Get ready for a thrilling ride as we unveil the secrets behind our customer tracking system, named the “Customer Health Score,” and how Snowflake, our trusty Data Warehouse, played a crucial role in revolutionizing our approach.
Imagine this: it’s late 2021, and our company faces a challenge — customers are leaving, and we’re puzzled about why.Determined to change things, we set out to find struggling customers before they think about leaving.
Our solution? A system capable of ranking our customers and providing us with a broader understanding of our customer base.
Welcome to the Customer HealthScore project, where we delved deep into the customer lifecycle and behavior, creating a formula that ranked customers on a scale of 0 to 100 — the higher the score, the healthier the customer.
Let’s break it down further into its three captivating sections:
1. Account Info — 40 Points
We set out to identify key indicators of customer health, focusing solely on metrics related to their account rather than their product usage.
- HyperCare — 5 Points: During the three months after a customer’s go-live and three months before every renewal, we entered a hyper-care phase. This meant we needed to provide extra attention and stay in close contact with these customers. Consequently, their HealthScore took a slight dip.
- Open Bugs — 5 Points: While bugs and support tickets can indicate active engagement between customers and our product, we discovered that customers with a high number of open bugs tended to be more frustrated. Thus, their health score suffered accordingly.
- Billing Issues — 5 Points: A customer who truly loves our product won’t have any qualms about paying for it. If our finance department flagged a customer for billing issues, we adjusted their health score downward.
- CSM Sentiment — 25 Points: Automation is nifty and accurate, but nothing can replace the human touch. After every quarter, following the QBR (Quarterly Business Review), our CSMs provided input on their assessment of the customer’s sentiment, choosing between good, stable, or bad. We assigned health score points based on this invaluable feedback. As it turned out, this metric proved to be the most important, earning a substantial 25 points out of the overall 100.
2. Product Usage — 25 Points:
This section focused on aggregating and analyzing metrics related to customers’ usage of our product, offering a glimpse into their behavior.
- Expected Usage — 15 Points: Our product operates on a consumption-based model, with usage commitments at each renewal. By comparing the committed amount to the actual utilization rate by the end of the contract period, we could gauge if customers were truly utilizing the system as intended. Points were awarded based on the utilization rate.
- Expected Usage Trend — 10 Points: Building on the previous metric, we also tracked the trend of utilization. Did the customer’s usage rate increase or decline over time? This information helped us understand whether customers were ramping up their usage or losing interest.
3. Stickiness — 35 Points:
This section measured the “stickiness” or advocacy level of our customers by evaluating the following key points:
- Unique Users — 10 Points: We monitored the number of unique agents logging into the system each day and compared it to the average of the previous 14 days. If the numbers matched or exceeded the average, bonus points were awarded.
- Unique Teams — 10 Points: Recognizing that our product played a pivotal role in our customers’ day-to-day work, we checked if entire teams were logging into the system daily. If a team suddenly went MIA, it signaled that something was amiss. Again, we compared the numbers to the previous 14-day average.
- Features — 15 Points: Each core feature of our product received a score, and we tracked daily feature usage. The more features a customer utilized, the stickier they were deemed, resulting in a higher score. We also compared feature usage to the previous 14 days to identify any sudden drops.
By implementing this system, we gained invaluable visibility into each of our customers’ health.We opted to integrate it into our Salesforce CRM as an LWC widget in the Account Object, providing a clear visualization within the system, where our CSM’s live.Additionally, we crafted an aggregation report using our BI Platform (Sisense) to divide our customers into industries, sizes, and regions, helping us identify accounts at risk of churn.
Thanks to this groundbreaking system, our internal conversations about customer issues reached new heights, allowing us to retain customers, enhance their loyalty to our platform, and ultimately boost our Net Dollar Retention (NDR).
How did Snowflake come into the picture? Well, it served as our trusted Data Warehouse, efficiently collecting and modeling the data needed to power our CRM.
Here’s a breakdown of how we utilized Snowflake:
- Product Details: We created aggregational views from our product’s metric tables, which were already housed in Snowflake.
- Bug Tickets: To seamlessly transfer our ticket data from Zendesk to Snowflake, we employed Rivery, an easy-to-use data integration tool.
- CSM Sentiment: After each QBR, we implemented a survey sent to our CSMs to gather details about the meeting. The results were then sent to Snowflake (using Rivery) for analysis, and we updated the CSM Sentiment accordingly.
Remember, Data is power, and by wielding it smartly, we were able to unlock a new level of customer success.
