You hand me a drive that is failing. Here is what happens next.

First, I do not go looking for your files.

That surprises people. The instinct is to plug it in and start copying the photos folder, and if the drive were healthy that is exactly what I would do.

On a failing drive I make an image first. An image is a copy of the whole drive, every sector of it, in order, whether or not anything is stored there. It goes onto a known good disk.

There is one reason for it. The drive has a limited number of reads left, and I would rather spend them once, in a single controlled pass, than a hundred times while browsing around. Once the image is made, the failing drive gets unplugged and never touched again. Everything after that happens against the copy, where I can take as long as I like and nothing gets worse.

The imaging is the part that takes the time.

Not the searching. A good copy of a healthy drive takes an hour or two. A bad spot on the drive makes it slower, because the software waits on the drive to try, give up, and move on. Imaging tools work around this by taking the easy parts first, fast, and then coming back for the difficult areas afterwards. That way the bulk of your data is safe early, and the stubborn bits get worked on with everything else already secured.

A badly failing drive can take days. That is normal and it is not a sign anything is going wrong.

Then the files come out of the image.

Sometimes they are simply there, in folders with names, and it is a copy and paste. That is the good day.

Sometimes the index is damaged, so the drive knows the data is there but not what it is called or where it belongs. Then the files get rebuilt by their contents rather than their names, and you get your photos back with names like IMG_0001 in a folder of four thousand. Everything is there. The organization is not. Sorting that out is tedious and it is still a good outcome.

Sometimes parts are unreadable. Then the answer is partial. Most of the photos, not all. A document that opens with a chunk missing.

When it stops being my job.

My bench covers a drive that still reads, even poorly. A drive with a logical problem. A drive behind a dead enclosure.

What I cannot do is anything mechanical. A seized motor, heads that have touched the platter, a snapped connector. That has to be opened in a clean room, and opening it anywhere else finishes it.

I will tell you plainly which one you have. I would rather send you to a lab on day one than take your money for a week and then send you to a lab.

Why nobody can quote this over the phone.

The price depends on what failed, and nobody knows what failed until the drive is connected and read. Anyone who gives you a firm number sight unseen is either guessing or has decided in advance what you are going to be charged.

What you should get instead is an evaluation. Somebody looks, tells you what they can get and what it will cost, and you decide then. An honest shop charges nothing if it gets nothing. Ask that question before you hand anything over, and be wary of anyone who takes the drive and goes quiet.

What this week came down to.

A drive tells you it is going. It tells you in noises, in stalls, in files that will not open, and in a message people click past.

Then something happens that is not actually the drive's fault, and Windows says it is not there.

Then the first hour decides the bill.

Then, if it has come to it, a copy gets made and the files come out of the copy.

Every one of those steps is easier and cheaper than the one after it. The whole series is really one sentence: the earlier you stop, the more you get back.

If you are looking at a drive you think is gone, bring it in before you send it anywhere. Ten minutes tells you whether you need a lab at all, and most of the time you do not.