TL;DR
A device budget per employee is usually the price of a laptop divided by a refresh cycle, and that number is wrong by a wide margin. It leaves out accessories, shipping, the machines that never come back, repairs, storage and disposal. Build it from six lines instead and the figure stops surprising finance halfway through the year.
- The laptop is roughly half of what a person’s equipment actually costs.
- Six lines belong in the number: device, accessories, logistics, loss, repair, disposal.
- Divide by the real service life, which is shorter than your refresh policy.
- Count the devices that never came back. They are a purchase, not an anomaly.
- Build it per role band, not as one company average.
- Recalculate when headcount or geography changes, not annually.
The six lines in a device budget per employee
Most budgets have one line. Here are the six that actually occur.
- The device. The only line most people count.
- Accessories. Monitor, dock, keyboard, mouse, headset. Collectively substantial and usually approved one at a time, which is why nobody adds them up.
- Logistics. Shipping to a home address, import duty where it applies, and the second delivery attempt when nobody was in.
- Loss. Devices that never came back from leavers, plus damage beyond repair. This is a real recurring cost presented as an occasional accident.
- Repair. Out of warranty work, screen replacements, battery swaps.
- Disposal. Collection, wiping and certified destruction at end of life.
Add them, divide by the service life, and you have a number per person per year that will survive contact with finance.
Best tools for HR Operations
Why the single line version fails
Because the missing five do not disappear. They appear as unbudgeted costs during the year, each small enough to approve without question and large enough in aggregate to overrun.
That pattern is what makes equipment look unpredictable when it is actually quite stable. You were forecasting one line of a six line cost.
Divide by the real service life, not the policy
Everyone divides by the refresh cycle in the policy. Almost nobody checks whether machines actually last that long.
They usually do not. Devices leave early because somebody left and kept one, because a screen broke and replacement was simpler than repair, or because a returned machine sat in a cupboard unwiped until it was too old to reissue.
How to measure it
Take every device retired in the last twelve months. Record the months between issue and retirement. Take the median, not the mean, because a few very old machines will flatter the average.
Most teams find the median is meaningfully shorter than the stated policy. That gap is the single biggest driver of the annual figure, and it is measurable from your own records in an afternoon.
It is also the most improvable line. Every extra month of service life spreads the device cost further, and unlike hardware prices it is entirely within your control.
Where depreciation is being modelled formally, the useful life assumption under IAS 16 on property, plant and equipment should match what you actually observe, and IAS 36 on impairment of assets covers what happens when an asset is impaired earlier than planned. Finance teams often carry an assumption nobody has tested against the register.
Capturing the accessories line, which nobody adds up
Accessories are the second largest line in most of these budgets and the hardest to see, because they arrive one request at a time and each one is small enough to approve without thought.
The result is that nobody has ever seen the total. A monitor here, a dock when somebody complains about cables, a headset before a customer call, a keyboard after a wrist problem.
How to get the number
Do not try to reconstruct history. Take one standard setup and price every item you would ship to a new joiner by default, then multiply by expected joiners for the year. That is the forward number, and it is the one you need.
Then add a replacement rate for the existing population. Accessories fail and get lost at a lower rate than laptops but across a larger item count, so the figure is rarely negligible.
Two refinements make it more accurate. Accessories usually outlive the laptop, often across two refresh cycles, so divide them by a longer life than the device. And they rarely come back from leavers, which means the replacement rate for accessories is closer to the joiner rate than to a failure rate.
That second point surprises people. If a monitor almost never returns, every joiner needs a new one regardless of how many monitors the company has already bought.
Build it per role, not as one average
Disclosure: RemoAsset, named below, is owned by the same people who publish HROpsLab. It is described with its limitations stated, on the same terms as every other supplier. This site takes no vendor money and sells no placements.
A single company-wide figure hides the thing you most need to see.
A support team on standard laptops and a video team on heavy machines with colour-accurate monitors do not average usefully. The blended number is wrong for both and gives you no lever.
Three bands is normally enough. Standard, elevated and specialist. Each gets its own six line calculation and its own service life, because heavier machines are often kept longer and specialist accessories often outlive two devices.
What this reveals
Usually that one small group accounts for a disproportionate share of the budget, and that the large standard population is cheaper per head than anybody assumed.
That changes conversations. Arguing about the standard laptop specification saves little. Deciding whether the specialist band genuinely needs annual replacement saves a lot.
Where devices are bought, delivered, retrieved and reissued through a single supplier, the per band figures are easier to assemble because the lines arrive in one place rather than four. That is part of what running the lifecycle through one provider gives you. It does not make the equipment cheaper by itself, and a managed service carries its own fee, so compare the total rather than the unit.
The decision table
| Situation | Scale | Setup | Primary pain | Recommended starting point |
|---|---|---|---|---|
| No budget line at all | Under 50 | Ad hoc purchases | Every order is a surprise | Build the six lines once, roughly |
| Budget overruns mid year | 50 to 200 | Laptop price only | Five lines missing | Add accessories and loss first |
| Finance questions the figure | Any | One blended average | Not defensible | Split into three role bands |
| Devices leaving early | Any | Policy says three years | Real life is shorter | Measure the median, then requote |
| New market opening | Distributed | Domestic assumptions | Logistics underestimated | Recalculate logistics per country |
Presenting a number finance will accept
Two things make the difference between a figure that gets approved and one that gets argued about.
Show the lines. A single number invites haggling over the number. Six lines invite a conversation about which line to attack, which is the conversation you actually want.
Show the measurement. Service life taken from your own retirement records is far harder to dismiss than a benchmark from an industry report, and it is specific to you.
Avoid quoting a market average. Somebody will find a different one, and you will spend the meeting defending somebody else’s methodology instead of discussing your own numbers.
What to do with the result
Pick the one line you can most improve and work on it for a year. For most teams that is loss, because devices not coming back from leavers is both the most avoidable line and the one nobody owns.
Second is usually service life, through better retrieval and a wipe process somebody trusts enough to allow reissue. Where that process is missing, machines that could serve another two years get replaced instead, which is the expensive version of a storage problem. NIST Special Publication 800-88 on media sanitization is the standard most teams point at when defining what a trusted wipe means.
For the adjacent question of how to value what you already hold, see our guide to equipment insurance valuation.
See how the lines arrive in one place with lifecycle management across procurement and retrieval, or read our guide to equipment insurance valuation.
Frequently asked questions
What should a device budget per employee include?
Six lines, not one. The device itself, accessories such as monitor, dock and headset, logistics including shipping and any duty, loss covering devices that never came back and damage beyond repair, repair work outside warranty, and disposal at end of life. The laptop alone is roughly half the real figure, which is why single line budgets overrun during the year rather than at the start.
How do we work out the real service life?
Take every device retired in the last twelve months and record the months between issue and retirement, then take the median rather than the mean so a few very old machines do not flatter it. Most teams find the median is well short of their stated refresh policy. That gap is the largest single driver of the annual number and you can measure it from your own records in an afternoon.
Should we use one average or several?
Several, normally three bands: standard, elevated and specialist. A blended company average hides the thing you need to see, because a support team on standard laptops and a video team on heavy machines with colour accurate monitors do not average usefully. Splitting them usually shows that one small group drives a disproportionate share while the large standard population is cheaper per head than assumed.
Is loss really a budget line?
Yes, and treating it as an occasional accident is why it never gets managed. Devices that never come back from leavers recur at a fairly stable rate once you start counting, which makes them a forecastable cost rather than a surprise. It is also usually the most improvable line, because the fix is a retrieval process rather than a cheaper laptop.
Should we quote an industry benchmark to finance?
Avoid it. Somebody will find a different benchmark and the meeting becomes a debate about whose methodology is better, rather than about your numbers. A figure built from your own retirement records and your own six lines is specific, harder to dismiss, and leads to a conversation about which line to attack, which is the conversation worth having.
Which line should we try to improve first?
Loss, for most teams, because devices not returning from leavers is both the most avoidable cost and the one nobody owns. Service life is usually second, improved through better retrieval and a wipe process trusted enough that returned machines get reissued. Where that process is missing, machines with two good years left get replaced instead, which is an expensive version of a storage problem.
How often should the number be rebuilt?
When headcount changes by half, when you open a new market, or when the mix of roles shifts materially. Annual recalculation tends to happen in years when nothing changed and gets skipped in the years when everything did. Logistics in particular needs redoing per country rather than carried over, because domestic assumptions travel badly.
