There is no credible asset recovery rate benchmark to borrow. Every published figure comes from a vendor using its own definition, with no independent body checking it. The useful move is to define the measure yourself, count it consistently, and judge vendors on your numbers rather than theirs.

TL;DR

  • A percentage without a stated denominator is not a benchmark.
  • Most disagreement is about what counts as a device that should have come back.
  • Timing of the trigger moves recovery more than vendor choice does.
  • Cost per recovered device is a better question than rate.
  • Put the definition in the contract before the number.
  • The trend across quarters matters more than the level in any one.

When you do not need to measure this

Four stages, and the first two do not justify a metric.

Everything comes back. A handful of exits a year, all local, all returned. You do not have a recovery problem and a percentage would be noise.

One or two go missing. Irritating, visible, still anecdotal. Fix the trigger timing before you build measurement, because that alone often clears it.

A pattern forms. Enough machines are unaccounted for that somebody asks how many. This is where a definition becomes worth writing down, and where most teams start counting inconsistently instead.

Somebody external asks. A customer, an auditor or a board paper wants a figure. Now the definition matters more than the number, because you will be asked how it was calculated.

Why an asset recovery rate benchmark is hard to borrow

A recovery rate looks like a simple fraction. Devices back over devices owed. The trouble is that both halves are judgement calls, and nobody has agreed on either.

Take the numerator. Does a device count when the courier collects it, when it arrives, or when somebody confirms the serial number matches the one you issued? Those are three different dates and three different numbers, and the gap between the first and the last is where machines disappear.

The denominator is worse. Does it include contractors? Machines the employee claims were already returned? Devices from a market you exited? Collections still open at the period end? Each exclusion is arguable and each one lifts the rate.

There is no standards body for this. Unlike data sanitisation, where NIST Special Publication 800-88 gives everyone a shared vocabulary, recovery reporting has no equivalent. So a vendor rate is a self-report with no audit behind it. That does not make it false. It makes it incomparable.

The nearest thing to a governing expectation is indirect. Asset management controls under ISO/IEC 27001 expect an organisation to know what equipment it holds and to account for its return. That is an inventory obligation rather than a performance target, and the distinction is worth keeping. Nothing requires you to hit a particular rate. Plenty expects you to know what your rate is and how you arrived at it.

This is why borrowed figures cause trouble in board papers. A number lifted from a vendor page describes that vendor’s book of business, not your fleet, your markets or your exit process. Two companies with identical performance can report rates ten points apart purely through counting choices, and neither is lying.

So the honest position is that a benchmark is the wrong thing to shop for. What you want is a definition you can defend and a trend you can act on.

Four ways the number moves without anything improving

  • Excluding open collections. Count only closed cases and anything stuck is invisible. The slowest, hardest recoveries are precisely the ones still open.
  • Counting the booking, not the box. A scheduled pickup is treated as a recovery. First-attempt failure at residential addresses is common, so this gap is not small.
  • Measuring only operated markets. Rates from countries the vendor serves directly, excluding partner territories, which is where performance is weakest.
  • Resetting the clock. A second attempt starts a new case, so the original failure leaves the data. Each case then looks fast and the device still is not back.

None of these is fabrication. Each is a defensible reporting choice. Stacked together they produce a figure that is technically accurate and practically meaningless, which is why the definition has to come before the number.

Five questions before you accept a rate

What is the denominator? Ask for it in words. If the answer is vague, the rate is too.

At what event does a device count as recovered? Collection, arrival or serial verification. Only the third is real.

Are open cases included? They should be, counted as not yet recovered.

Does this cover every market or only operated ones? Ask for the split rather than the blended figure.

What is the cost per recovered device? This single question reframes the whole conversation, because it prices failure rather than hiding it.

Who reports this, and where each one struggles

Named honestly, including the one we have an interest in.

Your own spreadsheet. Best for honesty, because you choose the definition and nobody benefits from flattering it. It struggles on completeness, since it only knows what somebody remembered to record.

Your asset management system. Best for tying recovery to the register, which is where the question ultimately lands. It struggles because most systems record status changes rather than physical events, so a device marked returned may not be.

Equipment lifecycle platforms such as Firstbase, Workwize, GroWrk and allwhere. Best for reporting generated from the work itself rather than typed in afterwards. They struggle on comparability: each defines recovery its own way, so two dashboards cannot be set side by side.

Local ITAD firms. Best on documentation for what they receive. They struggle on rate, because they only see devices that arrived and have no view of the ones that never started.

RemoAsset. Disclosure: RemoAsset is owned by the same people who publish this site. Because collection, wiping, storage and reassignment run through one service across 80 or more countries, the recovery event can be tied to a verified serial rather than a courier scan. Where it struggles: it reports on what it handles, so devices never entered into the process are invisible to it, exactly as they are to every other vendor. Ask how its retrieval reporting defines a recovery before comparing any figure.

The decision table

SituationScaleSetupPrimary painRecommended starting point
Few exits, all returnedUnder 50No IT functionNoneDo not measure, just record
Occasional losses50 to 150One ops ownerAnecdote, not dataMove the trigger earlier first
Pattern of unreturned machines150 plusIT teamNo agreed definitionWrite the definition, then count
Comparing vendorsAnyProcurementRates are not comparableAsk for cost per recovered device
Board or audit reportingAnyCompliance ownerMust explain the methodPublish the denominator with the rate
Many countries200 plusSmall IT teamBlended figure hides weak marketsReport per market, never blended

What a flattering number costs

A rate that is too generous does not just mislead a board paper. It stops you fixing the thing it is hiding, because a healthy metric is an argument against investigating.

The concrete version: a vendor reports a strong rate by excluding open cases. Twenty machines sit permanently open, never failing and never closing. Nobody chases them because the dashboard looks fine. A year later somebody counts the asset register against reality and finds the gap, usually during diligence.

The second cost is choosing the wrong vendor. Comparing two self-defined rates picks the one with the looser definition rather than the better service. Cost per recovered device is harder to game, because the denominator is money you actually spent.

So the question is not what rate you should expect. It is whether your current number would survive somebody asking exactly what it counts, and whether you would be comfortable with the answer.

When a percentage stops being the right measure

A single rate works while the question is whether things are broadly working. It stops working once you need to improve something specific, because a percentage tells you the result and nothing about the cause.

Two measures replace it usefully. Days from exit to verified receipt, which exposes the slow drift a monthly rate hides. And cost per recovered device, which stops a cheap vendor looking good while quietly recovering less.

Both need the same foundation: a record tying each device to each event, captured as the work happens rather than reconstructed later. That is also what an auditor asks for, so the measurement and the evidence are the same artefact. Seeing how lifecycle tracking produces that record is a reasonable starting point if you are currently counting by hand.

Related reading

Frequently Asked Questions

Is there a published asset recovery rate benchmark?

Not a credible one. Rates are quoted by vendors from their own data, with their own definitions, and no independent body audits them. A figure you cannot see the denominator for is marketing rather than a benchmark, so treat any percentage without a stated definition as unusable.

How should recovery rate be defined?

Devices physically received and verified by serial number, divided by devices that should have come back in the same period. Both halves need stating. Most disputes come from the denominator, where machines written off, abandoned or still in progress are quietly excluded.

How do vendors inflate the number?

By excluding collections still open, counting a booked pickup as a recovery, measuring only markets they operate directly, or resetting the clock when a second attempt is made. None of these is dishonest in isolation. Together they can move a rate by a wide margin.

What rate should we actually aim for?

Aim for a number you measure consistently and improve, rather than one you borrowed. A rate that is slightly lower but honestly counted is more useful than a higher one you cannot reproduce. Track the trend across quarters, because the direction tells you more than the level.

What drives recovery more than the vendor?

Timing. A collection booked before the last working day recovers far more often than one started after the person has gone. The single biggest improvement most teams can make is moving the trigger earlier, and it costs nothing to do.

Should recovery rate be in the contract?

Only with the definition written alongside it, including the denominator and the time window. A committed percentage with no definition is unenforceable, because the vendor can meet it by changing what counts. Define first, commit second.

What is a better measure than recovery rate?

Cost per recovered device, and days from exit to verified receipt. The first stops a cheap vendor looking good while recovering less, and the second catches the slow drift that a monthly percentage hides. Together they describe performance better than a rate alone.