Last updated: August 27, 2026
Every collections team celebrates a promise to pay. The customer picked up, the conversation went well, and they committed to a date and an amount. But a promise is not a payment. Teams that obsess over how many promises they secure — and ignore how many are actually kept — end up with inflated forecasts, unreliable cash flow, and a queue full of accounts that need to be worked twice. This guide covers what promise-to-pay metrics really measure, why promises break, and a practical playbook for converting more commitments into cleared payments.
A promise to pay (PTP) is a debtor's explicit commitment to pay a specific amount by a specific date. In collections operations it shows up as two related metrics, and confusing them is the most common reporting mistake in the function.
The PTP rate measures the percentage of contacts that end in a commitment. As Tratta's guide to debt collection KPIs defines it, you divide the number of contacts resulting in a promise by total contact attempts that reached the right party. It tells you whether your scripts, timing, and negotiation approach are landing — but nothing about whether money actually arrives.
The kept rate divides promises fulfilled by promises made. This is the number that belongs in your cash-flow forecast. A team securing commitments on half its calls but converting only a fraction of them into payments is manufacturing false confidence: the forecast looks healthy while the aging report quietly deteriorates. When you evaluate collectors or AI agents, weight kept promises over gross promises — a smaller book of reliable commitments beats a large book of optimistic ones.
Broken promises are rarely random. They cluster around a handful of causes, and each one has a specific operational fix.
| Why the promise breaks | What it looks like | The fix |
|---|---|---|
| Cash timing mismatch | Customer committed to a date before their own receivables or payroll landed | Anchor due dates to the customer's inflow calendar, not yours |
| Payment friction | Customer intended to pay but the portal, transfer, or approval chain got in the way | Send a direct payment link in the confirmation message |
| Memory decay | The promise was sincere but life moved on | Confirm in writing immediately and remind before the due date |
| Wrong channel | Commitment extracted under pressure on a call the customer wanted to end | Follow up on the channel the customer actually uses |
| Overcommitment | Amount was more than the customer could realistically clear | Offer structured partial payments up front |
Channel matters more than most teams assume. McKinsey's research on digitizing collections found that customers contacted through their preferred channels were 12 percent more likely to make a payment, and paid in full more often. A promise negotiated on a channel the customer tolerates is worth less than one negotiated on a channel they actually live in.
"I'll pay soon" is not a promise to pay. Train collectors — human or AI — to close on three specifics: exact amount, exact date, and payment method. Vague commitments should be logged separately so they don't pollute your kept-rate denominator. Specificity also changes debtor psychology: a named date and amount creates a concrete obligation rather than a mood.
The moments after a verbal commitment are your highest-intent window. Send a written confirmation within minutes summarizing the amount, date, and a one-tap payment link. This removes friction and converts a share of promises into immediate payments — the best kept promise is one fulfilled before the due date arrives. If you already run automated payment reminders, wire PTP confirmations into the same sequence engine.
Most teams contact the customer only after the promise fails, which turns a routine nudge into an awkward second collection call. A friendly reminder 48 hours before the committed date, with the payment link repeated, protects both the payment and the relationship. Post-break outreach should happen fast too: the first day after a broken promise is when re-commitment rates are highest.
A first-time promiser with a strong payment history and a customer on their third broken promise deserve different treatment. Score each open promise on history, balance size, and engagement signals, then route high-risk promises to earlier, more personal follow-up. Teams that apply this kind of segmentation see the effect show up directly in debt collection recovery rates.
When a promise breaks, don't simply re-ask for the same commitment. Diagnose the cause from the table above and change one variable: split the amount, move the date to align with the customer's cash cycle, or switch channels. Repeating a failed ask produces serially broken promises and trains the customer that commitments to you are costless.
This is also where automation earns its keep. Following up every open promise, on time, across WhatsApp, email, and phone, at portfolio scale is exactly the kind of repetitive, deadline-driven work humans drop when queues get long. An AI collections agent like Darwin's Rio confirms each promise in writing, sends the pre-due-date reminder, detects the broken promise the day it happens, and opens the renegotiation conversation — so your human collectors spend their time on the accounts that need judgment.
The same McKinsey research found payment rates from digital channels were uniformly higher than traditional outreach — 46 percent for online banking messages and 44 percent for mobile push, for example. Digital-first programs also compound on cost: lenders implementing them have seen their cost of collections fall by at least 15 percent while resolution rates improved. If your promise-to-pay workflow still runs on outbound dialing alone, channel expansion is likely your single biggest kept-rate lever.
For large balances, an all-or-nothing promise is fragile by design: one cash-flow hiccup on the customer's side and the entire commitment fails. Converting big promises into short structured plans — two or three scheduled installments with automatic reminders — trades a small delay in full recovery for a much higher probability of collecting anything at all. The same logic applies to voice outreach: if your team uses calls for high-value accounts, pairing them with digital confirmation and self-service payment (an approach we break down in our guide to AI voice agents for collections) keeps the commitment alive after the call ends. And because every broken large promise eventually lands in your dunning workflow anyway, designing the plan up front is cheaper than repairing the relationship later.
A promise-to-pay program is only as honest as the metrics around it. Four companions matter most.
Right party contact (RPC) rate. Promises can only come from conversations with the right person. If RPC is low, fix data quality and channel mix before blaming negotiation skills.
Kept-promise value. Track the dollar value of kept promises, not just the count. Ten kept promises of $100 and one broken promise of $50,000 is a bad month that count-based reporting will call a good one.
Roll rates. Watch how accounts with broken promises move across aging buckets. Serial promise-breakers rolling deeper into delinquency are candidates for a different strategy — structured plans, settlements, or escalation — as covered in our B2B collections playbook.
DSO. The ultimate downstream check. If kept rate is rising but days sales outstanding isn't moving, promises are being kept on paper while new delinquency forms elsewhere — our guide to reducing DSO covers the full-funnel view.
A promise to pay (PTP) is a debtor's explicit commitment to pay a specific amount by a specific date, usually secured during a collections call or digital conversation. It is recorded and tracked so the team can follow up and measure whether the commitment is fulfilled.
Divide the number of promises fulfilled by the total number of promises made in the period, then multiply by 100. Count a promise as kept only if the committed amount arrives by the committed date (or within your defined grace window), and keep that definition consistent.
It varies widely by industry, portfolio age, and how strictly promises are defined. Rather than chasing an external benchmark, baseline your own kept rate, segment it by debtor history and channel, and measure improvement against your baseline.
Usually yes, and quickly — continuity preserves context and the relationship. What should change is the offer: a smaller amount, a date aligned to the customer's cash cycle, or a different channel. Repeating the identical ask is the most common way to create a serially broken promise.
Turn more promises into payments. Rio, Darwin's AI collections agent, confirms every PTP in writing, reminds before the due date, and renegotiates broken promises the day they happen.
Meet Rio →