You’re focused on winning new customers
You’re focused on winning new customers
Existing relationships feel stable — until they’re not
Churn still surprises you
Losses feel personal, but only make sense afterward
This is about intuition breaking as you scale.
A founder-led B2B company experienced customer churn that felt increasingly personal and hard to predict.
The founder was still closely involved with key customers. Conversations were frequent. Relationships felt open. Yet several customers left within a short period, often described afterward as “unexpected”.
Each case could be explained in hindsight.
But none felt obvious beforehand.
We analyzed multiple years of historical customer communication data, focusing on how conversations evolved rather than what was explicitly said. The analysis included email exchanges, meeting cadence, response patterns, and changes in tone over time.
A consistent pattern emerged.
In most churn cases, clear intent shifts were already visible months earlier. Signals appeared as reduced initiative from customers, shorter replies, less future-oriented language, and subtle changes in confidence and commitment. These signals did not register as problems at the time and were easy to rationalize away.
By the time customers explicitly raised concerns or decided to leave, trust had already eroded.
After validating these patterns on historical data, a lightweight live setup was introduced using weekly updates from ongoing customer interactions. Early churn signals became visible almost immediately, well before customers voiced dissatisfaction.
Nothing changed overnight.
But the founder’s behavior did.
Today, the founder uses these signals to start earlier, more personal conversations — checking assumptions, restoring alignment, and addressing concerns while the relationship is still open.
As the founder reflected:
“I used to think I’d always feel this coming.
I realized I needed help seeing it earlier.”
Growth and pipeline dominate leadership attention
Dashboards look fine, renewals still create stress
Customer Success senses risk before metrics move
You’re surprised despite “having the data”
This is about false certainty and timing.
A European B2B SaaS scale-up with a dedicated Customer Success team experienced recurring churn that often came as a surprise.
Dashboards were mature. Health scores were in place. NPS was monitored. And yet, renewals regularly created pressure. Accounts that looked stable would suddenly downsize or leave, often explained only in hindsight.
We analyzed multiple years of historical customer interaction data to understand what happened before churn became visible in metrics.
The analysis revealed a consistent pattern.
In a large majority of churn and downsell cases, clear intent signals were already present three to six months earlier. These signals showed up as subtle changes in language, engagement rhythm, confidence, and responsiveness. Customer Success managers often sensed something was off, but lacked concrete evidence to escalate concerns early.
By the time health scores declined or usage dropped, customer decisions were largely made.
After validating these findings on static historical data, a production setup was implemented using weekly, dynamic data across active accounts. From the first weeks, meaningful churn signals became visible well ahead of renewal moments.
Nothing dramatic changed overnight.
But leadership gained time.
Today, the CEO and Customer Success leaders use these signals to guide earlier internal escalation and start renewal conversations while there is still room to reset expectations and alignment.
As the CEO summarized it:
“We weren’t missing data.
We were missing timing.”
New deals get more attention than existing customers
Evaluations are positive, churn still happens
You explain losses only in hindsight
Conversations start after trust already shifted
This is about preventing avoidable churn.
A European professional training company experienced recurring customer loss that often felt sudden.
Courses were well evaluated. Relationships felt solid. Yet contracts weren’t renewed as often as expected, sometimes with little warning. In hindsight, leadership could usually explain why customers left — but only after the fact.
We analyzed three years of historical customer interaction data to understand what happened before churn became visible in numbers.
The analysis showed a clear pattern.
In roughly 80% of churn cases, strong churn signals were already present six months before customers left. These signals did not appear as complaints or explicit feedback, but as subtle changes in language, timing, confidence, and engagement.
In most of these cases, there was still sufficient trust in the relationship at the time the first signals appeared. This suggests that a meaningful part of the churn could likely have been prevented if the right conversations had started earlier — when change was still reversible.
After validating the MVP on static historical data, a production setup was implemented using weekly, dynamic data.
Significant churn signals became visible from the first week of live monitoring.
Nothing dramatic changed overnight.
But the company was no longer surprised.
Today, the founders use these signals to start earlier, more focused conversations — both with customers and with trainers — while trust is still intact and alignment can still be restored.
As one of the founders reflected:
“The signals were there.
We just weren’t looking at them yet.”
Teams focus on delivery and new work
Clients disengage without complaining
Drift is sensed, not measured
Post-mortems come after the loss
This is about hidden relational drift.
A European B2B services firm with long-running client relationships experienced recurring contract losses that often felt unexpected.
Projects were delivered as agreed. Client satisfaction was generally positive. Account teams felt relationships were stable. Yet contracts were not renewed, sometimes after years of collaboration, with limited warning.
We analyzed several years of historical client interaction data across a selection of key accounts. The focus was not on delivery outcomes, but on how collaboration evolved over time.
The analysis showed a clear pattern.
In most lost accounts, relational drift was already visible four to eight months before contracts ended. Signals appeared as subtle changes in meeting dynamics, slower response times, reduced initiative from the client side, and shifts in shared language around goals and ownership.
These signals did not register in financial systems or project dashboards. Internally, teams often sensed that “something felt different”, but could not pinpoint when or why alignment was weakening.
After validating these patterns on static historical data, a live setup was introduced using weekly, dynamic inputs from ongoing client interactions. From the first weeks, early signs of weakening relationships became visible across multiple active accounts.
Nothing dramatic changed overnight.
But conversations changed.
Today, leadership and account teams use these signals to start earlier, more focused conversations with clients and internal teams — addressing misalignment while trust is still present.
As one partner reflected:
“We didn’t lose clients because of one moment.
We lost them because we noticed the drift too late.”
