The arithmetic of customer success at scale is unforgiving. You have a portfolio of long-tail customers – small initial deals, genuine growth potential, but not enough annual recurring revenue to justify a dedicated CSM spending half their week on each one. The question every CS leader eventually confronts is whether to scale by hiring more people or by getting smarter about how the existing team spends its time. David Becerra, who built a low-touch customer success organization at SAP from scratch, walked Thibaut de Lataillade through how his team chose the second option and what happened when they finally got the data they had been operating without.
Operating Blindly at Enterprise Scale
Before Mixpanel, Becerra’s team at SAP knew two things about their customers: they had bought the software, and a certain number of seats were active. That was it. No feature-level usage data. No visibility into which parts of the product were being adopted and which were gathering digital dust. No way to distinguish between a customer who was deeply embedded in the platform and a customer who had logged in once, poked around, and quietly moved on.
The team was active. They were running outreach campaigns, conducting check-ins, following their playbooks. But the activity was disconnected from any meaningful signal about what was actually happening inside the customer’s instance of the product. They were making decisions based on what they could see – seat counts, renewal dates, support tickets – and missing the entire layer of information that would tell them whether a customer was healthy, at risk, or already mentally checked out.
“We were doing a lot of things. We were having high activity, but it’s really hard to tie that and correlate that to an outcome of a renewal or of churn.”
This is a problem that sounds obvious in retrospect but is remarkably common in practice, particularly in large organizations where the CS team inherits a tech stack that was built for sales, not for post-sale engagement. Salesforce tells you who bought what and when. It does not, by itself, tell you whether the customer who bought an analytics platform six months ago is building dashboards every week or has not opened the application since February.
What Changed When Usage Data Arrived
The implementation of Mixpanel gave the team feature-level visibility for the first time. They could see which specific product features each customer was using. They could see adoption patterns, activation rates, and usage frequency at a granularity that seat counts and login data had never provided. The impact was not subtle.
For the digital tier – customers with the lowest ARR, managed entirely through automated touchpoints – the usage data enabled something that had been impossible before: intelligent segmentation. Instead of sending the same generic email to every customer in the cohort, the team could identify clusters of customers who were not using a specific feature that correlated with higher retention. Those customers got targeted outreach: a webinar invitation about that feature, an email explaining what it could do for them, a nudge at exactly the moment it mattered.
For the low-touch tier – Becerra’s primary domain, where CSMs were managing 60 to 80 accounts each – the data changed the nature of the conversation. A CSM walking into a customer review armed with “I see you have not activated feature X yet, and our most successful customers in your industry tend to use it for Y” is having a fundamentally different conversation than a CSM asking “So, how’s everything going?” The first conversation demonstrates that you are paying attention. The second demonstrates that you are not.
“It gave CSMs that insight to have a human interaction, but contextually around, ‘Gee, I see you’re not using this product feature. Did you know this or that?’”
The high-touch team benefited too, though their use was more surgical. They had fewer accounts but deeper relationships, and the usage data gave them ammunition for the strategic conversations they were already having. When a renewal was approaching and the champion was asking for internal justification, the CSM could provide specific adoption metrics rather than vague assertions about “value delivered.”
The Three-Tier Model
What emerges from Becerra’s description is a three-tier engagement model that uses data as the connective tissue between all three levels:
The digital tier – lowest ARR, highest volume – runs almost entirely on automation. Data triggers the outreach. No human touches the account unless a signal indicates something worth escalating. The economics only work if the automation is intelligent, and intelligent automation requires data that goes deeper than “are they still paying us.”
The low-touch tier – mid-range ARR, 60 to 80 accounts per CSM – blends automated signals with human judgment. The CSM cannot spend a full day preparing for each account review. They need the system to surface the three or four things that matter most, so their limited human time is spent on the conversations that will actually move the needle. Mixpanel did not replace the CSMs. It told them where to look.
The high-touch tier – largest accounts, dedicated CSMs – still relied on deep personal relationships, but the data elevated the quality of those relationships. It is one thing to tell a customer you are their strategic partner. It is another to walk into a meeting with a clear picture of which features their team is using, which they are ignoring, and what that pattern means for their business outcomes.
Measuring the Impact
The number Becerra returns to is a 10% reduction in churn over a few years following the Mixpanel implementation. That number is worth pausing on, because it represents a transition from zero product-level visibility to basic usage analytics. Not a sophisticated ML-powered health scoring model. Not a predictive churn algorithm. Just the ability to see which features customers were actually using.
The implication is stark: if you are running a CS organization without feature-level usage data, you are leaving a significant portion of your retention performance on the table. Not because you lack talent or effort, but because you lack the information to direct that talent and effort at the accounts and the problems that matter most.
Becerra is careful to note that data is not a silver bullet. You can have the best dashboards in the world and still lose customers if the product is broken, the support is slow, or the value proposition has drifted from what the customer actually needs. But without the data, you are making retention decisions based on incomplete information and hoping that effort alone will compensate for the gaps. In a scale organization managing hundreds of accounts, hope is not a scalable strategy.
For the full interview breakdown, see our complete Expert Insight with David Becerra.
Tools Mentioned in the Interview
The following tools and platforms were referenced during this conversation.
Related Content
Customer Success From Startup to Surf Camp: Expert Insights with David Becerra
David Becerra has built customer success organizations from startup to SAP enterprise to a surf retreat in Indonesia. He breaks down what actually drives retention, why your quietest customers are your biggest risk, and what hospitality teaches SaaS about onboarding.
Why Your Quietest Customers Are Your Biggest Churn Risk
David Becerra explains why angry customers are easier to save than silent ones, and how his team built a support culture around converting complaints into advocacy.
Time to Value: Why the First 30 Days Make or Break a SaaS Renewal
David Becerra explains why the sales-to-CS handoff is where most SaaS companies lose customers, and how a mandatory activation package cut early churn by 15% at his startup.
