Promise-to-Pay Analytics: Why a Commitment to Pay Is Not the Same Thing as a Payment

A promise-to-pay is an engagement signal—not cash, cure or recovery. Its value emerges only when the commitment becomes timely, sufficient and durable economic improvement.

The Entimema Promise-to-Pay Architecture

Kept requires explicit amount and timing architecture

“Kept” may mean exact amount, minimum agreed amount, payment within a documented tolerance or full arrears clearance. There is no universal standard, but the definition must be explicit and stable.

Partial payment has economic meaning but is not full promise fulfilment. Preserve amount, date and tolerance evidence rather than forcing every outcome into one Boolean.

Promise credibility is path-dependent

Where lawful and available, Promise Burden = Promised Amount / Available Capacity can support plausibility analysis without prescribing a threshold. A commitment far above demonstrated capacity may be less credible and may indicate the need for deeper support or restructuring assessment rather than repeated short-term promises.

Track N prior promises, N broken promises, historical PKR, time between promises and amount progression. A first promise can be informative engagement; Promise → Break → Promise → Break can become a deterioration signal. Broken severity should distinguish no payment, partial payment, late payment and repeated failure.

Keeping a promise is not curing an account

A borrower can keep a small promise yet remain delinquent. Compare Promised Amount / Arrears and Actual Payment / Arrears where relevant, and link payment timing to subsequent roll state.

A 65% promise-keeping rate becomes only 29% durable recovery. Cure & Re-Default Analytics measures whether cure survives. Roll Rate Analysis can compare 30 DPD → Current after kept PTP with 30 → 60 DPD after broken PTP.

Promise metrics are conditional on contact

Digital, human and self-service channels can generate different promise volumes and fulfilment, but selection matters. One channel may produce more promises with lower PKR; another fewer promises with stronger cure conversion. Compare the whole funnel and customer mix.

If incentives reward only PTP volume, teams or systems can generate unrealistic amounts and dates. Monitor quality at process level rather than using punitive individual surveillance.

Payment after a promise is not payment because of the promise

Customers willing to promise may already be more engaged, liquid and likely to cure. Observational P(Cure | PTP) therefore combines selection and treatment. PTP uplift is not directly identifiable without assumptions or governed design.

Champion/challenger testing may compare acceptable reminder timing, structures or channels, but must never withhold required support, hardship options or mandatory communications. Champion / Challenger Strategy supplies the wider governance discipline.

Broken promises can update risk without becoming a judgement

Upcoming promise, due-today status, partial fulfilment and repeated broken promise can update Collections Prioritisation, but priority still combines exposure, risk, recoverability, contactability and intervention value.

A broken PTP can precede roll-forward, default or re-default and can feed Early Warning or Behavioural Credit Scoring. Avoid circularity when scores drive the treatment that generates PTP data. A broken promise does not prove deliberate non-cooperation.

Promises support cash forecasting only after probability weighting

Expected cash is €667k rather than €1.20m. Track Forecast Error = Actual Cash − Expected PTP Cash by promise vintage. Uncertainty remains, and the same €500 collected now has greater present value than €500 much later—connecting PTP timing to IFRS 9 LGD.

PTP vintages make strategy and maturity visible

Collection-entry vintage, contact vintage and PTP vintage answer different questions. A promise due next week is not broken, and a newly cured account is not sustainably cured. Recent cohorts are censored and must not be compared with mature cohorts at unequal observation horizons.

Interpret changes through customer mix, contactability, amount realism, strategy version and macro context rather than attributing every KPI movement to treatment.

Fifteen thousand delinquent accounts become 1,350 sustainable cures

The 64.4% PKR becomes sustainable cure for only 9% of the original delinquent population. Promise history materially separates durability, but the segment table remains descriptive rather than causal.

Non-bank portfolios make promise evidence mature quickly

High contact volumes, rapid payment cycles and short tenors make PTP analytics operationally valuable. A promise several weeks away can consume a large share of remaining product life, so Days to Promise should be interpreted relative to product velocity.

Where broken promises are common, a binary PTP flag becomes weak. History, amount fulfilment, delay, cure conversion and sustainable recovery provide stronger differentiation.

Common failure modes

A Promise-to-Pay Analytics Agent can track commitments—not pressure customers

A future Agent can ingest contact and PTP events, match promises to payments, classify full/partial/broken outcomes, calculate timing and amount fulfilment, track history, estimate credibility, monitor due queues, identify repeated breaks, calculate cure conversion, track sustainable cure, create probability-weighted cash forecasts and compare vintages for human review.

Its role is PTP tracking + promise credibility + cash forecasting + cure analytics. It must not autonomously engage in intrusive contact or take ungoverned customer action.

Related research

Continue with Collections Prioritisation, Cure & Re-Default Analytics, Behavioural Credit Scoring, Early Warning Systems, Roll Rate Analysis and IFRS 9 LGD.