Parting out or selling whole: spot the diamonds in the rough

The instinct with an old PC you want gone is to photograph it, write “gaming PC” at the top, and list the whole thing. It is one listing instead of six, one buyer instead of six, one package instead of six. It feels like less work, and it usually is.

It is also usually the wrong call. Unless the machine is a coherent system someone can sit down and use as it is, you will make more by taking it apart. Not a little more. Often a lot more. The reason is not really about the hardware at all. It is about who is buying, and what they think they are paying for.

Why does a whole old PC only sell at a discount?

Nobody sets out to buy a whole used PC at full price. They set out to buy a compromise.

A built machine was assembled around one person’s priorities, at one moment, on one budget. The next buyer has different priorities. They love the graphics card, they are indifferent to the case, they already have storage, and they think the power supply is underpowered for what they want to do next. They are not buying six components they all want. They are buying one or two they want and four they will tolerate, and they price the whole thing off the parts they actually care about.

On top of that, they are pricing in risk. A whole machine is a sealed box of unknowns. It could have a failing drive, a power supply on its way out, or a fault that only shows up under load. The buyer cannot test each part in isolation before they commit, so they assume the worst on anything they cannot see proven and knock the price down to cover it. This is exactly why proving the machine works lifts a whole-system price at all: you are buying back some of that discount.

Put those together and a whole old PC is worth, roughly, the value of the one or two parts the buyer wants, minus a margin for everything they will have to work around, minus another margin for the risk they are taking on. That is a lot of subtraction.

Why does a single part sell for close to full price?

Now change who is buying. Instead of someone shopping for a machine, it is someone shopping for a part.

Their monitor is fine, their PC works, and their graphics card just died. They want one thing: a replacement graphics card, that exact tier, working, today. They are not comparing your card against other whole PCs. They are comparing it against buying the same card new, and against the handful of other people selling that same card used. That is a narrow, direct comparison, and it holds the price up.

There is no compromise in that transaction, because they picked the component. There is far less risk, because a single part can be tested on its own and described precisely. And there is real urgency, because a broken part is a problem they want solved now. All three push the same way: the person buying the individual part pays close to what the part is actually worth, while the person buying the whole machine expects a discount for taking the rest of it off your hands.

That gap, between full part price and discounted whole-machine price, is the money you leave on the table every time you list a tired build whole.

What actually counts as a diamond in the rough?

The phrase is doing real work here. A diamond in the rough is a part with a strong standalone market sitting inside a machine that looks, as a whole, like nothing special. The machine drags the eye down. The part is what matters.

The parts that reliably qualify:

The graphics card, almost always. It is the most liquid used component there is, because it is the upgrade people shop for on its own. In a lot of old builds the GPU is worth more than everything else combined, which is the textbook signal to part out. A current card in an otherwise dated build is the clearest diamond of the lot.

A CPU still on a live socket. A chip on a platform that is still selling new processors has a real upgrade-buyer market. The same chip on a dead socket does not, so this one depends entirely on the platform, not the raw performance.

A fast or high-capacity RAM kit. DDR4 stopped being the throwaway commodity it used to be as production wound down and prices climbed. A matched 32GB kit is now worth listing on its own, where a year ago you would have thrown it in.

NVMe and SSD storage. Drives sell quickly and separately because everyone building or upgrading wants one. Note the capacity and, for a used drive, the bytes written from its SMART data, and it sells itself.

A good branded power supply. A high-wattage, efficient, modular unit from a name-brand holds value well, and there is always someone whose build needs exactly that wattage. A no-name unit that came with a case is not this. It goes in the bin, not the listing.

Everything else, the motherboard, the case, the cooler, a mechanical drive, is usually thin. Motherboards depreciate hard unless they are high-end or on a scarce socket. Cases and coolers move slowly and cheaply. Those are the parts you bundle together, throw in with something else, or accept a low number on. The art is not squeezing every last dollar out of the case. It is making sure the diamonds get listed as diamonds.

How do you spot one before you list?

The test is quick and it is the same one that works out what the whole machine is worth, just pointed at one part at a time.

Take the component you suspect is carrying the build. Look up its recent sold price on its own, from completed and sold listings, not asking prices. Then look up what whole machines like yours actually sold for. Hold the two numbers next to each other.

If that single part sold for a large fraction of the whole-machine price, or more than it, the part is holding the listing up and the rest of the machine is dragging the buyer pool down. List it on its own. You will reach the buyer who wants precisely that part and pays precisely for it, instead of hiding it inside a box that only sells at a discount.

Do this for the graphics card and the power supply at a minimum, because those are the two that most often turn out to be worth more alone than the whole thing is worth together.

When should you still sell whole?

Parting out is the default, not a law. There is one situation where selling whole genuinely wins, and it is worth naming so you do not part out a machine that would have sold better in one piece.

Sell whole when the machine is a coherent, balanced system that someone can actually use as it is. Not “old PC someone could tolerate”, but a build where the parts make sense together, nothing is individually remarkable enough to be worth more alone, and the whole thing is current enough to be somebody’s working computer on day one. A sensible mid-range build where every part is pulling its weight is worth more assembled, to a buyer who wants a machine and not a project, than it is in pieces.

The tell is the same test run the other way. If no single part is worth a large fraction of the whole, the value is genuinely in the sum, and the convenience of a ready-to-use machine is worth paying for. That build sells whole. Everything else, everything where one part is carrying the rest, or where the machine as a whole is dated but one piece is not, parts out.

Doing this more than once

Value one machine and the maths above is a five-minute job with a browser and a notepad.

Do it regularly and the hard part is no longer deciding whether to part out. It is keeping track of what happened after you did. A machine comes in as one purchase. It leaves as a graphics card sold to one person, a RAM kit to another, a drive to a third, and a stripped chassis that sat in a cupboard for two months before you gave it away. To know whether any of that made money, you have to tie one buying price to several selling prices spread across weeks, minus the fees on each one.

That is what PartPlanner is for: one purchase cost split across the parts it became, and every separate sale tracked back to it, so parting out is a decision you can measure rather than a hunch you hope paid off.