TL;DR: On a completed eBay listing marked “Best offer accepted”, the price you can see is what the seller was asking. What the buyer actually paid is not published. Treat those rows as proof that the item sells - not as evidence of what it sells for - or your valuation drifts upward without anyone noticing.
The detail almost everyone misses
Checking completed sales instead of active listings is the single best habit in collectible pricing, and most guides stop there. But not every completed sale publishes a usable price.
When a seller runs a listing with Best Offer enabled and accepts an offer, the listing closes and is marked as sold. The number displayed beside it is the listing price. The accepted offer - the amount that actually changed hands - is not shown.
So a row that reads “$40.00, Best offer accepted” might have been a $40 sale. It might equally have been $28. You cannot tell from the page, and neither can anyone else.
Why this quietly inflates valuations
Offers get accepted because they are below the asking price. That is the entire reason the mechanism exists. So the hidden true prices are systematically lower than the visible ones - never randomly scattered around them.
Average a set of completed sales that includes those rows at face value and the result is biased upward. Not wildly, not obviously, and never in a direction that looks like an error. It just makes everything appear to be worth a bit more than it is.
That bias is worst exactly where it does most damage: categories where haggling is normal. Die-cast, figures, sports cards, anything with a large enthusiast market and a lot of mid-value listings. In those categories a meaningful share of completed sales can carry an unusable price.
What we do about it
We separate two things that are easy to conflate:
How many times did this sell? Every completed sale counts, including accepted offers. It sold; a buyer wanted it. This is the demand signal, and it is what tells us whether an item moves at all.
How many usable prices do we have? Only the sales that published what was actually paid. This - and only this - is allowed to set a value.
Ten sales of which four were accepted offers is a ten-sale market with six usable prices. Both numbers are true, they answer different questions, and reporting only one of them distorts the result. Report six sales and a healthy item looks thin. Average all ten prices and it looks more valuable than it is.
When too few usable prices remain, the honest answer is that the sample is small - not a confident number derived from a bad one.
How to check it yourself
- Search the item and filter to Sold items.
- Read each row for “Best offer accepted”, not just the price.
- Price from the rows that do publish a real figure.
- Use the offer rows as evidence of demand: they prove the item sells, and roughly how often.
- If almost every sale was an accepted offer, treat the visible prices as a ceiling, not an average - the real ones were lower.
It takes a few extra seconds per search and it is the difference between a price that holds and one that quietly overshoots.
Why we are telling you this
If you are selling to us, you should be able to check our reasoning. A valuation you cannot verify is just an assertion, and this is one of the specific places where a number can be wrong in a way that looks entirely reasonable.
It also cuts both ways. The same bias that makes an item look more valuable than it is can make a seller feel short-changed by an honest offer. We would rather you knew about the mechanism than assume we had simply lowballed you.