Almost everyone researching an antique starts by searching for similar items for sale, and almost everyone therefore starts with the wrong data. An asking price records what somebody hoped to get. Only a completed sale records what a buyer was actually willing to pay, and auction archives are the largest public collection of exactly that.
What is in an auction archive?
Records of lots offered at saleroom auctions, usually including the catalogue description, an image, the estimate and the price achieved. Many archives cover a great many salerooms over decades, which is what makes them so useful.
Some are free with limited results and some charge for full access. Individual auction houses also publish their own past results, which is worth knowing when you are researching a category one house specialises in.
How should you search them?
Start broad and narrow deliberately. Search the object type first, then add the maker, then add a distinguishing feature. Searching a long specific phrase usually returns nothing and tells you only that your wording differs from the cataloguers.
Use the vocabulary the trade uses rather than everyday words, since catalogues are written in a specialist register. Try alternative spellings and both singular and plural forms. If the maker name draws a blank, search the form and the period instead and look through the images.
How do you read a result correctly?
Check first whether the figure shown is the hammer price or the total including buyers premium, because archives differ and the gap between the two is substantial. Mixing the two across your comparables will distort the average badly.
Note the currency and the date of sale. A result from many years ago is evidence about that market rather than this one, and several categories have moved a long way since. Prefer recent results and treat older ones as background.
What makes a comparable genuinely comparable?
Same object, same period, similar condition, similar size, similar completeness. Change any one of those and the price moves, sometimes enormously.
Read the full catalogue description rather than the headline. Salerooms disclose damage and restoration in the body text, and a result that looks strong for your piece may describe an example in far better condition. Look for words indicating repair, replacement, associated parts or later additions.
How many results do you need?
More than one and ideally at least five. A single result tells you what happened once, and auctions produce outliers in both directions when two determined bidders meet or when nobody in the room wanted the lot.
Collect several, discard the highest and the lowest, and look at the middle of what remains. That figure is a reasonable expectation. Also count how many similar lots failed to sell, because a thin market with frequent unsold lots means your item may not find a buyer at all.
What do archives not tell you?
What the seller received. The hammer price is reduced by commission and various fees before it reaches the vendor, so a result that looks encouraging translates into meaningfully less in your account.
They also cannot tell you about condition beyond what was catalogued, or about the room on the day. A specialist sale full of collectors produces different money from a general sale, so note which kind of sale a result came from before relying on it.
When should you stop researching and ask someone?
When the results scatter widely with no clear middle, when you cannot find comparable examples at all, or when the figures are high enough that being wrong matters financially.
Wide scatter usually means the category depends on details you are not yet identifying, which is exactly the point at which a specialist adds most. Take your research with you when you ask, because a valuer will engage far more usefully with someone who arrives having already looked. Bring the printed results as well as the object, since a specialist can tell you in seconds whether the examples you found are genuinely comparable or whether you have been measuring the wrong thing.