Data Recovery Case File · Solid State & Flash · Capacity Is Not the Price

USB Stick Recovery Cost: Capacity Is Not the Price

Her enquiry was short and asked the question everyone asks first. "I have a USB stick which is 8GB. It does not respond. I would like to know if I can get the data off it and how much it would cost." She mentions the capacity because capacity feels like it should matter — a small stick sounds like a small job. It very nearly does not matter at all. An 8GB stick is not meaningfully cheaper to recover than a 128GB one, because the work is in the route rather than the volume, and the route is decided by things her description does not contain. This page sets out what actually sets the figure, so she can understand the answer she gets — and one genuine advantage her small, older stick may have.

The decode: four things that set the price, and none of them is capacity

One — does it enumerate at all? The biggest single factor. A stick that introduces itself to a computer, even unreliably, can often be imaged through its own controller: relatively quick, no invasive work, and the cheapest outcome available. A stick that produces no response has a failed controller, and the memory must be read directly — a different and more involved class of work. "It does not respond" suggests the second, but it is established properly rather than assumed, because occasionally a stick that appears dead answers behind hardware that can manage an unstable device.

Two — how is it built? Flash drives come in two constructions. Some contain a discrete NAND package alongside a separate controller, and that package can be desoldered and read directly in a programmer. Others are monolithic — memory and controller combined in one moulded block — with nothing to remove, so the read is taken through internal test points beneath the surface, located and probed against a database of layouts. The monolithic route takes longer and is more exacting. Neither is visible from outside, which is why an accurate figure follows an assessment rather than preceding it.

Three — is the controller a known family? Reading raw memory produces data that is scrambled, distributed by a wear-levelling scheme and error-coded, all in ways specific to the controller that wrote it. For a recognised controller family, those parameters are looked up and applied. For an obscure or unbranded one — and unbranded sticks frequently use whatever was cheapest that quarter — they must be determined by analysis first, which adds real time. This is the main reason two sticks of identical capacity can differ substantially in price.

Four — how worn is the memory? A heavily used device yields blocks that need aggressive error correction, and some that cannot be corrected at all. That affects both the time taken and the completeness of the result.

The advantage a small old stick may have: genuinely encouraging for her. Lower-capacity sticks from earlier generations tend to use simpler, well-documented controllers and less densely packed memory, which can make identification straightforward and the reconstruction cleaner than on a modern high-capacity device. So while the capacity does not reduce the price directly, an older small stick is sometimes the easier patient.

What she should expect commercially: an indicative range from a description, offered as an estimate; a free assessment establishing which route applies; then one fixed figure in writing, including VAT, before any work begins. Where a chip has to be removed, parts and labour are consumed either way, so a deposit applies with the balance only on success. What she should not accept is a confident fixed price quoted sight unseen, or a figure that can grow later.

On the bench

An enumeration attempt ran first behind the DeepSpar USB Stabilizer 10Gb, because the cheapest possible route is always tested before the invasive one and a device described as unresponsive occasionally answers behind hardware built to manage instability. Where it did not, the construction was identified — monolithic or discrete — and the memory read accordingly on the PC-3000 Flash, with the controller family and pinout confirmed against the Rusolut VNR database before reconstruction. The fixed figure had been issued in writing before any of that began, and it did not change. Her files were verified by opening and delivered on fresh media.

The outcome

The route established free of charge, the figure fixed in writing before work, and the contents verified and delivered. Free assessment, one fixed written figure including VAT; where a chip has to be removed, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The pricing decode: capacity is almost irrelevant, because the work is in the route rather than the volume. What sets it is whether the device enumerates at all, whether it is monolithic or has a separate memory package, whether its controller belongs to a documented family or has to be worked out, and how worn the memory is. Expect an indicative range from a description, a free assessment, and then one fixed written figure that does not move — and treat a confident fixed price given without the device as a reason to ask more questions.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.