A finished dissertation, a quarter of payroll, the only copy of the books, and every bit of it resting on a small plastic object that caught the edge of a desk or stopped halfway through a transfer. The size suits the geography, at least: an envelope posted in Coventry, Nuneaton or Leamington is on the bench the following morning. Snapped necks, silent controllers and folders deleted by accident all have a route back, and the first step in nearly all of them is resisting the urge to try one more port.
Looking at a USB stick costs nothing. What comes out of that examination is one written figure, agreed between us before anybody picks up a screwdriver: £250 + VAT on a USB stick, a pen drive or a memory card.
Logical work is no fix, no fee. Four things sit outside that: electronic and mechanical failures, chip-level work, DVR jobs and forensic work, and anything physical is half up front. All five bands are laid out on the data recovery cost page, while data recovery services follows a postal job from the parcel to the answer.
Tracing a symptom back to whatever caused it is where the job genuinely starts, and these thirty cover very nearly every box opened on the Oxford bench. A fault that is not on the list is not an unfamiliar one: describe it over the telephone and you will get a straight reading of the odds before you have spent a penny on postage.
A laptop lifted with a stick still in the socket puts every bit of the leverage on four or five small solder joints. They crack, and after that the stick works sometimes, or under pressure, or never. This is the most common physical fault on the page and one of the most hopeful, because the memory itself has not been touched and the entire job is restoring a stable connection for one clean read.
No sound, no light, no drive letter, nothing in Disk Management. Try a second port and a second computer before concluding anything, because front-panel sockets and unpowered hubs account for a fair share of dead sticks that are not dead. A stick that is genuinely silent everywhere has usually lost its controller rather than its memory.
The classic sign of a cracked joint between the connector and the board. Every insertion flexes it a little further, and the short that finally kills the stick almost always happens during one of those hopeful last attempts. If a stick has to be propped up or pressed sideways to be seen, that is the moment to stop rather than the moment to hurry.
Windows assigns a letter and then reports no media, no size, or asks for a disk to be inserted. The controller is answering the initial handshake and failing everything after it, which usually means its own firmware or its mapping tables have become unreadable. The memory below is generally intact and reachable another way.
Power is reaching the board and the controller is attempting to start. A stick in this condition is often a firmware problem rather than a hardware one, and the production diagnostic path for its controller family frequently brings it straight back with the mapping intact, which is much the best outcome available.
Heat is current going somewhere it should not, and a stick that gets noticeably warm within seconds is drawing that current through a fault. Every further insertion is a chance for it to take the memory packages with it. Leave it out of the machine from that point on; there is nothing to be gained by one more attempt.
Sat on, closed in a car door, or caught in a bag zip. Whether anything survives depends entirely on the memory package: if it is intact it can be lifted off and read directly, and if it has been cracked through, there is nothing left to read. Send every fragment, including the cap and the loose plastic, because the pieces are worth having.
A crushed case tells you very little about the board underneath, and less still about the chip on it. These arrive looking hopeless and come back complete more often than anybody expects. What does finish them is somebody straightening the case and plugging it in to see whether it still works.
Flash survives water surprisingly well, provided nobody powers it while it is damp and nobody applies heat. Corrosion afterwards is the real enemy, which makes the delay between the wash and the bench the thing that matters most. Do not dry it on a radiator and do not test it to see whether it survived.
Counterfeit sticks are small sticks with the firmware told to report a size they do not possess. They behave normally until the writing wraps around, at which point the earliest files are destroyed rather than merely mislaid. Nothing recovers what has been written over. If the capacity looked too good for the money, that is very likely the story.
Flash cells tolerate a finite number of erase cycles, and a stick that has been the working copy of somebody's documents for six years has genuinely done the miles. The symptoms are files that quietly stop opening and folders that take a long time to list. It is the one flash failure that gives warning, and the warning is usually ignored.
A stick spending its time on error correction and retries behaves like this. It is still readable while it is doing it, which makes today the right day to copy it and next month the wrong one. Do not defragment it, do not run a disk check on it, and do not let anything write to it while you are getting the files off.
Much the best position to be in. Sticks are generally not given the same background erase instruction that internal drives receive, so deleted material frequently sits in place until something writes over it. Keeping the stick out of every machine from the moment you notice is what preserves that, and it costs nothing.
The application reported the file as written before the stick had finished storing it. What is left is a partial file, sometimes with the original still present underneath in the space the new version was meant to occupy. Both are worth looking for, and both are found by reading the stick as an image rather than by opening the file again and hoping.
Either a damaged allocation table or the aftermath of an infected machine, and telling those apart matters because one is repaired and the other has to be handled on an isolated bench first. Anything arriving with an autorun infection is worked on a machine with no network connection, and the files come back clean.
No switch, no policy, and the stick has decided it will accept no more writes. Controllers do this when they run out of spare blocks to shuffle into. It is a terminal state rather than a setting, and it means the same thing every time: copy everything off now, in one pass, because the next stage after read-only is silence.
Sticks with a keypad on the case or an on-board encryption chip protect the data properly, which cuts both ways. Without the password or the administrative key the memory can be read and the result is noise. This is a genuine limit and it is better said here than discovered after postage. Some models have a recoverable key held in the controller, and the diagnostic establishes which is which at no charge.
Installer sticks and rescue sticks live hard lives, get rewritten repeatedly and are frequently the same stick somebody also keeps documents on. A stick that has lost its boot structures has usually not lost the files sitting alongside them, and the two problems are separate: the boot part is disposable and the documents are not.
Somebody changed the file system so that both machines could use it, without appreciating that the change discards the index rather than converting it. If nothing has been written since, this is a good outcome. If a fortnight of work has gone on afterwards, it is a poor one. Stop using it either way.
Often a stick formatted by a device rather than by a computer, or one carrying a proprietary layout that only its own software understands. It is not broken; it is speaking a dialect. Say which machine wrote it when you send it, because that single line frequently turns a puzzle into a ten-minute job.
The stick takes the full weight of the machine against the desk, or the socket takes it, and both the plug and the port lose. The board damage is usually at the joints rather than in the memory, and the repair is at the microscope. The port on the laptop is somebody else's problem.
Industrial sticks hold programs, calibration files and job records, and they are often years old, formatted oddly, and the only copy in the building. They come in from workshops around Coventry, Rugby and the Warwickshire trading estates. The recovery is ordinary; the useful part is describing what wrote it so the layout can be interpreted properly.
Flash leaks charge over time. A stick put away with the photographs on it and forgotten for five years can come back with regions that no longer read reliably. Low-level reading with adjusted thresholds recovers much of it. It also explains why the answer to where should this live is never a single stick in a drawer.
Not fatal, and worth declaring. Two things do harm: software installed onto or saving its output onto the stick it is reading, and hours of continuous scanning on a stick with a cracked joint or a failing controller. Neither ends the job. Both change what the bench does first, and the booking form is the right place to mention it.
On the larger sticks the data is spread across more than one package, so a single failure does not lose a tidy fraction of the files; it damages a part of nearly all of them. Where the controller kept a parity scheme, the missing share can often be rebuilt, and where it did not, what comes back is partial and is described honestly as partial.
The very small and very thin sticks are frequently a single moulded assembly, controller and memory together, with no separate package to lift. Access means grinding down to the test points inside the package and wiring onto them under magnification. It succeeds often enough to be worth trying, it is chip-level work at 50% up front, and it sits outside no fix, no fee.
The identical pair from the same multipack, one of which has just been prepared for a fresh backup. What survives depends on how much was written over the top and on whether the wipe was a quick format or a full one. Send both, labelled, and stop using either.
Some sticks tolerate that for years and some are corrupted by it the first time. Pulling a stick without ejecting leaves the file system half updated, and the visible result is a stick that asks to be formatted or one that lists folders it cannot open. The contents are normally intact behind the damage.
The cheap branded stick handed out at an exhibition, used because it was to hand, and now holding the only version of a tender, a set of accounts or a dissertation. Build quality on these is poor and the failure rate is high. The recovery is the same as for any other stick and the price is the same 250 pounds + VAT.
No drop, no water, no warning. Sticks have no moving parts and consequently no way of telling you they are struggling, so the transition from working to not working is instant. That is not a reason to assume the worst: silence points at the controller far more often than at the memory, and the memory is where your files are.
Worth sorting out before anything is posted, because two prices are involved. The first object is an ordinary thumb drive or pen drive: a controller and a memory package on a small board inside a plastic shell. That is £250 + VAT. The second is a memory card living inside a reader or an adapter, which is a card job at the same £250 + VAT — take the card out of the adapter and send the card. The third is a portable solid state drive in its own case, a Samsung T-series or a SanDisk Extreme or one of the many similar units, and that is a drive rather than a stick, so it sits in the £300 + VAT band alongside every other single SSD. The difference matters because people assume anything with a USB plug on it is the cheapest band, and because the failures are not alike either. A portable SSD has a bridge chip between the plug and the drive, so half of those turn out to be interface faults with the storage behind them in perfect order. A thumb drive has no such layer, so when it goes quiet the controller itself has generally stopped.
The commonest arrival on this page by a distance is a stick with its connector bent over or torn away completely. A bag catches it, a knee goes into the side of a laptop, a lid closes on one that was left in. The plug takes a strain the little board behind it was never built to carry, and the result looks final and very seldom is. Where the contact pads are still on the board, it is a straightforward rework under a microscope. Where the pads have lifted, the connection is rebuilt track by track at magnification. Either way the memory package in the middle of the stick, which is where your files live, has not been touched. Two habits spoil it. The first is trying the damaged stick in a fresh port on the off chance, when every attempt threatens to lift the pads away altogether. The second is binning it, which is what a surprising number of people do. Put it in an envelope with a line explaining what happened. Rebuilt connectors are electronic work, so half is payable up front and the guarantee does not reach it, and that is said before you commit.
Sort the sticks that arrive here into two piles and you have a soldering pile and a firmware pile, which is why the bench looks the way it does. No filtered enclosure, because nothing is opened to the air, and no mechanical tooling, because nothing turns. Fine rework, controller access and the means to read a memory package that has lost its voice cover practically the whole page.
Bent and torn plugs replaced, lifted pads rebuilt, cracked tracks bridged. A stick that snapped at the plug is among the quickest jobs here and among the most likely to come back whole, provided nobody has spent a fortnight forcing it into sockets first.
Phison, Silicon Motion, Alcor, Chipsbank and Innostor controllers all have a production path. Restoring a collapsed translation table through it hands back a readable stick with its mapping intact, which is a far better result than reconstructing the mapping afterwards from raw pages.
Where the controller is beyond reviving, the memory package comes off under controlled heat, is cleaned, and is read on a programmer. Chip-level work takes 50% up front and falls outside no fix, no fee, and which category a stick is in is settled during the free diagnostic rather than at the end.
Where there is no package to remove, the moulding itself is taken down under magnification until the contacts buried in it are exposed, and each one is then wired to by hand. It is slow, delicate and never guaranteed, and on an object built in one piece there is no alternative route in.
Raw flash is pages, spare areas and scattered metadata. Turning that back into files means reproducing the page order, the interleave, the XOR scheme and the error correction the original controller used, and it is where most of the hours on a chip-level job actually go.
Every port has hardware write blocking in front of it, and anything carrying an infection is worked on a machine with no network connection. Nothing on your stick is altered at any stage, and what comes back to you comes back clean.
Whatever the badge and whatever the size, a pen drive is 250 pounds + VAT, exactly the figure a memory card attracts, and the assessment that produces that figure is free and finishes on the second working day after the stick is logged in. Where the fault is a deletion, a format or a broken file system, the work is logical and no fix, no fee applies to it. Rebuilding a torn-off connector at the microscope counts as electronic work instead. Lifting the memory package off the board, or grinding into a sealed one-piece stick to find its contacts, counts as chip-level work. Those last two want half the money in advance and appear by name in the published exclusions, and nobody is left to discover which category applies to them at the end of the job. Two things do genuine damage: hunting for one more socket in one more computer, and pointing a repair tool at a stick while it writes its own scratch files onto it. Stopping is free and it works. Send the stick in a small box or a stiff envelope where nothing can crush it, with any snapped-off pieces bagged up alongside.
Almost everything worked on here arrived in the post. A drive that is already in trouble has an easier time boxed, padded and insured than it does being carried round in a bag for an afternoon, and a parcel handed over in Coventry today is normally booked in at Oxford tomorrow morning.
As a rule the storage comes out and the machine stays where it is. That applies to a laptop, a tower, an iMac and to the recorder sitting under a counter. Taking equipment apart is not something this bench does, and a repair shop will free a drive in a few minutes. Two things go the other way: an external drive stays sealed inside its own case, and a NAS travels as a complete unit with its disks still in their bays. A Fusion Mac is a third case — both of its drives come out and travel together, each one labelled. The single situation nobody can work around is memory soldered flat onto a mainboard, which is how Apple Silicon Macs and a good many slim laptops are built: if the storage will not unbolt, there is no parcel to send.
↓ Print the shipping & booking-in form (PDF)
Put Oxford Data Recovery on the label. From Coventry it is roughly fifty-five miles straight down the M40, about an hour if you would rather drive it in than post it. Either way you are told the moment it is logged, and the free diagnostic finishes two working days later.
Unsure what ought to go in the box? Ring 0800 689 0668 before you seal it, or work through the free online diagnostic and let it do the asking.
The stick page in full: monolith chips, broken plugs, dead controllers
A stick reporting the wrong size is reporting a controller fault
Secure digital, microSD and CompactFlash, with the whole card cluster
Formatting in the body writes an index, not the pictures beneath it
Put it down and leave it alone; the diagnosis can wait an hour
Looking at it costs nothing, one written figure follows, and the band governing this page is £250 + VAT on a USB stick, a pen drive or a memory card.