Finding the real price of used Macs with data and AI
4 min read
When I want to buy something used, I do not guess what a good price is. I look at the market. A short research routine tells me what the normal price is and which listings are unusually cheap.
Start with the median
Collect the listings for the item you want, then calculate the median price. The median is more useful than the average, because a few overpriced or broken listings do not distort it. Once you know the median, a good deal is easy to recognise: it is clearly below it, for a reason you can explain.
The messy part: descriptions
Prices are only comparable when the items are comparable. Used Macs are a good example: sellers often describe them badly. The amount of RAM, the processor or the storage is missing, written in different ways, or only visible in the photos. Two listings at the same price can be completely different machines.
Where AI helps
This is a good job for AI. I let it read each description and extract a clean set of fields: model, processor, RAM and storage. Where the information is missing, it says so instead of guessing. After that, I can group listings by real specification and calculate a median for each group.
The result
- A realistic price for each configuration, not for “a used Mac”.
- A short list of listings that are actually below the market.
- Less time spent reading descriptions by hand.
The same approach works for any market where descriptions are inconsistent: collect, normalise, compare, and decide.