Asset register accuracy is almost never measured, which is why almost every register is worse than its owner believes. The failure is predictable: records are created once, correctly, and then reality moves and nothing tells the register about it.
TL;DR
- Measure before you fix. Pull ten random records and verify each against reality.
- More than two wrong out of ten and the register is not usable for decisions.
- Four failures cause almost all of it: silent reassignment, unreturned leaver kit, purchases outside the process, and stale last-seen data shown as current.
- Each failure has a different fix, and a new tool addresses only one of them.
- Report two numbers, never one: devices known, and devices confirmed recently.
- Monthly reconciliation against your directory catches what no scan can.
Measuring asset register accuracy in an afternoon
What to do: take ten records at random, not ten you choose, and establish the truth of each by contacting the named holder or checking a shipment record. Then score them.
| Records wrong out of ten | Verdict | Action |
|---|---|---|
| 0 to 1 | Trustworthy | Keep the process, measure quarterly |
| 2 to 3 | Usable with caution | Fix the specific failure you found |
| 4 to 6 | Not usable for decisions | Change how records are created |
| 7 plus | Fiction | Rebuild from procurement records |
Record which kind of wrong each one was, because the pattern matters more than the score. Four wrong for four different reasons is a process problem that needs a change to how records are created. Four wrong all because people left is one fixable gap and a far cheaper piece of work. Teams who skip this distinction tend to buy a general-purpose tool for a specific failure, then find asset register accuracy no better a year later.
The four failures, and the fix for each
Silent reassignment. A laptop moves from one person to another without anybody updating the record. Extremely common and almost never logged, because the handover happens between two colleagues who have no reason to tell IT. The fix is to make reassignment an event somebody has to perform, usually by requiring a wipe between holders.
Unreturned leaver equipment. Somebody left, the laptop did not come back, and the record still shows them as the holder. The fix is reconciliation against your directory, monthly, because no scan can distinguish a device in a drawer from one that is merely switched off.
Purchases outside the process. A manager buys a laptop on expenses for an urgent start. It never enters the register because nothing in the purchase route touches it. The fix is partly policy and partly making the sanctioned route faster than the unsanctioned one.
Stale data presented as current. The record exists and was accurate eight months ago. Nothing marks it as old, so it is read as fact. The fix is a last-verified date on every record and a rule that anything beyond thirty days is flagged rather than listed normally.
Why a better tool fixes only one of them
This is the part worth understanding before any purchase, because it determines whether buying something helps.
A discovery tool addresses stale data, for devices it can reach. It does nothing about reassignment, nothing about leavers, and nothing about purchases outside the process, because none of those produce a network signal.
A manual register with better fields addresses none of the four. It is the same discipline in a tidier interface.
A lifecycle platform addresses reassignment and leaver returns, because both are events it performs. It is blind to purchases made outside it, which is the one failure no system solves on its own.
So the honest sequence is: measure, identify which of the four dominates, then decide whether that one is a tooling problem or a process problem. Most teams who skip the measurement buy for stale data and still have a register that is wrong about people.
Reporting it honestly
One number invites the wrong conclusion. Two numbers tell the truth and take no longer to produce.
Checklist:
- Devices known: every record in the register
- Devices confirmed: those verified within your chosen window
- Never quote the first without the second
- Show last-verified on every record by default
- Flag anything beyond thirty days instead of listing it as normal
- Reconcile against the directory monthly and record the differences
This matters beyond tidiness. If an auditor, an insurer or a security review asks how many devices you manage, the single number is the one that gets quoted and the one most likely to be wrong by a third.
What the options publish
Read from each vendor’s own page on 16 October 2026, with script and style blocks stripped first.
| Option | Published pricing | Failures it addresses |
|---|---|---|
| AssetTiger | $15 a month unlimited items, $20 a month to 500 assets, then $40, $75, $140 and $275 | None directly; a tidier register |
| Snipe-IT | Free self-hosted; hosting from $399.99 a year | None directly; a tidier register |
| Lansweeper | No figure published; free tier available | Stale data, on reachable devices |
| Device42 | No figure published | Stale data, plus relationships |
| RemoAsset | No figure published, quote-based with a demo required | Reassignment and leaver returns |
RemoAsset. Relevant to two of the four failures because reassignment and retrieval are actions it carries out, so the record updates as a consequence rather than as an afterthought. Disclosure that should shape how you read this: RemoAsset is owned by the same people who publish this site. Where it struggles, stated plainly: pricing is quote-based and requires a demo, so it cannot be compared against the published ladders without a call; and it is a newer platform with a smaller footprint than the legacy global resellers. It is blind to anything bought outside it, which is precisely the failure no platform fixes, and it performs no discovery.
Verify the figures against the source: AssetTiger and Snipe-IT both publish openly.
Final thoughts
The ten-record audit costs an afternoon and changes most plans. Teams who run it generally discover their register is wrong in one specific way rather than generally unreliable, which turns a vague procurement exercise into a narrow fix.
And whichever way you go, report two numbers. A register that admits what it does not know is more useful than one that presents eight-month-old guesses with the same confidence as this morning’s facts.
How do we measure asset register accuracy?
Pull ten records at random rather than ten you pick, and establish the truth of each by contacting the named holder or checking a shipment record. More than two wrong out of ten and the register should not be informing decisions. Record what kind of wrong each one was, because the pattern tells you which single failure to fix and that matters more than the score.
What makes registers go wrong?
Four things, and almost all inaccuracy traces to them: a laptop reassigned between colleagues without anybody updating the record, leaver equipment that never came back, purchases made outside the sanctioned route, and old records displayed as though current. Each has a different fix, which is why a general sense that the register is unreliable rarely leads to the right purchase.
Will a new tool fix it?
It will fix at most one of the four failures. Discovery addresses stale data on reachable devices only. A tidier manual register addresses none of them, since it is the same discipline with a better interface. A lifecycle platform addresses reassignment and returns because it performs both. Nothing addresses purchases made outside the process, which is a policy problem.
Why report two numbers?
Because a single count of known devices invites the reader to treat every record as current, and some will be months old. Reporting devices known alongside devices confirmed recently takes no extra effort and prevents the specific embarrassment of telling an auditor or insurer a figure that is overstated by a third. Quote them together, every time.
How often should we reconcile against people?
Monthly, because the failure it catches is time-sensitive. An unreturned laptop is far more recoverable in week three than in month nine, and no scanning tool can surface it, since a device in a drawer is indistinguishable from one that is switched off. Compare the register against your directory, chase every difference, and write down what you found.
What about devices bought on expenses?
That is the failure no system solves, because nothing in the purchase touches your process. Partly it needs a policy, and partly it needs the sanctioned route to be faster than the unsanctioned one, since managers buy on expenses when the official path cannot deliver before somebody starts. Fixing the speed usually does more than restating the rule.
