Get the demo
New-gen lead scoring

Lead scoring, finally explainable.

A scoring engine built for RevOps and sales leaders who do not trust black-box numbers. Every score arrives with a reasoning trail. Models retrain on real won and lost outcomes, not a static rubric.

Four-signal model · Outcome-trained · Auditable scores
The four score signals

One score. Four sources.

A single number on the lead card, built from four independent signal layers. Each layer is weighted by its historical lift on closed-won outcomes in your own pipeline.

01 / Firmographic

Who they are

The ICP layer. Vertical, headcount band, funding stage, geography, tech stack. Static facts that anchor the rest of the score.

IndustryStageHeadcountGeo
02 / Behavioral

What they are doing

Job posting trails, exec moves, product launches, team restructures. Behaviors that signal a buying motion is underway in the account.

HiringExec movesRestructuresLaunches
03 / Intent

What they are reading

First-party visits, third-party intent feeds, review-site activity, content engagement. The "they are looking now" signal layer.

Pricing visitsReviewsTopic surgesContent
04 / Network

Who can vouch

Warm-intro graph against your team, your investors, your customers. Tells the SDR whether to cold reach or route through a relationship.

MutualsPast customersInvestorsAlumni
Every score, a reasoning trail

If the score is 0.92, we tell you why.

Every score breaks down into the four signal contributions, with a one-line reason and a confidence band. No black box, no vibes, no "trust the algorithm." Hover and read.

REASONING / 01

Each signal is line-itemed

The score is a sum of four contributions. Each one is plotted, named, and explained in one short line. The math is open.

REASONING / 02

Confidence is shown, not hidden

Confidence travels with the score. A 0.92 at 60% confidence reads differently than a 0.92 at 92% confidence. Both are shown.

REASONING / 03

RevOps can audit any lead

Click any score, see the reasoning trace, see the underlying signal events. The reasoning is the same one the SDR sees.

REASONING / 04

Sales reps can disagree

A rep can flag a score as wrong with a one-line reason. That flag flows into the next retraining cycle.

Outcome-trained retraining

Scores get sharper as deals close.

Every won and lost deal flows back into the model. The weights, the signal coefficients, and the confidence bands all reshape against your actual pipeline. Static rubrics are out.

01

Outcomes capture

Closed-won, closed-lost, and stalled deals get tagged at the account level and tied back to the original signal trail and score.

02

Signal coefficients

The model retrains on the new outcome set. Signals that predicted close get a heavier weight. Signals that misfired get downweighted.

03

Score recalibration

Every open lead is rescored against the fresh model. Reasoning trails update to reflect the new weights. Nothing silent.

04

RevOps review

Before the new model ships, RevOps sees the delta. Which leads went up, which went down, and which signals shifted. Approve and ship.

Module breakdown

Built from three modules.

The platform ships as a connected set, not a bundle. Each module does one job well. Bring the ones you need, leave the rest for later.

Module 01

Score Engine

The core scoring service. Four-signal composite with confidence bands.

  • Firmographic, behavioral, intent, network
  • Composite score with per-signal contribution
  • Confidence band on every score
  • API, webhook, and CRM sync
Best for: RevOps standing up scoring from scratch.
Module 03

Retraining Loop

The outcomes-to-weights retraining pipeline.

  • Won, lost, and stalled outcome capture
  • Signal-weight recalibration on schedule
  • Pre-ship delta review for RevOps
  • Rollback to any prior model version
Best for: Teams whose ICP and motion change quarter to quarter.
Sample wins

Two RevOps teams. Cleaner pipelines.

Composite numbers from anonymized engagements. The reasoning trails were the unlock, not the score itself.

CODENAME ORCHID
Vertical SaaS, Series C

Reps stopped fighting the score.

Reps used to ignore the old static score. After the reasoning layer shipped, every account had a line-itemed trail. Pipeline meetings shifted from "is this score right" to "which signal do we lean into."

91%
Scoring accuracy on closed-won cohort
+38%
Lift from qualified to closed-won
-46%
False-positive high scores
CODENAME LANTERN
FinTech infra, mid-market

The retraining loop fixed the rubric drift.

Their static rubric was two quarters old and their ICP had shifted to a new vertical. The retraining loop reweighted firmographic signals against fresh closed-won outcomes. The top of the funnel re-sorted itself in one cycle.

94%
Scoring accuracy after first retrain
+52%
Lift from qualified to closed-won
-61%
False-positive high scores
See it on your own pipeline

Bring a lead list. We will score it.

Send a sample account list and a quick read on your closed-won pattern. We will score it with the four-signal model and walk you through the reasoning trail on each lead.