Tools Health score playground
What is your health score not telling you?
A weighted health score lets strength in one area cover for a warning in another. This is the model from my article, running on a made-up segment of 1,200 small-business accounts. Edit the accounts, tune the weights, and switch the hard overrides off to see what the average hides.
- Engagement
- 40 points for how recently and how often people log in
- Feature depth
- 40 points for how many of the six core modules are in use
- Support & sentiment
- 20 points for recent tickets and NPS
- Hard overrides
- No login for 30+ days, or an NPS of 6 or below, drops an account to Critical whatever its score
1,200 accounts, one switch
Each dot is an account in the segment. Accounts with a ring are the ones the overrides catch: the customer has gone quiet or told you they're unhappy, but the weighted score alone would call them fine.
Ten accounts up close
Open any account and change its numbers. The score, the tier and the reason update as you go. Westbrook Physio is the example from the article: busy, broad usage, and an NPS of 4.
Tune the model
Everything here feeds the segment and the accounts above. Weights are rescaled so they always add up to 100.
What each result sets in motion
A score is only worth building if it changes what someone does on Monday morning. These are this week's numbers for the segment above.
The playbook around it
The score is one piece. Here's how I'd stand up a scaled programme around it, roll customers onto a new platform without losing any, and report progress honestly.
-
Days 1–30
Stop the bleeding, build the base
- Build the health score as a live CRM report
- Turn on email open and click tracking
- Work the worst 20 to 30 Critical accounts by hand
- Agree one definition of "at risk" across CS, support and sales
-
Days 31–60
Systemise and automate
- Refresh scores automatically every night
- Ship three playbooks: inactivity nudge, detractor outreach, new-account activation
- Scope a pilot group for any platform change from the Healthy tier
-
Days 61–90
Prove it and scale
- Report the first read on how accounts are moving between tiers
- Send an automatic weekly alert for every new Critical account
- Lock the pilot group and start the rollout communications
Wave by profile
Start with Healthy single-product accounts, then Healthy multi-product, then At-Risk once the playbooks have shown they work.
Batch to your capacity
Move 50 to 100 accounts a fortnight, so support response times don't slip while everyone learns the new platform.
Let the data set the pace
Use the pilot group's email engagement and support volume to sharpen the message before each wider wave.
Hold some accounts back
Any account that is Critical, has an open support ticket, or is under three months old waits until it's stable. A change of platform is the last thing a struggling customer needs.
Report these as progress
- Score coverage
- The share of the segment carrying a live, automated score. The target is close to 100%.
- Net tier movement
- Accounts moving out of Critical and At-Risk, watched alongside the ones sliding down.
- Playbook reach
- At-Risk accounts actually touched by a playbook, against those flagged and left alone.
- Re-engagement speed
- The share of dormant accounts that log back in within 14 days of an inactivity sequence.
Flag these as risks, not wins
- Early renewals
- Renewals in the first 60 days reflect how customers already felt. They don't prove the new playbooks work yet.
- Missing NPS
- Accounts with no recent response are a gap in the data. Silence isn't health.
- Opens versus value
- An opened email is attention, not retention. Logins and module usage are what keep customers.
Questions about the model?
Built through LAB59, my company. Everything on this page is made up; the thinking isn't.