Why asking prices are not the same as sold comps
You search the thing in your hand, you see a row of listings at $400, and you decide that $60 at the estate sale is a steal. Nothing in that sentence is evidence. Here is why, and what to look at instead.
The two numbers
An asking price is a number a seller chose. That is the whole of it. Nobody agreed to it. Nobody paid it. It has passed no test.
A sold comp is a completed transaction. Somebody wanted the thing enough to move money, and somebody else was willing to let it go at that figure. Two parties, one number, done.
Those are different categories of fact, and only one of them is a fact about the market.
Why the gap survives
The reason bad numbers stay visible is mechanical, not mysterious.
When something sells, the listing disappears from the default view. When something does not sell, it sits there. Relisted, renewed, bumped, still up two years later at a price nobody has ever agreed to. So the longer a hopeful price fails, the more search results it accumulates.
Search any category and you are looking at a filtered sample: the items that could not find a buyer, plus whatever is currently live. The successes have already left the room. That is survivorship bias running in reverse, and it reliably makes a market look richer than it is.
Five things that break a comp even when the sale was real
Finding a completed sale is the start of the work, not the end. A sold price is only useful if the item behind it is genuinely the same trade you are contemplating.
1. Condition
The most common way to overpay is to comp a damaged object against clean ones. A first edition with a price-clipped jacket and a cracked hinge is not the same object as a bright, tight copy, and the gap between them is often most of the value. Photographs of a comp are worth more than its title.
2. Shipping friction
Heavy, fragile, oversized and awkward items sell for less than their value suggests, because the buyer is paying twice and the seller is absorbing risk. A $200 vase that costs $45 to pack and ship safely, with a real chance of arriving broken, does not behave like a $200 item. Furniture and large ceramics are where this bites hardest.
3. Category mismatch
Similar-looking is not comparable. Same maker, different line. Same line, different year. Signed against unsigned, which is frequently a multiple rather than a margin. A comp set that has quietly mixed two variants will hand you an average of two different markets and a number that describes neither.
4. Lots and bundles
A box of thirty items that sold for $150 is not thirty $5 items. Bundle sales carry a discount for the convenience of moving everything at once, and they contaminate a comp set badly if they are not pulled out. The reverse trap also exists: one strong piece can sell an entire lot, and the rest was filler.
5. Thin data
Two sales is not a market. It is two anecdotes that happen to have prices attached. For genuinely uncommon objects, thin comps are the normal condition rather than a failure, and the correct response is a wider range and more caution, not a confident average of a tiny sample.
Sometimes the honest answer is that there is no answer
For odd, unidentified or genuinely rare objects, there may be no usable comp at all. Nothing close enough, nothing recent enough, nothing in comparable condition.
That is a real result and it should be reported as one. The failure mode is not saying "I do not know." The failure mode is generating a tidy number anyway, because a specific figure feels more useful than an honest shrug, and then acting on it with your own money.
Doing it by hand
None of this needs an app. It needs discipline, and about four minutes.
- Name the object properly firstMaker, line, year, variant. Pricing an object you have misidentified produces a confident number about the wrong thing. On unfamiliar items this is most of the work, as in this $5 book nobody could name.
- Filter to completed salesOn the marketplaces that offer it, switch the search to sold or completed items. This single toggle is the difference between a hope-based number and a market-based one, and most people never touch it.
- Throw out the ones that are not your itemWrong variant, wrong condition tier, bundles, obvious misdescriptions. Be ruthless. A small clean comp set beats a large dirty one every time.
- Look at the spread, not the averageThe range tells you how confident the market is. Tight spread means a known quantity. Wide spread means condition or variant is driving the price and you need to work out which side of it you are on.
- Subtract the cost of sellingPlatform fees, payment fees, shipping, materials, and the hours. A $100 sold comp is rarely $100 in your pocket, and the gap decides whether a $60 buy was clever or not.
How this app handles it
This is the part What's It Worth? is actually built around, so it is worth being specific about the design choice.
The baseline estimate is anchored to completed sales. Active listings are still collected, but they are kept in their own bucket and used to inform the upper end of a range, never to set the baseline. Before anything is averaged, the comp set is filtered: duplicates removed, outliers dropped, bundle listings pulled out, and variants separated so signed and unsigned versions of the same thing are not silently blended.
The app also reports how thin the evidence is instead of hiding it. Seeing that a category has plenty of listings but very few completed sales is exactly the signal that should make you cautious, and it is the kind of thing a single averaged number is designed to obscure.
When identification is uncertain, the range widens and the app asks you for the detail that would settle it, rather than guessing. That is deliberate. Enthusiastic about working out what something is, conservative about telling you what it is worth, and those two behaviours are not in tension. They are the same commitment to not making things up.
Why this matters standing in a stranger's garage
Estate sales, thrift stores and flea markets are exactly where the asking-price illusion does the most damage, because you are deciding fast, with somebody behind you who also wants the thing.
The trap is not that you cannot check. It is that the easy check, the top row of a search, is the misleading one. It is optimised to show you things for sale, and things for sale are things nobody has bought.
Getting this one distinction right will do more for what you pay than any amount of category expertise. Buy against what people actually paid, not against what people wish they could get.
Check it before you hand over money
What's It Worth? is free to try on iPhone and Android. Photograph the item, get a read on what it is, and see a range built from completed sales rather than hopeful listings.
More how-to material in Guides. Real objects worked start to finish in Field Notes.
Estimates are informational and not a certified appraisal. For a formal valuation, use a qualified appraiser.