing comes from live signals: hiring, funding, technology change and intent. The list ranks on both, so a strong-fit account showing a signal this week sits above a strong-fit account that is quiet, and the reasons are written on the row.
Can scores live in our CRM?
Yes. Scores and the reasons behind them are written to the fields you name as an approval-gated batch, refreshed on a schedule, so routing rules and reports in the CRM run on the same definition the lists do. Each score records the ICP version it was computed under.
Can we edit the ICP by hand?
Yes. The definition is rules and weights in plain language, editable in Settings or by telling the mission "weight stack higher" or "exclude agencies". Every change creates a new version, and scores record the version they were computed under, so any past score can be explained.
How does scoring handle missing data?
Missing fields are filled through the waterfall before scoring where a source can answer, and rules that still lack an input abstain rather than guess, with the abstention noted in the reasons. The report shows coverage per rule, so a low score from missing data looks different from a poor fit.
Can it size a segment against the market?
Yes. A segment defined by the ICP is expanded to every matching company in licensed indexes and counted by geography and size band, with technology adoption charted beside it. The count becomes a list you can prospect from and an audience you can push, from the same definition.
What does scoring cost?
Plans start at $65 per month with credits included. Scoring, segmentation and score writes are free; the enrichment that fills missing inputs costs a few credits per row and only when a source answers. Every run states its estimate first against a ceiling you set.
⟨ RELATED READING ⟩
⟨ DEPLOY ASTROFABRIC ⟩
Decide who is worth the next dollar.
The ICP from your best customers this week; every account scored by Friday.