Anti-Fragile GTM

Use case library

Retrain account scoring on every closed deal

A scoring model that sharpens with every win and loss, not a static firmographic filter.

The problem

Your account scoring model was built six months ago from 40 closed-won deals. Since then, 60 more deals closed, your ICP shifted, and new signal types emerged. The model is stale but reps still trust the numbers.

The system

  1. Extract features. Firmographics, technographics, signal history, and engagement from every closed opportunity.
  2. Retrain weekly. Model updates with Laplace smoothing for small samples, full retrain as data grows.
  3. Write scores to CRM. Updated rankings visible to reps with score change indicators.
  4. Measure lift. Track conversion rate by score decile to validate model performance.

Field notes

Retraining is the anti-fragile loop. Every deal makes the model sharper. Static scoring is a snapshot that rots.