An SSD does not degrade the way a hard drive does. There is no noise to listen for and no gradual slowdown to notice — the useful signals are all in the numbers, and only a few of them mean anything.
Percentage used, reallocated or reserved blocks, and uncorrectable errors. Those three tell you nearly everything; the remaining dozens of attributes are mostly noise for a normal user.
On Windows, the built-in options are thin: wmic diskdrive get status gives a crude pass/fail, and Optimise-Drives confirms TRIM is running. For real numbers use CrystalDiskInfo, which reads the SMART attributes directly, or the manufacturer’s own tool — Samsung Magician, Crucial Storage Executive, WD Dashboard — which interpret vendor-specific values correctly where generic tools guess.
On macOS, System Information shows basic SMART status, but for anything meaningful you need smartmontools via the terminal. Apple’s own reporting is deliberately minimal, and on soldered Apple Silicon storage there is very little exposed at all.
Percentage used (or Media Wearout Indicator) estimates how much of the drive’s rated write endurance has been consumed. Under 80% is unremarkable. Above that, plan a replacement — though note this is a wear estimate, not a prediction of failure, and drives frequently outlive it.
Reallocated or reserved block count shows how much spare capacity has been consumed retiring bad blocks. A slow rise over years is normal; a sudden jump is not, and is worth acting on immediately.
Uncorrectable error count is the one that matters most. It means the drive found data its error correction could not repair — that is real data loss, however small, and any non-zero value climbing over time is a drive to replace now rather than monitor.
SMART on an SSD reports flash wear well and controller health barely at all. The commonest catastrophic SSD failure is the controller or its firmware, and that gives essentially no advance warning — the drive is healthy on Monday and undetected on Tuesday. A perfect SMART report does not mean a drive is safe; it means the flash has not worn out.
Which is why the practical advice for SSDs differs from hard drives: with a spinning disk you can often listen for trouble and act. With an SSD you cannot, so the backup has to exist beforehand rather than being arranged once symptoms appear.
Where the controller has failed but the NAND is intact, the drive can often be reached in vendor mode and imaged — that is the recoverable case, and it is common. Where the flash itself has degraded, or where files were deleted from a working drive and TRIM has erased them, there may be nothing left to read. SSD and NVMe recovery is from £300 +VAT after a free 48-hour diagnostic that establishes honestly which situation you are in.
Read its SMART data with CrystalDiskInfo on Windows or smartmontools on macOS, or use the manufacturer’s own utility, which interprets vendor-specific values correctly. Focus on three numbers: percentage used, reallocated or reserved blocks, and uncorrectable errors.
It estimates how much of the drive’s rated write endurance has been consumed. Below 80% is unremarkable; above that, plan a replacement. It is a wear estimate rather than a failure prediction, and plenty of drives run well past it.
A sudden rise in reallocated or reserved block count, and any non-zero uncorrectable error count that climbs over time. The second is the serious one — it means data was found that error correction could not repair, which is real loss however small.
Only partially. It reports flash wear reasonably well and controller health barely at all — and controller failure is the commonest catastrophic SSD fault, arriving with no warning. A clean SMART report means the flash hasn’t worn out, not that the drive is safe.
Often, yes. Where the controller has failed but the NAND is intact the drive can be reached in vendor mode and imaged read-only. Where the flash itself has degraded, or TRIM has already erased deleted files, there may be nothing to read. A free 48-hour diagnostic establishes which.