28 structural factors, statistically weighted against actual churn outcomes, run by an AI agent that recalculates every account's risk the moment new data lands. It doesn't just flag the risk — it hands CS the next-best retention step to close it, before the customer reaches a decision.
Remove the structural reasons a customer might leave — before they reach a decision.
IRP reads structural risk before it shows up in NPS, health scores, or QBR sentiment.
Every vendor has their own IRP DNA — the structural pattern that determines who stays and who drifts.
The CS industry runs on signals — NPS, CSAT, health scores, usage data. Each one accurately measures what it's designed to measure. The problem is what they were never built to measure: whether the customer is actually set up to renew. They read the wrong person, at the wrong time, on the wrong dimension.
A CSAT survey is filled in by whoever picks up the phone. An NPS score reflects how one person feels about one interaction. The renewal decision is made by someone else entirely — the economic buyer who signed the contract and controls the budget. That structural gap is not visible in any signal.
By the time signals turn negative, the structural conditions that produced them have been in place for months. Signal-based CS catches churn after the customer has already begun drifting. IRP catches it while the conditions are still changeable.
Sentiment measures how the customer feels. Usage measures what they do. Neither measures whether the structural conditions for renewal exist — whether the champion has internal authority, whether leadership is engaged, whether the organisation has the domain experience to build a business case.
Signal-based CS reads what customers express — behaviour, sentiment, activity. Capability-led CS reads whether the structural conditions for success exist in the account, before any signal appears.
Same logic, different stage of the customer lifecycle — each with its own qualification discipline.
Every vendor has their own IRP baseline — a retention profile built from their own account data. Each customer account is scored against that baseline on a 1 to 5 capability scale. The output is not a health score. It is a gap: the delta between where the account sits and where it needs to be for renewal to be the likely outcome.
A low capability score on a low-correlation factor is acceptable. A low capability score on a high-correlation factor is where churn lives:
The scoring itself runs on all 28 factors. Each point on the charts below is a dimension — the rolled-up average of the 4 factors underneath it, kept to 7 points so the shape stays readable. Averaging also softens the picture: the most critical extremes live at the single-factor level and are sharper than any dimension score shown here.
A low capability score on a low-correlation factor is acceptable. A low capability score on a high-correlation factor is where churn lives — visible above as the gap between the purple baseline and the coral scorecard on People and External.
28 structural factors, one qualification discipline for retention — reverse-engineered from actual churn and renewal outcomes and weighted by statistical correlation.
Every vendor builds their own IRP baseline first — a retention profile specific to their product, their customer type, their go-to-market motion. It is built from their own account data and updates every quarter as the dataset grows. No two vendors have the same baseline.
Each customer account is scored against the vendor baseline on the same 28 factors. CSMs fill in a 1 to 5 capability score per factor — no scoring engine visible, just the questions. Where a CSM cannot answer a question, that gap is itself diagnostic.
Capability scores are weighted against the vendor IRP baseline. Accounts are ranked by weighted risk — gap × correlation coefficient — with the signal layer added for reference. The agent recalculates continuously as new scorecard data arrives, not on a fixed quarterly clock, so priorities shift the moment an account's picture changes.
Below is the CSM-facing artifact those risk scores and escalation flags land in every quarter — one sheet, hard numbers only, no scoring engine visible.
| Account | CSM | ARR | Renewal | Mo. left | IRP risk | Dimension flags | NPS | CSAT | Usage | Signal divergence | Action type | Priority |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Arcline SaaS | J. Meyer | €210,000 | Oct 2026 | 2 | 87 | 42 | 4.0 | High | Signal ↑ / IRP ↓ | Escalate | 1 | |
| Nordlight CX | S. Okafor | €95,000 | Dec 2026 | 4 | 58 | 8 | 4.3 | Med | — | Monitor | 2 | |
| Blueshift IO | J. Meyer | €340,000 | Feb 2027 | 6 | 14 | 9 | 4.7 | High | — | — | — | |
| Marrow Health | R. Dupont | €128,000 | Jan 2027 | 5 | 22 | 8 | 4.4 | Med | — | — | — |
A single quarter tells you which accounts are at risk. A confirmed DNA pattern changes how the whole organisation operates.
Recurring structural problems become documented, repeatable interventions — not one-off fixes reinvented every quarter.
Skills gaps revealed by IRP DNA become the basis for a structured capability programme.
If domain experience is a systemic blindspot, it becomes an explicit hiring criterion — for CSMs and for the customers you target.
Structural gaps CS cannot close internally become partner requirements instead of quietly-accepted risk.
Accounts that arrive without a strong champion are flagged before CS ever inherits them.
DNA doesn't appear in one scoring cycle. It's confirmed over four.
Gainsight, Totango, ChurnZero, Planhat are CS operating systems — well-built for what they do. What they do not do is measure structural conditions.
Gainsight · Totango · ChurnZero
Ideal Retention Profile
IRP is not sold by the hour and it is not a software subscription. Each tier reflects a different depth of engagement.
A 30-minute conversation is enough to talk through what your IRP baseline might reveal. This goes straight to Romek's inbox.