Picture a Monday morning. A CSM has a renewal call in 40 minutes with a $180K account. They open the CRM. Last activity note: six weeks ago. Health score: yellow, updated last Thursday. They open the product analytics dashboard. Login frequency looks normal. They scan the support ticket history. Nothing critical. They close four tabs and open the email thread. The last exchange was about an implementation issue from two months ago, marked resolved.
They walk into the call with a rough sense of the account. Not a complete picture. A rough sense.
The call goes sideways. The customer has been quietly evaluating a competitor for eight weeks. There’s been internal turnover the CSM didn’t know about. The new VP had concerns about ROI that were never formally escalated but had been circulating internally for months. The account churns thirty days later. The post-mortem lists “lack of executive sponsorship.” But the real answer is simpler: the information that would have changed the outcome existed. It just wasn’t connected.
The CSM Talent Problem Nobody Names Correctly
Kristi Faltorusso, former Chief Customer Officer at ClientSuccess, made a point on the Across the Funnel Podcast that cuts right to it:
“We hire the brightest people. And then we don’t empower them to do the right work. We have them filling in CRMs and CSPs and sending follow-up emails. That is not a great use of our people.”
The CS industry has spent years hiring smart, relationship-oriented professionals and then systematically burying them in data entry. Health score updates. Manual activity logs. Meeting notes typed into a CRM field that nobody reads until something goes wrong. The assumption is that having the data in the system is the goal. It isn’t. Having the data surfaced at the right moment, in the right form, is the goal.
Faltorusso’s model for fixing this starts by designing digital first, then adding humans where they genuinely add value. Not digital as a cost-cutting measure, but digital as a precision tool that handles the mechanical work so the CSM can focus on the judgment call, the relationship conversation, the moment that actually moves the account forward.
The problem is that “digital first” only works if the digital layer is actually intelligent. And most of the time, it isn’t. It’s a stack of disconnected tools, each capturing a slice of the customer relationship, none of them talking to each other in a way that produces a coherent picture of what’s happening.
What Gets Lost in the Gap Between Tools
The data exists. That’s the frustrating part.
Product usage is in one system. Support tickets are in another. Billing events are somewhere else. Email history is in the inbox. Call notes, if anyone took them, are in the CRM. And then there’s everything that lives in the meeting itself: the tone of the conversation, the questions that signaled concern, the stakeholder who went quiet halfway through, the comment about budget that didn’t make it into the summary.
Most CS teams can access most of this, with enough time and enough tabs. The problem isn’t access. It’s synthesis. Nobody has time to pull it all together before every call, for every account, every week. So the CSM makes a judgment call with partial information. Sometimes they’re right. Sometimes they’re not. When they’re not, it looks like a missed signal. It’s actually a structural problem.
In a conversation on the Across the Funnel Podcast, Ramsey Pryor, CEO of Rumi.ai, framed the scale of what gets thrown away:
“Most of that data has been completely discarded, and humans are really poor at remembering things. AI is amazing at remembering everything.”
He was talking specifically about meeting data, but the observation extends to everything that passes between a vendor and a customer across a relationship lifecycle. The signals are generated constantly. Most of them disappear.
Rumi.ai was built around the idea of perfect recall: not just recording meetings, but building a centralized repository that captures institutional knowledge across every conversation, internal and external, so that the context from twelve months of customer interactions is queryable before the next one begins. The CS team prepares for a renewal by asking the system what the customer’s main concerns were, what was promised, what changed. The answer takes seconds. Without it, that preparation either doesn’t happen or takes most of an hour.
The Signals That Never Make It Into Any System
The internal signals — product usage, support tickets, email engagement — are the ones CS teams are at least trying to capture, even if imperfectly. The signals that most teams aren’t tracking at all tend to be external.
A key contact at the account is promoted. The company announces a restructuring. A press release mentions they’re pivoting their product focus in a direction that changes the use case they bought your software for. A LinkedIn update from the champion signals they’re job hunting. The company just had a down round and needs to cut SaaS spend by 20%.
None of this shows up in a health score built from product telemetry and CRM data. All of it is relevant to whether the account renews. And the reason most teams don’t track it isn’t that they don’t know it matters. It’s that monitoring it manually, across 80 accounts, isn’t possible. So the CSM finds out when the customer brings it up in the renewal call, or doesn’t, and churns quietly instead.
The customer memory graph solves this by expanding the signal set beyond what’s captured inside your own tools. When external signals, press releases, hiring data, executive movement, company news, are part of the account picture, the CSM walking into that renewal isn’t piecing together six data sources in their head. They’re working from a single, continuously updated view of what’s happening with the customer and what’s changed since they last spoke.
What Intelligence Looks Like When the Picture Is Complete
The word “proactive” gets used a lot in CS. Most of the time, what people mean is “reacting a little earlier than usual.” True proactive CS requires knowing something is changing before the customer tells you, before the health score updates, before the support ticket spikes. That requires reading signals across a full context layer, not just the slice that lives inside any one tool.
Pryor described where Rumi.ai is taking this: a system that knows what you care about — which deals, which renewals, which accounts — and surfaces changes automatically. “Anytime there is a change to that, you’ll get an alert.” Not a dashboard you have to check. Not a weekly digest. A signal, when something shifts, with enough context to understand why.
That’s the direction customer intelligence is moving across the post-sales space. Not smarter dashboards. Not better summary emails. Continuous monitoring of the full signal stack, with intelligent routing that puts the right information in front of the right person at the moment it’s actionable.
Faltorusso’s framing holds here too: the goal is not to add more tools that CSMs have to check. It’s to build a system where the mechanical work happens automatically, and the CSM’s time is spent on what only a person can do. “I guarantee everything you do will only be amplified,” she said, on adding automation intelligently. The amplification only works if the automation is reading a complete picture.
The Architecture That Makes It Real
If you were to sketch out what a real customer intelligence system looks like, one that goes beyond a CRM and a product analytics tool bolted together, it has a few distinct layers.
The data layer unifies every source: CRM, product usage, support, billing, email, calls, and external signals. Not just pulling the data in, but modeling the relationships between entities — accounts, contacts, events, subscriptions, interactions — so that the connections between data points are preserved, not just the data points themselves.
The signal layer identifies patterns across that unified data that are meaningful, not noise. A single missed check-in is noise. A pattern of reduced engagement from a key contact, combined with three support tickets and a drop in product adoption over six weeks, is a signal.
The intelligence layer interprets those signals in context. Health forecasts, churn probability, expansion readiness, relationship strength — calculated continuously as new data comes in, not updated once a week by someone logging into a dashboard.
And the action layer is where it reaches the CSM: in Slack, in the CRM, as a pre-call brief. The right information at the moment it’s usable, not buried in a tool they’d have to go looking for.
Most CS teams have pieces of this. Very few have all of it connected. And the gap between “pieces” and “connected” is where most churn hides.
Conclusion
The context problem in customer success is structural. The information needed to be genuinely proactive is scattered across too many systems, too much of it entered manually, too little of it unified into something that reads as a complete account picture before something goes wrong.
Customer memory graphs built from the full signal stack — meeting data, product telemetry, support history, email sentiment, external signals — are what make the complete picture possible. The CSMs who will define what excellent post-sales looks like in the next five years aren’t the ones who work harder to gather more context. They’re the ones whose systems do the gathering, so they can spend their time on the conversation no tool can have for them.
The signal was always there. The question is whether the infrastructure is built to catch it.
