Start from the only honest sentence in the field. A domain is worth what a specific buyer will pay for it on a specific day, and every estimate before that is a model of how likely such a buyer is. Any figure a tool shows you should carry a qualifier such as "estimate, not an offer", because a figure without that qualifier reads as a price and it is not one. Treat any tool that hides the qualifier with more caution, not less.
The four things that are measurable
Almost every credible estimate reduces to four inputs, and you can read all four yourself in about a minute.
The extension. This dominates everything else. A .com carries a premium over the same string on any other extension, and the gap widens as the name gets better, because the buyer most likely to pay a serious sum is the one who wants the name to look established. On the same string, .io and .ai carry real value in their own categories and .co is a distant third. Beyond that the resale market thins out quickly.
Length. Short is scarce, and scarcity is the whole aftermarket. Every character you remove cuts the population of possible strings by a factor of twenty six, so a five letter pronounceable .com sits in a category a nine letter one never enters. The step down from four characters to three is the steepest in the market.
Whether it is a word. A dictionary word, a plausible invented word and a random string are three different assets. A name that can be said once and typed correctly costs nothing to market. A name that has to be spelled out is a permanent tax on every conversation about it, and buyers price that tax in.
Commercial pull. A name that sits in a category where customers are expensive to acquire is worth more to the businesses in that category than an equally pretty name in a category where they are cheap. This is the input most easily mistaken for cleverness. It is not about how good the name sounds, it is about who has budget for it.
The one thing that is not measurable
Demand for that exact string, right now, from someone with the money. This is the whole difference between an estimate and a sale, and no model sees it. A name can score well on every measurable input and sit unsold for six years because the two companies who would have wanted it named themselves something else in 2019.
It also runs the other way. A name with mediocre inputs sells for a serious sum because one buyer has already built a product around it and the domain is the last thing standing between them and a launch. That is not value, it is leverage, and it never appears in an appraisal.
Why three tools give you three answers
Because they are answering three different questions. One is modelling what comparable names have sold for. One is modelling what a name like this would list for. One is modelling what an end user would pay if you found the right one. Those numbers should not agree, and when they do it is coincidence rather than confirmation.
The practical read is to treat any single figure as a class rather than a price. Under four figures, four figures, five figures: which band a name lands in is the useful signal, and it is usually stable across tools even when the exact numbers are not. If you want the tools themselves compared, domain appraisal tools goes through what each one publishes about its own accuracy.
How to sanity check an estimate in five minutes
- Look up what genuinely comparable names have sold for. Same length, same extension, same kind of word. Reported sales are public and they are the only real evidence in the whole exercise.
- Say the name to somebody and ask them to type it. If they get it wrong, discount everything above.
- Check whether the obvious buyer exists. If you cannot name a company that would want this string specifically, the top of the range is theoretical.
- Check the .com. If you are valuing a .net or a .io and the .com is registered and in use by a business, you are valuing the second choice and it prices like one.
What the estimate is good for
Not for setting an asking price on its own, and not for deciding that a name is a bad name. It is good for two things, and both are comparisons rather than absolutes.
The first is choosing between candidates. When you have three names left and no test separates them, an estimate that puts one of them in a different class tells you something real about how the name will wear, because the same inputs that make a name valuable make it easy to say and easy to remember.
The second is setting an expectation before you negotiate. If you are approaching a holder, the number tells you which conversation you are in. Going in at $500 for a name whose comparables cleared five figures wastes everybody's afternoon. How much to offer for a domain name covers the negotiation itself.
The cheapest version of all of this
If the name is not registered, none of the above applies. A free .com costs the standard yearly registration fee at any registrar, and no valuation is involved, which is why the availability check comes before the appraisal rather than after it. The expensive question only arrives when somebody already owns the string.
So run the checks in that order. Search the name and its close variants first with a domain availability checker, register the one that is free, and only open the valuation question for the names somebody already holds. If the name you want is listed at a premium by its registry rather than held by a person, premium domain names explains how that price works and why it often continues at renewal. If it belongs to a person, buying a taken domain covers the steps from the first message to the transfer.