EN | TR
Sign in Try it free
ML Prediction Engine

Know who's leaving.
Know what they're worth.

ML Prediction Engine scores every CRM contact with RFM-based churn risk and predicted lifetime value — then writes the results straight onto the deal records that produced them.

Try ML Prediction Engine View Pricing
RFM Scoring Predicted LTV CRM Auto-Sync Run History One Shared Engine
RFM Scoring

Churn risk, computed
from real pipeline behavior

Every contact is scored on Recency, Frequency, and Monetary signal pulled straight from your CRM pipeline — no manual tagging, no separate data export.

  • ✓
    Recency derived from pipeline stage and last-contact date — "closed-lost" and "churned" stages are penalized automatically
  • ✓
    Frequency weighted by pipeline stage strength — active and closed-won deals count more than a cold lead
  • ✓
    Monetary value normalized against the full pipeline range, not a fixed threshold
  • ✓
    Three configurable weight sliders (Recency / Frequency / Monetary) auto-normalize to 100%
Open AI Predictive Models
totalapp.app / ai-predictive-models
Default RFM Weights
Recency
34%
Frequency
33%
Monetary
33%
Cluster Churn Risk
Champions
8%
Loyal Core
20%
Potential Growth
45%
At Risk
75%
Hibernating
88%
Predicted LTV

Scores that land
directly on the deal

Every recalculation writes churn-risk score, predicted LTV, and a scored-at timestamp straight onto the CRM deal record that produced them — no separate export to reconcile.

  • ✓
    churnRiskScore, predictedLtv, and scoredAt written onto the matching deal
  • ✓
    Read-modify-write preserves every other field on the deal untouched
  • ✓
    Predicted LTV is the deal's own monetary value at time of scoring — no external LTV model needed
  • ✓
    Deal-level scores immediately visible back in the CRM pipeline board
Open CRM Pipeline
totalapp.app / crm-pipeline
Deal — Acme Industries
Enterprise Renewal closed-won
churnRiskScore 8
predictedLtv $84,000
scoredAt 2026-07-29 09:14
All other deal fields (owner, stage, notes, tags) preserved unchanged by the read-modify-write.
CRM Auto-Sync

Preview before
it touches live data

A single tenant-wide toggle decides whether scoring results get written back onto CRM deals. Turn it off to compute and review segments risk-free before committing to production data.

  • ✓
    Sync on — every recalculation mutates the matching CRM deal records
  • ✓
    Sync off — segments still compute and save to run history, but no CRM deal is touched
  • ✓
    Safe way to test a new RFM weight split before it reaches live pipeline data
  • ✓
    Current sync state shown live in the KPI strip on AI Predictive Models
Configure Auto-Sync
totalapp.app / ai-predictive-models
CRM Auto-Sync
Write scoring results back onto CRM deals
Last 3 Runs
all · 214 deals SYNCED
all · 214 deals DRY-RUN
crm-main · 189 deals SYNCED
Run History

A full audit trail
of every recalculation

Every scoring run — manual or automated — is logged with timestamp, scope, deals scored, CRM-mutated status, and total revenue covered, most recent first.

  • ✓
    Up to the last 20 runs retained per tenant
  • ✓
    CRM Mutated badge shows whether that specific run wrote to live deals
  • ✓
    Revenue pool per run helps spot when a scope filter changed coverage
  • ✓
    Empty state shown clearly if the tenant has never triggered a run
View Run History
totalapp.app / ai-predictive-models
TimeScopeDealsSync
Jul 29, 09:14all214YES
Jul 28, 17:02all214NO
Jul 26, 08:47crm-main189YES
Jul 22, 14:30all201YES
Revenue pool: $1.8M across last 4 runs
One Shared Engine

Two screens.
One scoring engine underneath.

Customer Segmentation (Marketing) and AI Predictive Models (Orchestration) both call the exact same ML Prediction Engine — same RFM math, same clustering, same churn-risk and LTV output.

  • ✓
    Customer Segmentation — marketer-facing, ad-hoc weights for a specific campaign, no engine-wide config touched
  • ✓
    AI Predictive Models — admin/ops console for tenant-wide defaults, CRM sync, and run auditing
  • ✓
    Default weights set on AI Predictive Models become the starting point marketers see in Customer Segmentation
  • ✓
    One engine, one run history, no drift between the two views
Open Customer Segmentation
totalapp.app / marketing-audience-ml
Customer Segmentation
Marketing · ad-hoc weights
AI Predictive Models
Orchestration · tenant defaults
ML Prediction Engine
Same RFM scoring · same clusters · same run history

Stop guessing who's
about to churn.

RFM-based churn risk and predicted LTV, written straight onto the CRM deals that need them.

Try ML Prediction Engine Free Read the Docs