A damaged file system and a surface full of dying sectors produce the same messages, the same freezes and the same sinking feeling. They need opposite responses, and the tool most people reach for first is safe on one and destructive on the other. That is what the free diagnostic is for.
Nobody is quoted before the cause is known. The free diagnostic separates a damaged file system from a failing surface, then any one drive or SSD is £300 + VAT, fixed in writing.
Windows owns exactly one word for a large family of faults, and it hands that word out just as readily to a drive with a scrambled directory as to a drive with ten thousand failing sectors. Nearly every mistake made on this sort of job comes from acting before that difference has been established.
The most quoted message in the trade and the least informative. It appears when the index entry for a file cannot be parsed, which happens both when the index itself is damaged and when the sectors holding the index simply will not read. Identical wording, opposite causes, opposite correct responses.
A transfer runs at a sensible speed for a while, drops to a few kilobytes a second and eventually locks the whole machine up. That timing pattern is the signature of a physical fault rather than a logical one. The operating system is sitting on a failing sector waiting out a retry cycle that was never designed for this.
A directory that opens instantly and shows nothing at all, while the free space on the drive says the files are still there, is usually index damage. The file records themselves are frequently untouched and every one of them can be walked out of the master file table on a copy of the disk.
It passes a scan on Monday and throws errors on Wednesday, and both results are honest. Sectors go from readable, to readable after several attempts, to gone, and a drive quietly reallocating in the background looks healthy right up to the point it runs out of spares to reallocate into.
The partition is still listed, the capacity is right, and the file system where the boot record should be is simply not recognised. On a sound drive that is a handful of damaged sectors in one very unlucky spot. On a failing drive it is the first visible symptom of a surface that has been degrading for months.
Somebody ran the built-in repair tool, it worked for several hours, and afterwards the folders were gone and a directory of numbered files had appeared. That tool writes, and when it cannot verify a structure it resolves the doubt by deleting. This is the single most common avoidable cause of loss on this page.
Health reporting is a starting point rather than an answer. Reallocated sector counts, pending sectors, uncorrectable errors and raw read error rates are all worth reading, and a drive with hundreds of pending sectors has told you what it is. The trouble is the opposite case: plenty of drives with genuine surface damage report themselves as healthy, because the attributes only update when the drive itself notices, and a drive that has stopped being able to read an area reliably is not always honest about it.
What settles it is a read pass across the whole capacity with the timing of every block recorded. On a drive whose fault is logical, that pass comes back flat: every region reads at full speed and the damage is confined to the few megabytes where the metadata lives. On a drive whose fault is physical, the graph is unmistakable. Failures gather into ranges. Response times climb from milliseconds into whole seconds before an area gives up entirely. On a multi-head drive, damage restricted to one head shows as regular bands of trouble at fixed intervals across the disk, because consecutive sectors are spread over the surfaces in a repeating pattern.
That pass is what the free diagnostic is, and it explains the timing. Part of it runs at whatever speed the drive will tolerate, and a drive that has to be coaxed cannot be hurried without doing damage. This is why the diagnostic closes 2 working days after arrival and never sooner, and why nobody here will offer an opinion on the telephone about which of the two you have. The answer arrives in writing with a fixed figure attached to it.
If it is logical, the drive is copied for safety and the work happens on the copy: the master file table is parsed and rebuilt from its mirror, the journal is read to see what was in flight when things went wrong, orphaned records are reattached to their parents and the directory tree comes back with names and structure intact. Any one drive or SSD is £300 + VAT and the job keeps its guarantee in full.
If it is physical, the priority changes completely. The drive goes onto an imager, the healthy regions are taken first at speed, and the damaged ranges are approached afterwards from several directions with the retry behaviour tuned to the fault. Only when as much of the surface as can be read has been captured does anybody look at the file system. The band is the same £300 + VAT, but the terms are not: electronic and mechanical faults take 50% of the figure upfront, in the network wording, no fix, no fee all jobs except electronic and mechanical failures, chip level work, DVR and Forensic jobs.
The check-and-repair tool built into Windows is a good piece of engineering aimed at a problem you probably do not have. It exists to make a file system internally consistent again, typically after a power cut, and it does that on the assumption that the hardware underneath it is sound. Both halves of that sentence matter. It writes its corrections into the metadata regions, and on a drive with pending sectors the metadata regions are exactly where a write is most likely to fail and leave things worse than it found them.
Then there is what it does when it finds something it cannot verify. It does not preserve the ambiguity for later; it removes it. Index entries that do not check out are dropped. Cross-linked clusters are cut apart and one side is discarded. File records with no valid parent directory get written into a folder called FOUND.000 as numbered fragments with no names, no dates and no folder structure, which is the state a great many drives are in when they arrive here. The data may still be present, and the map that told you what any of it was has been thrown away.
None of that means the tool is useless. On a healthy drive with a genuinely inconsistent file system it often works, and it is free. The difficulty is that you cannot know which situation you are in without the read pass described above, and running it on the wrong one is not recoverable. So the order is simple: stop writing to the drive, do not install recovery software onto it, take it out of the machine and let somebody establish what is wrong before anything tries to put it right.
Where the file system can be rebuilt, you get your folders and your file names back as they were, which is what most people mean by a successful recovery. Where the metadata is genuinely gone, the fallback is carving: the image is searched for the recognisable headers that documents, photographs, video and databases begin with, and files are lifted out by shape. That works remarkably well for photographs and office documents and it cannot invent what it never had, so those files come back numbered rather than named and out of any folder structure. It is worth knowing in advance which of the two a job is heading for, and the diagnostic says so.
Corruption cases come in from the office estates along the A329M and from the technology corridor towards Slough and Reading, from home workers in Wokingham and Crowthorne, and from London commuters whose laptop was fine until it was closed on a train and never came back properly. Unbolt the drive rather than posting the machine, wrap it so nothing rattles, and send it tracked to Guildford Data Recovery, Building 2, Ground Floor, Guildford Business Park, Guildford GU2 8XH, roughly forty minutes down the A322 and the A3 or next working day by post. See also hard drive recovery, recover lost files, formatted drive recovery and the published bands on data recovery cost.
Stop writing to the drive, leave the repair tools alone, and let a timed read pass answer the question properly.
Almost every job on this bench arrived as a parcel. Tracked, insured post is the gentlest way to move storage that is already in trouble, and a box handed in around Bracknell, Wokingham or Ascot is normally on the Guildford bench the next working day.
Is the storage still bolted into a machine — laptop, tower, iMac, MacBook, rack server, a DVR under the till? Free it first and send the bare unit. Stripping hardware is not something this lab does, though it is ten minutes' work for any repair shop on your high street. There is a single case with no way round it: memory chips soldered flat onto the mainboard, which is how Apple Silicon machines and certain ultrabooks are built. Where the storage cannot be unbolted, there is no parcel to make up.
↓ Print the shipping & booking-in form (PDF)
The name on the parcel wants to be Guildford Data Recovery. Driving it over from Bracknell is roughly forty minutes on the A322 then the A3; posting it costs you a stamp and a day. Either way, a message goes out to you as soon as it is logged onto the system, and two working days later the diagnostic is finished.
Unsure whether something should go in the box? Ring 0800 689 0668 while the lid is still open, or work through the free online diagnostic and let it tell you.