Published 2026-09-03 · LuckyMDM Blog
In short: delinquency rate is the single most quoted number in device rental and device-as-a-service, and it is almost always quoted without its denominator. That matters more than it sounds. Take one portfolio — 1,000 active leases, 1,250 units, USD 2.4M in net receivables — with 60 contracts past due covering 105 units and USD 316,800 of balance. Depending on the denominator, the same portfolio is running at 6.0% (by contract), 8.4% (by unit) or 13.2% (by receivable balance). Same business, same month, a 2.2x spread. This page defines the three denominators, the two numerators, the aging and roll-rate views that turn a snapshot into a trend, and closes with a five-metric table you can copy, each row carrying its own denominator and threshold.
Ask ten operators how their portfolio is doing and nine of them will answer with one number. It is usually a percentage, it is usually in the single digits, and it is almost never accompanied by the one sentence that would make it meaningful: over what?
Delinquency rate is a fraction. Everyone agrees on that. The disagreement — and the reason two operators can look at the same portfolio and disagree by a factor of two — is entirely about which quantity sits on the bottom of that fraction.
This page is about fixing that, because until the denominator is written down, none of the downstream decisions have a foundation. Underwriting changes, collections staffing, when to stop scaling: all of them hang off a number whose definition is usually still in someone's head.
Start with a portfolio that is structurally normal for a mid-size operator.
| Denominator | Numerator | Denominator | Rate | What it actually answers |
|---|---|---|---|---|
| By contract | 60 contracts | 1,000 contracts | 6.0% | How customers are behaving |
| By unit | 105 units | 1,250 units | 8.4% | How much of the asset base is impaired |
| By balance | USD 316,800 | USD 2,400,000 | 13.2% | How much money is actually exposed |
None of the three is wrong. Only one of them can be compared against your cost of capital.
This is not an artefact of the example, and it is worth understanding, because the direction of the gap is itself a risk signal.
Multi-unit contracts sit disproportionately in the high-balance, high-default end of the book. Organised fraud does not lease one phone; it leases three flagships on one application, declines to negotiate, and asks for same-day dispatch. When a multi-unit contract fails, it fails as a batch. So:
So the practical answer is not to pick one. Run two as a minimum: balance-weighted for the solvency question, unit-weighted for the capacity question.
With the denominator settled, the numerator still has two versions, and conflating them is the second most common error.
Gross delinquency counts every account that failed to pay on schedule. It is a state measure: how much of the book is in breach right now.
Net loss counts only what was genuinely lost after recoveries — payments collected late, deposit applied, device recovered and resold. It is an outcome measure.
The bridge between them is one line:
Net loss rate ≈ gross delinquency rate × (1 − recovery rate)
At a 13.2% gross rate, a recovery operation that returns and redeploys 40% of everything entering collections produces roughly 7.9% net. At 20% recovery, the same book produces 10.6% net.
That line is the most useful thing on this page, because it shows there are two independent levers. Most operators spend all their effort on the first one — tightening underwriting to shave a point off gross. The second lever is usually cheaper and sits entirely inside operations: how fast collections start, whether the device can still be reached, whether the return-to-stock process is clean.
A delinquency rate is a photograph. It cannot tell you whether the book is improving or deteriorating; it can only tell you what it looked like on the day it was taken. Two additions fix that.
Split past-due balance by days delinquent — conventionally 30 / 60 / 90 / 90+. The shape of that split tells you something the headline rate cannot: whether the problem is fresh and still collectable, or has already cured into something that only a recovery process can address.
A working rule of thumb: once more than 30% of past-due balance sits in the 60+ buckets, collections started too late. Not "collections are ineffective" — too late. The money did not become uncollectable; it became uncollectable while it was waiting.
Roll rate measures movement between buckets. Specifically: of the balance sitting in 30-day delinquency at the close of last month, what proportion has moved into 60-day delinquency by the close of this month?
Roll rate (30→60) = balance that moved from 30 to 60 ÷ last month's closing 30-day balance
If USD 100,000 was in the 30-day bucket last month and USD 35,000 of it aged into the 60-day bucket this month, the roll rate is 35%.
Roll rate is the earlier warning, and that is the whole argument for tracking it. When the delinquency rate moves, the loss has already happened. When the roll rate moves, the loss is happening and the window is still open. As an operating cadence: watch roll rate monthly, review delinquency rate quarterly.
Do not adopt these thresholds as targets on day one. Adopt the definitions, run them for one quarter without judging anything, and then set targets against your own baseline. A borrowed threshold is a borrowed portfolio.
| Metric | Denominator | Formula | Reference threshold | Cadence |
|---|---|---|---|---|
| Delinquency rate (balance-wtd) | Net receivables | Past-due balance ÷ net receivables | Top-quartile operators run well under 10%; two consecutive months of increase is the trigger | Monthly |
| Aging mix | Past-due balance | Share in 30 / 60 / 90+ buckets | 60+ share at or below 30% | Monthly |
| Roll rate 30→60 | Prior month 30-day balance | Aged-in balance ÷ prior 30-day balance | At or below 35%; a rise here outranks a rise in delinquency | Monthly |
| Recovery rate | Units entering collections | Units recovered and redeployable ÷ units entering collections | 50% is a reasonable starting expectation; calibrate against your own four-quarter history | Quarterly |
| Net loss rate | Net receivables | Gross rate × (1 − recovery rate) | Must sit below unit contribution margin, or scale destroys cash | Quarterly |
Two notes on reading this table. First, the 50% recovery figure is a starting expectation, not an industry statistic — recovery performance is highly specific to geography, device mix and how early the process starts, so calibrate it yourself. Second, the last row is the only one tied directly to solvency: if net loss exceeds unit contribution margin, every additional unit placed makes the business larger and poorer at the same time.
Four of the five metrics above are financial and come straight out of the billing system. The fifth — recovery rate — does not, and neither does the denominator of the roll rate once you start segmenting by whether a unit is still reachable.
Those depend on device-side facts: is this unit still enrolled, when did it last check in, do commands still land, has it been wiped. None of that arrives automatically in a finance report. It has to be modelled as fields and carried into the same table.
This is the practical reason platforms such as LuckyMDM exist in the stack: not to produce the delinquency rate, but to keep enrolment state, last check-in time and command acknowledgement as ordinary columns in the asset register, so that recovery and roll rates are computed from observed data rather than assumed. Most operators whose risk reporting "stops working after a quarter" have not chosen the wrong metrics — they have left the device-state columns empty.
Delinquency fell from 6% to 5%. Possibly underwriting improved; possibly the month's originations were unusually large and inflated the denominator. A trend requires a like-for-like series, or better, a roll rate.
Charging off an account and then removing it from the dataset keeps historical delinquency permanently flattering, at the cost of destroying the only calibration sample you had. Keep the record, tag it, report it separately.
This is the original sin. Any cross-company comparison should begin by aligning denominator, numerator and measurement date. Without that, the comparison produces a conclusion with no content.
The definitions are not overkill; the full dashboard probably is. Minimum viable version: write the denominator down somewhere permanent, and split aging into three buckets. Both are spreadsheet work.
Balance-weighted. It is the one that answers whether exposure is inside the margin that pays for it, and it is the one that can be compared to funding cost.
Roll rate. Delinquency is the outcome; roll rate is the process, and acting on the process while the window is open is materially cheaper.
It improves the accuracy of two of them and the controllability of one. Enrolment and command-reachability data make recovery and roll rates measurable; graded restriction and reliable check-ins make recovery rate movable. It does not change why a customer stopped paying, and capability is always conditional on enrolment state, connectivity, OS version and the authorisation relationship — it reduces risk and shortens detection lag, it does not guarantee an outcome.
It moves it, within limits. A deposit changes the economics of opportunistic default but does nothing about organised fraud, which is a pattern problem: clustered applications, resale-dense delivery addresses, reused devices. Deposits do not detect patterns. Deposit sizing also runs into consumer-protection limits in most markets, so it is a bounded lever, not a strategy.