We are supposed to track customer health scores, but we tend to throw together a few numbers and just carry on with them for months. Here are four quick ideas to refresh our thinking.
Tip #1: Use different formulas for different customer segments
Such a simple, but powerful idea (shared by Abby Hammer from ChurnZero). If you segment customers, and most vendors should, health scores should also use segmented definitions. For instance, the health score of enterprise customers may rely more on the judgment of the customer success manager (see tip #3) than on mechanistic factors such as product usage.
One prime idea for segmenting is treating new customers differently from established customers. Health score formulas should reflect that.
Tip #2: Automate whenever possible
Carefully pondered and harvested qualitative factors are wonderful — and also expensive, and quickly dated. Instead, look for meaningful, relevant, automated factors such as the number of users, usage of specific features, and of course support history, including bugs and feature requests.
Tip #3: Carefully define subjective factors
I’m a great fan of qualitative factors, since it’s awfully hard to automate the measurement of whether your solution brings real value to customers. But don’t just give up and rely on individual CSMs’ assessments of qualitative items. Some will be overly optimistic, other pessimistic, and perhaps predictably so if they are responding to badly-chosen objectives.
Create realistic descriptions: what does it mean to have a rating of 5 on relationship quality? How it is different from 4?
Tip #4: Use actual referrals instead of NPS
I know everyone just loves NPS — but likelihood to recommend is no match for actual referrals. Sure, referrals are hard to track, but they are not impossible to track. Give it a try!
What techniques are you using for health scores? Please share in the comments.