workspace / platform/scoring

How a 0-100 fit score gets computed.

Every account in your CRM gets one number that tells your reps whether to call this week, this quarter, or never. Here is how it is built.

Four input classes.

Filing signals
35%

Capex authorizations, capacity expansions, supplier-concentration risk, outsourcing language. Item 1A / 7 / 7A extractions.

Customs signals
35%

HTS-class match, volume trend, tariff exposure, sourcing-shift detection. 24-month rolling window.

Hiring signals
20%

Role-class clustering, geographic correlation with capex signals, 14-day window.

ICP match
10%

Industry, revenue band, geography, certifications you require. Defined during pilot setup.

A typical $40M manufacturer's pipeline.

1,800 accounts in CRM. ~50 score above 70. Top 25 become priority briefs.

90-100
12
P1
80-89
38
P1
70-79
71
P2
60-69
124
P2
50-59
196
P3
40-49
252
P3
30-39
304
noise
20-29
421
noise
0-19
382
noise

Old signals lose weight.

A capex signal in a 10-K filed eighteen months ago is no longer actionable. Each signal class decays with its own half-life:

filing signals
half-life · 12 months
customs signals
half-life · 6 months
hiring signals
half-life · 90 days
ICP match
no decay

What the score does not measure. The score is fit plus timing, not propensity to close. A P1 account is not guaranteed to buy from you. A P1 account is one your reps would regret missing.

We do not use intent signals from third-party trackers (no Bombora, no G2 surge data, no anonymous-visitor de-anonymization). Those signals are noisy in industrial verticals.

The model is calibrated quarterly against pilot outcomes. Customers see the calibration delta in their quarterly re-scoring report.