DocsIntelligence
User Guide

Intelligence

Customer health scoring, churn prediction, MRR trend analysis, and AI-generated insights.

Part 1

How Intelligence works

Intelligence runs nightly across all your Stripe customers. For each customer it computes three health bands — engagement health, recovery history, and customer value — and combines them into a 0-100 health score and a recommended action.

High risk
2
$495/mo at risk
Medium
107
$14,145/mo
Healthy
98
$11,579/mo

The health score is a UI-continuity number. The recommended action is what matters operationally.

Part 2

Health scores

The 0-100 score

The score is derived from three independent bands:

Engagement health. Based on whether the customer opened or clicked a recovery email in the last 30 days. Green: opened recently. Yellow: no recent engagement. Red: no engagement signal.

Recovery history. Based on prior failed payments and whether they recovered. Green: no failures or all recovered. Yellow: one unrecovered failure. Red: multiple unrecovered failures.

Customer value.Based on the customer's MRR as a percentile within your account. High: top 25%. Medium: middle 50%. Low: bottom 25%.

CustomerPlanHealthMRRAction
Sycamore CoStarter20$0Personal outreach
Bluewave StudiosScale28$495Personal outreach
Raj PatelGrowth43$99Watch closely
OakCoreStarter50$29Continue automated
Zane DiazStarter50$29Continue automated

Score thresholds and colors

200-33 red, high risk. 5034-66 amber, watch. 8267-100 green, healthy.

Part 3

Recommendations

Seven action types

Intelligence produces one of seven recommendations per customer:

  • Personal outreach. High value customer showing disengagement or failure history. Reach out directly before automated emails.
  • Continue automated. Sequence is running, no action needed from you.
  • Continue automated (priority). High value, automated is working, flag for monitoring.
  • Continue automated (deprioritize). Low value, automated sequence will be delayed 48 hours.
  • Accept loss dispute. Customer has an open dispute and low health score. Accepting may be better than fighting.
  • Expansion outreach. Healthy customer with growth signals. Consider an upsell.
  • Win-back priority. Recently churned customer with high prior value. Win-back email sends at 1 hour instead of 4 hours.

Customers needing action appear on the Recommendations page grouped by action type. Customers with continue_automated are excluded: they need nothing from you.

Part 4

Churn risk

The Churn page shows your MRR distributed across three health bands.

Real Teamflow data:

  • High risk: 2 customers, $495/mo
  • Medium: 107 customers, $14,145/mo
  • Healthy: 98 customers, $11,579/mo

The high-risk table shows your worst-scoring customers sorted by MRR contribution descending: highest-value at-risk customers first.

Part 5

The Intelligence dashboard shows four MRR metrics updated daily from your Stripe subscriptions:

  • Total MRR $26,218 (Teamflow demo)
  • Active customers 185
  • NRR 99.93% (near-perfect retention in the demo)

Plan breakdown: Starter ($2,471, 100 customers), Growth ($6,779, 65 customers), Scale ($16,969, 42 customers).

The MRR movement waterfall shows new, expansion, reactivation, contraction, and churn MRR week by week. Use it to spot trends before they show up in your headline MRR number.

Part 6

AI analyst

Every night the AI analyst reviews every at-risk customer and writes a 2-3 sentence plain-English assessment. It explains why the customer is flagged, what the risk is, and what the recommended action means in context.

The assessment appears on each customer's detail page under the health score. It is regenerated nightly. You always see the most current read.

Tip. The AI analyst skips customers who are green on all three bands and showing continue_automated. It focuses its analysis where action is actually possible.
Part 7

Settings

Weekly narrative email

Monday morning email summarizing your Intelligence data for the week. Enable in Settings.

Health alert

Email or Slack notification when a customer's health score drops below a threshold you set. Configure the threshold (1-99) in Settings.