A domain valuation is built from four things: comparable completed sales, an analysis of who could realistically buy the name and why, the economics of the category it serves, and how liquid the asset is. The output should be a range with the reasoning attached, because a single figure implies a precision that this market does not support.
Comparable sales, used carefully
Completed domain sales are published in various places, and they are the closest thing this market has to evidence. They are also routinely misread.
The first error is comparing to asking prices rather than sale prices. Listed prices are aspirations, and the population of names listed at a figure is not the population of names sold at one.
The second is treating superficially similar names as comparable. Two names in the same category can differ by an order of magnitude in value because one is the obvious address for the category and the other is a variation on it. A comparable is only useful when the name occupies a similar commercial position, not merely a similar shape.
The third is ignoring when the sale happened. Categories move, and a price from an era when a market was fashionable tells you about that era.
Buyer analysis
This is the part automated tools cannot do and the part that determines most of the answer.
The question is who could plausibly want this domain, how badly, and what the alternatives cost them. A company currently trading under a longer or more awkward name has a real reason to move. A company with an established brand and no confusion problem does not, however well the domain matches what it sells.
The number of such buyers matters as much as their enthusiasm. One highly motivated buyer produces an unpredictable outcome; a category of twenty companies with the same problem produces something closer to a market price.
Category economics
What is a customer worth in this industry, and what does one cost to acquire?
That single relationship explains most of the variation in domain prices between categories. Where a customer is worth thousands and competition for attention is expensive, a domain that improves credibility and reduces acquisition cost can justify a substantial figure. Where a customer is worth very little, no name can justify much.
Liquidity and time
A valuation that ignores time is incomplete. A domain that might achieve a strong price from one buyer within five years is not equivalent to one that could sell to any of thirty buyers this year, even if the headline figures match.
This is why a considered valuation usually contains two numbers rather than one: what the name might achieve given the right buyer and enough patience, and what it would realistically achieve if it had to be sold in the near term. Owners are frequently quoted the first and plan around it as though it were the second.
Why the output is a range
A domain has no market price in the sense a listed share does. It has one buyer at a time, negotiating privately, with information neither side fully shares. The outcome depends on who that buyer turns out to be and what pressure they are under, neither of which is knowable in advance.
A single figure conceals all of that. A range with reasoning exposes it, which is less satisfying and considerably more useful — particularly when the reasoning includes the conditions under which the top of the range would be reached.
Questions
Can anyone tell me exactly what my domain is worth?
No. A valuation is an informed opinion about a range, supported by comparable sales and an analysis of who might buy. The price is set at the moment a specific buyer agrees to it, and until then no figure is more than an estimate.
Why do two valuations of the same domain differ so much?
Because they are answering slightly different questions. One may be estimating what the name could achieve from an ideal buyer given years of patience; another may be estimating what it would fetch this quarter. Both can be reasonable, and neither is useful without knowing which was asked.