Score
An XGBoost churn model turns usage, support, renewal and champion signals into a live 0–100 health score and churn-risk % for every account.
Live account monitoring
Know which accounts are leaving — before they do
Churn Radar scores every B2B account 0–100, tells you exactly why it's slipping, and hands your CSM the save-play to run — with a confidence estimate attached.
- 238 accounts monitored
- Explainable, not black-box
- One CSM workflow
238 accounts
112%
+14 pts
NRR · 12 mo98% → 112%
Healthy
At-risk
Critical
0
Accounts monitored
0%
Net revenue retention
0
At-risk accounts
flagged now
0
Saves this quarter
retained
Why accounts churn
Churn is never a mystery. It's a weighted sum of signals.
Churn Radar attributes every at-risk score to the behaviour behind it — so the number always comes with a reason your team can act on.
Usage decline
High
Falling logins, seats and core-action volume — the earliest tell that value is fading.
Support strain
Medium
Ticket spikes, unresolved escalations and slipping sentiment across recent conversations.
Renewal proximity
High
Contract windows closing without expansion signals or an engaged buying committee.
Champion departure
Critical
Your internal advocate goes quiet or leaves — the single sharpest predictor of a lost account.
One account, fully attributed
Every score decomposes into the drivers that produced it — with weights, not vibes. Here's how one at-risk account breaks down.
42
/ 100
- Champion departure 36%
- Usage decline 31%
- Renewal proximity 22%
- Support strain 11%
From signal to save-play
Score it. Explain it. Prescribe the save.
Churn Radar closes the loop from raw behaviour to the next action a CSM should take — no spreadsheet archaeology in between.
Explain
Each score is decomposed into weighted drivers, so the CSM sees why an account is at risk — not just a number to distrust.
Prescribe
A sequenced save-play lands with owners, steps and a confidence estimate — the exact play to run to keep the account.
Inside the product
The command centre and the account file, side by side.
BuildspaceLabs built the MVP front end: a portfolio-triage Account Health board and a per-account detail view for the CSM.
Account Health board
Triage the whole portfolio in one glance.
Live NRR, at-risk and saves KPIs sit above a colour-coded, sortable table — so the accounts that need a human are always on top.
- 12-month net-revenue-retention trend rising to 112%
- At-risk accounts colour-coded by severity
- Top risk drivers surfaced across the book
Account detail
Everything a CSM needs on one account file.
A health ring, weighted why-at-risk drivers, usage sparklines and an activity timeline — capped by an AI save-play with its confidence score.
- 42/100 health ring with churn-risk context
- Weighted risk drivers and usage sparklines
- An 84%-confidence AI save-play, sequenced
What's in the box
Built for the customer-success workflow, end to end.
Health scoring, risk drivers and save-plays consolidated into one CSM surface — nothing left in a spreadsheet.
Account Health board
Live NRR, at-risk and saves KPIs above a triage-ready, colour-coded portfolio table.
0–100 health score
A predictive score and churn-risk % per account, refreshed as behaviour changes.
AI save-plays
Sequenced steps with owners and a confidence score — the recommended play to keep the account.
Risk-driver attribution
Every score decomposed across usage, support, renewal and champion-departure signals.
12-month NRR trend
Net revenue retention charted over a rolling year, from 98% up to 112%.
Account detail file
Usage sparklines, why-at-risk driver weights and an activity timeline on one screen.
Explainable, not black-box
A number your CSM will actually trust.
A score nobody understands gets ignored. Churn Radar shows the weighted drivers behind every health score — so the recommendation carries its own justification.
- Every score decomposes into named, weighted drivers.
- Save-plays ship with a confidence estimate, not false certainty.
- Real-time NRR and MRR exposure across all 238 accounts.
Health score · account #A-118
42
Champion departure
36%
Usage decline
31%
Renewal proximity
22%
Support strain
11%
Under the hood
Modern data stack, production-grade front end.
Delivered as a production-quality MVP in an eight-week engagement.
Next.js
React
TypeScript
Tailwind CSS
PostgreSQL
Python · XGBoost
dbt + BigQuery
See churn coming — for your book of business.
We build AI-native products like Churn Radar. Tell us about your customer-success motion and we'll show you what an explainable health system looks like on your data.