The Score is Changing in Customer Success

Over the past seven years, I have spent a lot of time learning about the world of customer success management (CSM), a new functional area focused on driving customer retention and revenue growth. Over the past decade, customer success has emerged as a “must-have” function for software-as-a-service (SaaS) companies and for many other organizations with subscription-based business models. Additionally, the customer success profession has experienced rapid growth in terms of both job opportunities and compensation. 

At the University of San Francisco School of Management, this has been on our radar for many years, largely because of our proximity to Salesforce.com and so many other technology companies. We have developed new courses focused on customer success, built an academic track in CSM for our MBA students, ran dozens of student consulting projects with companies large and small, worked with industry partners to launch continuing education classes, and conducted some of the first academic research associated with CSM. For me, it has been energizing to start this initiative from scratch (we were the first university in the world to have any kind of program in customer success and SaaS), engage with a new community of technology industry professionals, and immerse myself in new ideas and concepts. Indeed, I felt an unreasonable sense of pride upon learning late last year that I had been voted one of the Top 100 Strategists in Customer Success.

As a young profession, customer success is constantly evolving. In the aftermath of a relatively challenging year for much of the technology industry (including significant layoffs among customer success professionals), one emerging trend for SaaS companies is to make the customer success function explicitly responsible for exi$ting cu$tomer$ and to be accountable for the $ame type$ of metric$ as a traditional $ale$ organization, with customer success staffing levels linked directly to goal$ and target$. In theory, having clear metrics should make resource allocation and investment decisions clearer for customer success organizations. In practice, proactively managing to these metrics is quite a bit more challenging than one would think at first glance.

Lack of Metrics and Data

One problem is that retention and growth revenues are lagging metrics, and historical data is often not available to correlate past customer-related activities and choices with future financial outcomes. In addition, customer success leaders are challenged to determine where and how to combine technological capabilities and human resources, a particularly complex decision space for companies with thousands of customers and multiple products. And in the wake of widespread budget cuts in 2023, most SaaS companies are operating with tight budgets and significant financial pressures.

In response to these challenges, many customer success groups have developed “Customer Health Scores,” which serve as a proxy for the ri$k (or up$ide) associated with each customer. These health scores are also used to provide guidance about where and when to engage in particular types of customer success activities. For example, a customer with a low health score might garner additional attention from a member of the customer success team before the contract renewal date in an attempt to reduce a perceived high level of risk.

In theory, this too is a great idea. In practice, this is trickier than it appears. Many health scores are either totally or partially dependent on subjective judgments from human beings with a decidedly imperfect understanding of future customer outcomes and limited time and energy for quantifying that understanding. Also, health scores are often too complex to interpret – and with input data coming from different systems and people, it is hard to keep the inputs (and therefore the health scores) updated regularly. Finally, most companies lack the data infrastructure to carefully examine the relationship between today’s health scores and tomorrow’s cu$tomer outcome$.

Brent Grimes has had first-hand experience with these types of challenges. Upon taking over customer success in 2014 at MuleSoft (a software company whose integration platform helps its customer connect SaaS and enterprise applications), Grimes quickly realized that “as we started to scale, we were never going to have enough resources to try to cover every customer and do enough with every customer.” Tasked with maintaining – and growing – revenues from MuleSoft’s customer base, Grimes worked with members of the data science and operations teams in an attempt to develop a data-driven health scoring model that would allow his team to understand and predict future renewal risks and expansion potential several months before the end of a customer’s contract. Crucially, these health scores were automatically updated, allowing MuleSoft to respond dynamically as additional customer data became available.

In fact, once the organization came to trust these predictions, Grimes eventually shifted the company’s customer success team to what he calls the “outlier model” of service delivery, focusing more resources on customers with higher risk for churning and/or upside revenue potential. Grimes explained that “we were able to categorize accounts based on our health scores and really drive the kind of operational investment based on these categories. And what we discovered was that this approach was less expensive in terms of resource outlay – and actually performed better than our previous more traditional approach.”

The rest is history. MuleSoft enjoyed a successful IPO in early 2017 before being acquired a year later by Salesforce for $6.5 billion, in no small part because of its 117% net revenue retention rate. In addition, Grimes credits his experience at MuleSoft for motivating him to start a new software company called Reef, focused on helping its clients use data and build scoring models to effectively manage and grow net revenue: “What I realized was that if we can take something that was very manual, very one-off, very labor-intensive at MuleSoft and productize that, a lot of companies could potentially benefit from it.”

Founded in 2021, Reef is still early in its journey. But I am confident that the future of customer success will surely include data-driven customer health scores that truly correlate with lagged metric$, along with insightful, actionable guidance driven by disparate sources of customer data (with many of these recommendations eventually being automated and/or powered by large language models, or LLMs). It is also clear that the combination of emerging technologies and increased financial attention means that “doing more with less” is likely to be a permanent state of affairs for most customer success teams. In a world of limited resources, using data- and model-driven metrics will be a key part of optimizing operational and financial results. As a data science researcher and analytics professor immersed in the world of customer success at the dawn of the LLM era, I will be keeping a close eye on this space.