Some are, most are not, and the only way to tell is to check an appraiser's past calls against sales that happened afterwards. Free estimators from registrars return a figure for any string without explaining it. We publish our pre-sale reports next to the reported sale, hits and misses alike, and tested the engine against 1,217 of the largest sales of the last decade: 74.8% closed within a band from 20% of our low to twice our high, and 12.5% landed inside the range itself.
Short answer: an appraisal is accurate if the sales that happen after it land where it said they would. That is testable, so test it. Ask any appraiser for sales that came after their valuation, not before, and for the ones they got wrong. If they cannot show you either, you are looking at a number generator.
Below: why free tools disagree with each other, what our own record looks like when we hold it up to the same standard, and what "accurate" can even mean for an asset that sells for different prices to different buyers.
Free estimators exist to sell you something else: a registration, a listing, a brokerage. They return a figure for any string you type, including nonsense, and they do not explain it. Most score surface features, such as length, extension and whether the words appear in a keyword list, so unrelated names of the same shape get similar values. That is why a free tool will value a meaningless string and a real brandable word within a few hundred dollars of each other.
A paid appraisal should earn its fee by doing three things a free tool does not: explain the number (strengths, weaknesses, who the buyer is), show the comparable sales it rests on, and put its past calls on the public record so you can check them. Price is not the test; evidence is.
Two public tests, both with the misses left in.
Read that honestly: the market does not land on the range, it distributes around it, with the centre of mass below retail because most reported sales are wholesale and most end-user sales are never reported. The range works as the ceiling anchor that real prices spread down from. That is what an accurate retail appraisal of an illiquid asset looks like. Anyone promising a tighter target is promising to predict which buyer walks in.
Valuations, human or automated, vary between sittings. Measured across more than 1,200 domains appraised five times each, the typical gap between a name's highest and lowest single run was 29%; one domain in three exceeded 50%; on roughly one in five, at least one run spiked above 1.5x the median.
So we do not sell single runs to the public. Every $19.99 report is a consensus of five independent valuations with the highest and lowest dropped. In our measurements, swapping any one of the five opinions typically shifts the consensus by 0%. The spread is still there in the market; it is just no longer in the appraisal.
Because most of them score the string, not the meaning: length, extension, a keyword list, sometimes search volume. Two tools weighting those differently produce different numbers, and neither knows whether the words describe something a business would pay for. They also never show their reasoning, so you cannot tell which one to believe.
An appraisal cannot predict which buyer shows up, so it cannot be a single number. It can be accurate about the retail range: what an end user who needs the name would reasonably pay. A reseller buying wholesale pays well under that range on purpose, and that is not the appraisal being wrong.
Any single valuation is one sample from a spread. When we appraised more than 1,200 domains five times each, the typical gap between a name's highest and lowest run was 29%, and one domain in three exceeded 50%. A consensus runs five valuations, drops the highest and lowest, and converges the middle three. Replacing any one opinion then typically moves the result by 0%.
We calibrate each extension's valuation curve against publicly reported comparable sales, and the report shows the comparable sales we consider relevant so you can judge them yourself. The valuation itself is built from the qualities of the name.
Yes. The accuracy gallery shows the reported sale next to the full report we produced before it, and every entry is graded in range, near, or miss. TXT.com, for example, is a published miss: it sold for far less than our range. We would rather you see the misses than take our word for the hits.
One domain, $19.99, no account. Five independent valuations converged into one range, with the reasoning and comparable sales you need to check our work.
Get a consensus appraisal for $19.99