Your NPS score is lying to you.
NPS tells you how someone feels today. It does not predict what they will do tomorrow. ethyx applies behavioural science to predict software churn — and prescribe what to do about it.
- Peer-reviewed methodology
- 250,000+ participants validated
- 100% ROI guarantee
- Privacy-first, consent-based
of companies see zero meaningful correlation between NPS and who actually churns.
ProfitWell / Paddle NPS–Retention Analysis
more expensive to acquire a new customer than to retain an existing one.
Bain & Company
increase in profits from just a 5% improvement in customer retention.
Bain & Company
NPS is not predictive of churn.
You're making million-dollar retention decisions based on a metric that has no predictive validity for individual churn. The signals that actually predict behaviour are sitting in plain sight — you're just not measuring them.
Meta-study
265K
companies, zero evidence
A 10-year study across 265,000 companies found no statistical evidence that customers with poor NPS scores churn faster than those with high scores.
Customer Cross / Skalin, 2023
Metric fatigue
8th
NPS is falling out of favour
NPS dropped from the 2nd most popular CX metric in 2023 to 8th place in 2024. Companies are abandoning it because it doesn't predict what matters.
CMSWire State of Digital CX, 2025
Signal quality
54%
barely better than a coin flip
NPS has only 54% correlation with retention likelihood. That's a weak signal buried in noise — not something you should bet your retention strategy on.
Multiple 2024 SaaS retention studies
Selection bias
Survivors
only — the wrong people
NPS only measures customers who stuck around long enough to be surveyed. The people who already left — the ones you most need to understand — are invisible.
Sprig NPS Analysis
Even the creator of NPS now advocates for complementary metrics beyond NPS. He calls the way most companies use it — selective reporting, no transparency on response rates, scores tied to bonuses — a system that actively destroys its own validity.
lost to failed payments alone in 2025. One-third of all churn is involuntary — customers who wanted to stay.
Churnkey State of Retention 2025
median CAC payback period for SaaS. Every churned customer before 18 months is a net loss.
Benchmarkit 2025
increase in enterprise value from reducing churn by just 5 percentage points over five years.
McKinsey
A 12-item survey.
That's it.
No enterprise implementation. No months of integration. ethyx is a non-intrusive, 12-item behavioural survey grounded in the Transtheoretical Model of Change — the same peer-reviewed framework validated across 250,000+ participants in seven countries.
It can be embedded directly into your platform via API — fully white-labelled, matching your brand, triggered at the moments that matter (onboarding, renewal, post-support, any milestone). Or it can be conducted off-platform by your CS team as part of a check-in or QBR. Either way, under 4 minutes to complete. No survey fatigue. Every question is load-bearing.
The output doesn't just tell you who is likely to leave. It tells you what to do to keep them — specific, actionable prescriptions tailored to where each user sits on the behavioural commitment spectrum. Your CS team gets a playbook, not a probability score.
In-platform
API integration. White-labelled. Embedded in your product at key moments. Users never leave your app.
Off-platform
Conducted by your CS team during check-ins, QBRs, or renewal conversations. No tech lift required.
Traditional vs ethyx
NPS
Looks healthy
ethyx
High risk
Mentally left
Same customer. Same day.
The model gets smarter the longer it runs.
The Transtheoretical Model gives you validated behavioural classification from day one. But ethyx doesn't stop there. Our machine learning layer trains on your organisation's data over time — learning the patterns that are unique to your product and your users. The AI algorithm begins to draw correlations between feature usage patterns and churn outcomes. It models what users do in the days and weeks before they cancel — the login frequency drop, the feature they stop using, the support ticket they never filed. So you can see it happening before it hits your CS team's inbox. Over time, the model builds an organisational fingerprint — learning which behavioural signals are strongest predictors for your product specifically, not just SaaS in general. The longer it runs, the more precise the prescriptions become.
From first pulse to organisational fingerprint.
Validated classification on day one; precision improves as your data accumulates.
Behavioural classification
12-item survey classifies every user into 5 commitment tiers. Validated model works from first pulse.
Usage correlation begins
ML layer starts mapping feature usage patterns to commitment scores. Early signal detection.
Pre-cancellation modelling
Algorithm identifies what users do right before they cancel. Pattern recognition from your real churn data.
Organisational fingerprint
Fully trained on your data. Predictive accuracy improves continuously. Prescriptions get sharper with every cycle.
Measure. Predict. Prescribe.
Three steps from survey deployment to prescriptions your CS team can run with.
Survey
12 items. Under 4 minutes. Embedded in your platform via API (white-labelled) or conducted off-platform by your CS team. Every question maps to a validated behavioural construct — no filler, no fatigue.
Predict
Classify each user across five behavioural commitment tiers — from fully committed to already gone. Our ML layer then correlates feature usage, login patterns, and support interactions to refine the signal over time.
Prescribe
Not just who will leave — what to do about it. Each account gets specific CS actions, product recommendations, and intervention timing. The AI models what users do right before they cancel so you can act before it hits the inbox.
100% ROI guarantee
You're making million-dollar retention decisions based on a metric that has no predictive validity for individual churn. The signals that actually predict behaviour are sitting in plain sight — you're just not measuring them.
Sources: ChartMogul 2024 · ChartMogul SaaS Benchmarks (n=2,100) · High Alpha 2025
Not just who will leave. What to do about it.
Each account gets a churn risk profile — health status, risk score, confidence level, and AI-generated priority actions. Your CS team gets prescriptions, not dashboards to interpret.
The model surfaces the specific interventions that move the needle for each user. A new executive stakeholder who hasn't been onboarded. A paid feature going unused for 90 days. A usage pattern that mirrors accounts that cancelled last quarter. Your team sees it before the customer sends the cancellation email.
Customers who perceive fair value are 72% more likely to renew — even when shown cheaper alternatives. ethyx identifies the value gaps before your customer finds one.
Acme Corp — Enterprise
Last pulse: April 2026
- Risk score
- 38
- Confidence
- 82%
Priority actions
- New executive onboarding neededHigh
- Feature adoption gap: cohort analysisMedium
of all SMB customer losses happen in the first 90 days. If you're not measuring commitment at onboarding, you're already too late.
SaaS Churn Rate Benchmarks 2025
of customers who perceive fair value will renew — even when shown cheaper alternatives. The problem isn't price. It's perceived commitment.
McKinsey
CS, Product, and Revenue teams.
Different teams, different insights — all from the same behavioural data.
Minimal user friction
Deploy at Key Moments
Trigger surveys at onboarding, renewal, post-support, or any milestone. Catch commitment signals when they matter most.
User-level churn classification
Upsell Detection
Surface expansion and upsell opportunities that sentiment surveys miss entirely. Turn your retention program into a revenue channel.
Peer-reviewed methodology
Academically validated
Risk-free deployment
API Integration
Plugs into your existing tech stack — CRM, support tools, product analytics, and billing systems. No rip-and-replace.
Global evidence base
CS Team Prescriptions
Specific, actionable interventions for customer success — not generic retention playbooks. Each account gets tailored next steps.
Cohort Analysis
Compare churn risk across pricing tiers, onboarding cohorts, usage segments, and account size. See where commitment breaks down.
Product Insights
Understand which features drive commitment and which drive churn — from behavioural data, not satisfaction surveys.
See churn happening before it hits the inbox.
Usage drops before sentiment does. Our AI layer models the behavioural fingerprint of cancellation — what users do in the days and weeks before they churn.
Login frequency drops. Key features go unused. Support tickets dry up — not because they're happy, but because they've stopped caring. The model draws correlations between feature usage patterns and churn outcomes specific to your product. It doesn't just flag risk — it shows your product team which features drive commitment and which drive churn.
For SaaS companies over $50M ARR, expansion revenue from retained customers now surpasses new sales. Every user you keep isn't just a saved subscription — it's the expansion revenue, the referral, and the case study you'd never have had.
Benchmarkit 2025 SaaS Performance Metrics
Pre-cancellation signals detected
- -14 days
Login frequency dropped 64%
vs. 30-day average
- -10 days
Core reporting feature unused
21 days since last access
- -7 days
Admin seats reduced from 5 to 2
team disengagement pattern
- -3 days
Data export requested
matches cancellation pattern
Cancellation
ethyx flagged this account 14 days earlier
The Transtheoretical Model has been validated across decades and hundreds of thousands of participants. It predicts behavioural commitment, not sentiment. That's the difference between knowing someone is happy and knowing they'll stay.
What is churn actually costing your SaaS?
Most teams underestimate revenue impact because they only count lost subscriptions. The real cost includes acquisition spend, onboarding investment, and the expansion revenue you'll never see.
Include sales, marketing, onboarding, and implementation costs per account.
Total annual revenue lost to churn
$1,700,000
Lost ARR + wasted acquisition cost
Accounts lost per year
100
Wasted acquisition spend
$500,000
CAC invested in accounts that churned
With 15% churn reduction
$255,000
Potential annual savings
Projected ROI with ethyx
13x+
Estimated return on investment
Revenue impact based on lost ARR + sunk CAC. Excludes expansion revenue loss and brand/referral impact. Actual results vary.
Revenue impact based on lost ARR + sunk CAC. Excludes expansion revenue loss and brand/referral impact. Actual results vary.
Behavioural science, applied to software.
The same methodology that predicts workforce retention — validated across 250,000+ participants in seven countries — now applied to software user behaviour.
Peer-reviewed methodology
Grounded in published psychometric frameworks and the Transtheoretical Model of Change — one of the most validated frameworks in behavioural science.
7 countries validated
Applied across government programs internationally. The model works across cultures — unlike NPS, where a score of 7 means different things in different countries.
API integration
Plugs into your existing tech stack — CRM, support tools, product analytics, and billing systems. No migration, no rip-and-replace.
Built by product people
Devon Passmore brings a decade across SaaS, HR tech, and regulated industries. This isn't an academic experiment — it's built for your stack.
100% ROI guarantee
If we don't deliver the return, you don't pay. We can make this promise because our models are evidence-based and continuously improving.
Stop guessing why users leave.
Book a free initial chat. No pressure. No free trials. Just real insights and honest conversations.
Contact UsNo spam. No free trials. Just real insights and honest conversations. Contact Us