How a Series B SaaS Company Reduced Churn by Detecting Silent Detractors
High NPS scores masked a hidden churn problem. Voice interviews revealed emotional signals that surveys completely missed.
The Challenge
This Series B SaaS company had a problem that didn't look like a problem. NPS scores were strong. CSAT surveys came back positive. Customers said “everything's fine” in check-in calls.
Then they didn't renew. Quarter after quarter, accounts that appeared healthy churned without warning. The CS team was blindsided. Leadership suspected a product gap, but the data didn't support it. Something was being missed, and surveys couldn't find it.
The Approach
ReadingMinds deployed Emma to conduct voice interviews with 50 accounts in their renewal window. Conversations focused on product experience, support interactions, and overall satisfaction.
The difference: Emma didn't just capture answers. She detected emotional undertones (resignation, declining enthusiasm, forced positivity) and flagged accounts where the words said “fine” but the voice said “done.”
50
Accounts interviewed
Voice
Emotion detection active
Renewal
Window timing
Key Findings
34% of “satisfied” accounts showed resignation
More than a third of accounts that rated themselves as satisfied showed clear emotional signals of resignation and declining enthusiasm in voice responses.
“Fine” with low energy = 4x churn risk
When customers said “fine” or “good” but delivered it with low energy and flat affect, they were 4x more likely to churn than genuinely satisfied accounts.
Support friction was the hidden driver, not product gaps
The churn driver wasn't missing features. It was accumulated frustration with support response times and feeling “unheard,” invisible in product usage data.
Results
28%
Reduction in unexpected churn
Proactive intervention prevented losses that would have gone undetected.
12
At-risk accounts identified in week one
Emma flagged accounts that NPS scores said were healthy.
6 of 8
At-risk accounts saved after CS intervention
The CS team intervened on 8 flagged accounts and retained 6.
“We thought we had a product problem. It was a relationship problem. ReadingMinds heard what our surveys couldn't.”
VP Customer Success, Series B SaaS
About this case study: This is an anonymized design partner story from our early access program. Company details are generalized to protect partner confidentiality. Emotion signals referenced are derived from transcript analysis; no permanent voice recordings are stored. All findings include intensity scores and link to source transcripts.
Get Similar Results
Detect the churn signals hiding in plain sight. See how ReadingMinds can help your CS team intervene before it's too late.