Three in four laptop returns were avoidable: what a device fault really costs before anyone diagnoses it

Somewhere in your organization this week, a laptop is going into a box. The employee has been complaining for a month (the machine stutters, or the Wi-Fi crawls, or Microsoft Teams freezes whenever anything else is happening on the line), and after two or three attempts to fix it over a remote session, somebody on your desk has made the sensible call. Ship a replacement, bring the old one back, reimage it or send it to the depot, and move on. Nobody on the desk did anything wrong, and nobody at the depot will either, which is the part of this story worth spending some time on.

This piece is for whoever owns end-user computing, and for the CFO who signs the refresh every four years, because the cost of that box is almost never recorded as a cost of the original fault. The ticket closes as "device replaced." The shipping lands in a logistics budget, the engineer's time in the desktop support budget, the employee's lost days nowhere at all, and the laptop's shortened life shows up three years later as a line in a refresh plan that nobody connects back to a Microsoft Teams call that kept freezing. Add those pieces up (the arithmetic is not complicated) and the swap turns out to be the single most expensive thing your desk does, and the thing it does with the least evidence.

What has changed, and the reason to write this now rather than two years ago, is that it has become possible to find out what is actually wrong with a specific machine before it goes in the box.

How many devices go back, and how many needed to?

The best public number comes from the Forrester Total Economic Impact study commissioned by Lakeside Software in 2022. Forrester built a composite organization of 40,000 devices from its customer interviews, and in that composite, 1.5 percent of devices were returned each year for repair or reimaging, 600 laptops, of which 75 percent were judged avoidable once the organization could see what was going on inside the machine. Three in four. Forrester's own cost assumptions for each return were $100 in round-trip shipping, two and a half hours of an engineer's time and four hours of the employee's.

On the consumer side the picture is more extreme, with a caveat. Accenture's reverse-logistics work, reported by RCR Wireless in 2017 alongside FutureDial's returned-device data, found that 68 percent of consumer electronics returns tested as "no trouble found" and that only 4 to 5 percent of reported malfunctions had a hardware cause at all. TechSee's 2020 survey of no-fault-found returns tells the same story from the support side. The caveat is that nobody publishes an enterprise-laptop no-fault-found rate, and corporate fleets should not be assumed to match consumer returns (the complaints are different, and so is the person triaging them). But between Forrester's 75 percent avoidable and the consumer industry's 68 percent no-fault-found, it is fair to say that most laptops that go back did not need to.

What does one return actually cost?

Take Forrester's assumptions and put a conservative loaded rate on them, say $75 an hour for the engineer and $60 for the employee. One return is then $100 of shipping, plus 2.5 × $75 = $187.50 of engineer time, plus 4 × $60 = $240 of employee time, which comes to roughly $530 before anyone has looked at the laptop. If the device goes to a depot rather than being reimaged in-house, Premier's depot services literature puts an onsite dispatch at around $1,000, and a depot cycle somewhere below that depending on contract.

The number to treat with more care is the employee's time, because four hours is almost certainly too low for anyone who works away from an office. AdvisoryMSP's guide for small businesses estimates that an employee without a loaner is offline for one to three business days per broken laptop, rising to three to five if a repair is waiting on a part. Those are the authors' assumptions rather than a survey, cited here only to show the range, but for remote and hybrid fleets the truth sits nearer that range than Forrester's four hours.

So, for the 40,000-device composite: 600 returns × $530 = about $318,000 a year on the conservative arithmetic, of which three quarters, around $240,000, was spent on laptops that had nothing wrong with the hardware. Scale that to your own fleet by dividing by 40,000 and multiplying by your device count, then decide for yourself whether the employee-time line should be four hours or two days. If you need to put numbers like these in front of finance, here is how to answer the CFO on AI costs.

Why the swap shortens the life of the whole fleet

The return cost is the visible part. The larger cost is what the swap does to your refresh cycle, through a chain of reasoning in which nobody is behaving unreasonably.

Laptops do fail. Gartner's hardware reliability research, as reported by EDN in 2006, put notebook annual failure rates at 15 to 20 percent across each year of life, and SquareTrade's 2009 study of 30,000 laptops found 20 percent had suffered a hardware malfunction by the end of year three. Those are old figures and modern machines are better built, but the shape has not changed: a meaningful fraction of any fleet has a real hardware problem each year. That is exactly why a desk that cannot tell a real fault from a software one defaults to the swap. The swap is right often enough to feel safe.

The trouble is what a desk learns from that pattern. If a four-year-old laptop with a dirty filesystem stutters, and the desk cannot see the filesystem, the ticket reads as "old laptop, slow, replaced," and the refresh plan absorbs the lesson that four years is as long as these machines last. The Texas Legislative Budget Board's issue brief on IT replacement schedules cites the standard industry finding that PCs older than four to four and a half years cost 59 percent more to support. That figure is true, but true partly because the faults on older machines are the ones nobody diagnoses, and it is one of the quieter drivers of keeping-the-lights-on cost.

Forrester's composite is the clearest evidence that the cycle can be moved. Once the organization could see the software causes behind "slow laptop" tickets, it extended its refresh from four years to five, which meant 800 devices a year (2 percent of the fleet) were not replaced, at $1,100 each: $880,000 a year that had previously been spent replacing hardware that was working. Nexthink reported something similar from the employee side in 2022: of 332 employees with a healthy device experience score who were offered a refresh, 73 percent chose to keep the machine they had for another year. People do not want a new laptop; they want the one they have to work.

What has been missing?

It is only fair to every desk that ships the box to say that the alternative has not really existed. A remote session gives a technician twenty minutes on a machine that may not misbehave while they are watching. Experience monitoring tools give the desk a score that says the device is unhealthy, which is useful, and then stop, because the score does not say what is wrong with this laptop, this morning. And scripts fix the faults somebody already wrote a script for. Faced with a complaint that has survived all three, the swap is the only move left on the board, and it has the advantage of closing the ticket.

What we have built at Rezolve.ai with Rapport is an agent on the device that investigates the way your best engineer would if they had an afternoon. It forms a hypothesis, runs a check, reads the result and decides what to look at next, without the employee stopping work. Two cases from our own pilot show what the swap would have missed.

The first was a desktop that "stutters and locks up." Memory, crash history, GPU resets, hardware error events and disk latency all came back clean, five results that would have justified a replacement on any desk. Then the agent looked at something nobody runs on a stutter complaint: the NTFS volume was flagged dirty, and Windows had been asking for a full disk check at every boot for a month (45 times in 30 days) without being able to repair it while running. The fix was a scheduled check and a reboot, not a new machine.

The second was a home-office PC with "painfully slow Wi-Fi" while a Mac on the same network was fine. The agent measured both (105 Mbps against 19), found the PC stuck on 2.4 GHz, scanned from the PC and found that it could not see the router's 5 GHz radio at all, with no networks on channels 52 to 144. It traced that gap to a 2019-era wireless card whose region settings block those radar-shared channels, switched on the card's country-detection setting with approval, rescanned to prove the limit was in the hardware, and recommended a $30 replacement card rather than a $1,100 laptop or a call to the ISP.

Neither of those is a hardware failure in the sense the refresh plan means, and both would have gone in the box.

Where this leaves your refresh plan

If you are the person who signs the refresh, the practical question is not whether to move from four years to five. It is whether you can see the faults on your oldest machines well enough to know. Start with your own return numbers: how many devices went back last year, what fraction came back from the depot with no fault found (ask the depot; they know), and what the ticket said before the swap. If the answer to the last question is "slow" or "unable to replicate," you are paying the $530 and shortening the fleet's life on faults nobody looked at.

Then put an investigator on the machines that are about to be replaced. A diagnosis before the box costs minutes of agent time and nothing of the employee's; a swap costs a day of theirs, half a day of yours and a year off the laptop's life. The laptop going into the box this week may well have a real hardware fault, and if it does, the investigation will say so. But for the three in four that do not, you will have found out before the shipping label was printed rather than after. The case for rethinking DEX starts here.