ArticleCustomer SuccessAugust 14, 2019
Our Top Indicators of Renewal Image

Our Top Indicators of Renewal

There is nothing more important to retaining customers than understanding why they renew. In 2015, we unveiled our Risk Management Framework, which helped us to establish a common view of customer risks. Since then, we have successfully used this framework to track customer health and address early warning signs.

But we weren’t totally sure that we had captured every critical type of risk, particularly as our business evolved. Our customer base had grown significantly since 2015. That meant we had lots of data we could analyze to understand the leading indicators of renewal.

The Process:

In partnership with our Data Science team, we studied the effects of 60+ attributes from a variety of data sources on our customers’ likelihood to renew.

We first identified customers who had come up for renewal. Therefore, we excluded new customers who had not yet had the chance to renew or churn. For each of these customers, we indicated whether they had “renewed” or “churned”. (As much as we’d like to say we don’t have churn, we do have it!)

Then, we gathered data for each of these customers across a variety of indicators. For our analysis, we aggregated the data by month, and we focused, for the most part, on analyzing the 3- to 6-month period before the customer’s renewal date. We used Random Forest, one of the most powerful machine learning algorithms, to determine the relative importance of 60+ attributes based on our data set.

Through this analysis, we discovered that some metrics were much more indicative than others, with a few surprises in both strong and weak indicators of renewal.

Strong Indicators of Renewal:

1. Breadth of Usage

According to our analysis, one of the leading indicators of renewal for Gainsight is breadth of feature usage. As a product, Gainsight has a variety of features, including:

  • Rules Engine - build Rules to trigger CTAs, update scorecard measures, move data, etc.
  • Customer 360 - get a 360-degree view of a customer in one place
  • Cockpit - manage your priorities from a centralized location
  • Reporting - create personalized reports and dynamic dashboards
  • CoPilot - launch automated outreaches based on customer attributes, behaviors, events, and more

Our findings suggest that, of the ten features that we examined, customers who actively used more features were more likely to renew. This makes sense for Gainsight, because one of our value propositions is to provide a comprehensive set of tools to make customer success management easy. Customers who leverage the suite of functionality comprehensively often see more value in using Gainsight.

Furthermore, certain features had a stronger impact on retention than others. While Customer 360, Cockpit, and Reporting are all heavily used features, CoPilot and Rules Engine were the features most associated with retention -- they are the “stickiest”. Knowing this, we have begun to promote breadth of usage, particularly in CoPilot and Rules Engine, and implement solutions to better track feature usage.

2. Daily Active User %

As many would expect, DAU % is a strong predictor of renewal. If a customer is not using the product, then chances are that they are not deriving value from the product and are less likely to renew.

We also looked at similar metrics such as Total Count of DAUs and Total Page Views, and as expected, these metrics are also strong predictors of renewal. We believe that DAU % is a slightly better indicator than Total DAU because:

  • DAU % factors in the total number of licenses, which captures an element of value that is missing in Total DAU
  • DAU % is a more standard metric across customers of different sizes

3. Support Tickets

Most of us would imagine that both too many and too few support tickets would be a bad thing:

  • Having too many tickets would suggest product issues, messy user interface, etc.
  • Having too few tickets would suggest lack of engagement

However, our analysis showed that only the latter was statistically significant. Customers who submitted more support tickets (during a 3-month time frame) were more likely to renew; the low number of support tickets was a sign of disengagement, not health. Meanwhile, many of our customers who submit the most support tickets are some of our healthiest customers, including one customer who has submitted ~300 support tickets throughout their time with Gainsight.

We believe that a large number of outstanding tickets may not be a predictor of renewal because of the strength of our Support Risk process, in which a CSM helps quarterback the entire situation with the Support team to ensure that the customer’s needs are met. So we’re going to continue to track situations in which there is either a large number or a small number of outstanding support tickets.


Now, let’s look at some of the metrics that were not very indicative of renewal in our customer base.

Weak Indicators of Renewal:

1. NPS

NPS is used throughout the industry and is deeply embedded into our product. However, our analysis indicates that NPS may not be so effective in predicting renewal. This is not to say that NPS is not an important metric. We have a couple hypotheses for why there is a weak relationship between high NPS and renewal in our customer base:

  • A low NPS usually triggers mitigation actions on the CSM’s part to “fix” the issue
  • Simply responding to NPS surveys is a sign of engagement from the customer

The latter is consistent with our observation that having NPS survey submissions at all had a bigger impact on retention that the scores themselves.

2. Time to Launch

Another metric that was not a strong indicator of renewal was time to launch. We know that time to launch is important to customers. In fact, we are well aware of the risks that can occur during the implementation stage. However, time to launch as a metric was not significant in predicting the likelihood to renew. One explanation for this may be the high degree of variance in time to launch - there are numerous variables (company size, data quality, etc.) that affect how long an implementation will last, and when taken as an average, may not say much about retention. This is an area where deeper segmentation may yield more insights for us.

Rest assured, we’re still obsessed about helping customers launch Gainsight in a speedy way :)

3. Attendance at Events

Every year, we host Pulse, our Customer Success conference that attracts over 4,000 customer success professionals. We also participate in Dreamforce, a conference run by Salesforce. While these events are undoubtedly a big part of our brand, we see that attendance at these events is not a strong predictor of renewal. This may be due to the nature of these specific events (perhaps these events are a better predictor of sales than retention), or it may suggest that events in general are not what ultimately drive retention.


Our findings ultimately suggest that product engagement (measured by breadth of usage, DAU %, and support tickets) is the biggest factor in customer retention.
This makes sense, because the more engaged a customer is with our product, the less likely they are to want to give it away. We see that other factors, such as sentiment (NPS), implementation (time to launch), and events (attendance) are not as predictive of renewal, or at least not through these metrics, but they continue to influence the way we interact with our customers.

In the future, we want to insert Value Attained as a metric in this analysis, since we all know that Adoption isn’t the same as Value, and helping customers get Value is the ultimate goal of our CSM team. We’ll be posting more about that in the future.

It’s worth noting that these indicators of renewal are bound to vary from company to company and product to product. Depending on your product, you may have different data points to measure different aspects of your product. It could be a useful exercise to evaluate your own data set to see what indicators are most relevant to you. If you are interested in getting help to do a data-driven retention analysis for your own company, please reach out to Sam Cummings (scummings@gainsight.com).

We would love to hear from you! What other indicators are we missing here, either for Gainsight or for your own product?

Picture of Jim Huang
Jim Huang Business Operations Associate | Gainsight Business Operations Associate | Gainsight
5 Comments

Leave a reply

Your email address will not be published. Required fields are marked *

  1. April Reynolds
    Apr 7th, 2017Reply

    One indicator that I have always found to be a strong indicator is the ability for a CSM to identify an Executive Sponsor. With strong usage and healthy ticket volume, reps still need access to someone with influence who can access budget to get the deal done.

    CSMs need to build relationships not only with the end users and management, but also with folks above the power line to understand true business objectives and show value.

  2. Casey
    Apr 10th, 2017Reply

    This is great! We’re actually working on figuring out the same thing with our customers and finding similar trends. And as April said, knowing you’re talking to the *right* person makes a world of difference.

  3. Andy Powell
    Jul 27th, 2017Reply

    Purchase history is an oversight and deserves mention. If the customer has recently said “Yes” with their wallet in any way (product, services) it is a good indicator they will say “Yes” to the renewal. E.g. If a customer expanded their license count 3 months prior, but haven’t fully onboarded, then DAU% would take a hit. Without looking at purchase history, the customer would show as less likely to renew.

  4. estefaniabartolomeo@gmail.com
    Oct 31st, 2018Reply

    Email open rate seems to be a good indicator of engagement too. If a customer opens the company e-mails, he´s more likely to be engaged and, therefore, renew.

  5. sam k
    Nov 25th, 2018Reply

    Amazing article with great insights. I would like to understand little more on how you used random forest..can you point me to any article or tutorial or sample that can educate me on using random forest?

Shares