The problem with health scores.
In the CS world, we are ALL about health scoring. We invest in solutions to help us make it easier, more at-a-glance, less biased, and more accurate. But there is a fundamental problem with health scoring: it is entirely reactive. We are boiling everything we know about our customers into a number, based on what has happened in the past. By the time we identify a customer as a churn risk, we are too late. They are already a churn risk!
What’s worse? That’s true even in the best case scenarios, when companies are using holistic data to guide their health scores and automation or strong processes to ensure they remain up to date. Many of us aren’t even close to that idyll, haphazardly updating health scores in a manual way, using our best guesses about customer health based on biased conversations and a data source or two.
How do we fight sparks before they become fires?
In recent years, CS teams have been starting to suspect that holistic data and health score automation are key to accurate and useful health scores. What we’re only starting to discover, however, is that we can do even better. We can use historical data to predict health, not just identify it.
But - I’ll warn you - it’s hard to do on your own. As I mentioned before, simply gathering customer data and automating updates is hard! Turning that information into predictive health scores is even harder. But if you’re up for the challenge, it’s worth it. You can turn your organization from one that fights fires into one that channels them into revenue growth!
What it looks like
So, what does it take to transform health scoring from reactive to proactive?
- Bring at least a year’s worth of historical customer data together - the more data the better, and bonus points if you can include sentiment.
- Use that data to determine which customers belong in like segments. Most of us rely on arbitrary metrics like ACV to segment customers, but your data might tell a different story.
- Compare historical data for each segment to churn and upsell to determine which metrics historically have had the most impact on each.
- Weight these metrics appropriately in the calculation of health scores going forward. If, for example, support ticket severity has been most highly correlated with churn, it should be factored more heavily into the health score than, say, meeting frequency.
- Most importantly - make sure to run these calculations and update the resultant health scores regularly.
It sounds too good to be true, right? Fortunately, technology is keeping up.
A handy shortcut: let AI handle the lift!
If you don’t have a team of data scientists devoted to the full time upkeep, unification, and analysis of your customer data, consider an early warning solution like involve.ai, that does this in real-time with AI. But also remember that something is better than nothing. If you can’t invest in an early warning system, many traditional CRMs are making strides toward health score automation; combining this with a quarterly, or even annual, manual effort to reconfigure segments or identify high priority metrics will still be a valuable effort at proactive health scoring.
----------
Mary Poppen is the Chief Strategy and Customer Officer at involve.ai, responsible for driving the Customer Intelligence category in the market. She is building a world-class delivery team focused on helping companies leverage the power of Artificial Intelligence (AI) to become customer-centric. Prior to involve.ai, Mary was Glint’s Chief Customer Officer at LinkedIn and Chief Customer Officer for SAP’s Global Cloud business before that. Mary holds a Master’s Degree in Industrial/Organizational Psychology and has over 20 years of customer success, business consulting, and executive leadership experience.
And if you’d like to learn more about Customer Intelligence, she literally wrote the book on the subject! Sign up to receive a complimentary copy of “Goodbye Churn. Hello Growth!” to learn more about the transformation of the Customer Success function and all things Customer Intelligence.