How Estibot Values a Domain, and When to Ignore It
Estibot is the oldest automated appraisal still in wide use, and its logic is unusually legible. Once you know what it counts, you know exactly which of its numbers to trust.
On this page
Estibot prices the keyword, not the name. Its model is built on search demand, advertiser competition and how widely a string is registered across extensions, which means it produces credible numbers for exact match commercial phrases and near zero for anything invented. Investors who get value out of it stopped reading the headline valuation years ago and use the sub-scores as a research panel instead.
That distinction is the whole review. Here is how the model gets there.
The model in one paragraph
Estibot is one of the oldest automated appraisal engines still in general use, and its logic predates the current wave of sales trained machine learning. It breaks a domain into its component words, looks up what those words are worth in search advertising terms, checks how many people have registered the same string in other extensions, factors in length and extension, and cross references reported sales of similar names. Then it prints a number.
The lineage matters. The tool was designed in an era when a keyword domain could earn money from parked type in traffic, so the underlying question it asks is closer to “what would this name earn?” than “what would somebody pay for this name?” Those were nearly the same question once. They are not now.
Why exact match keyword names score so well
Feed it something like CommercialInsurance.com and the inputs all line up. Real search volume, expensive clicks, heavy registration of the string across extensions, and a stack of comparable sales in the insurance category. The model reads a business asset and prices it accordingly.
On genuinely strong exact match .com names, that is defensible. Where it breaks down is on the long tail of the same shape. Add two modifiers to the phrase and the search volume splinters, the buyer pool collapses, and no similar name has sold in years, but the model still recognises valuable ingredient words and still returns a substantial figure.
The failure mode is specific and worth naming: the model prices the phrase, not the odds of a sale. A phrase can be commercially valuable while the domain built from it is unsellable, and nothing in the keyword data distinguishes those two states.
Why brandables score near nothing
Invented names return almost nothing, and the reason is mechanical rather than aesthetic. A made up word has no search volume, no advertiser bids, and typically one registration in one extension. Strip those away and the model has length and extension left, which is not enough to value anything.
So a genuinely excellent brandable, short, pronounceable, clean to spell, can come back at scrap while a clumsy four word keyword phrase comes back at four figures. If you hold brandables, this tool is not merely imprecise, it is measuring the wrong dimension entirely. The traits that make those names valuable are the ones covered in what makes a domain brandable, and none of them appear in keyword data.
The sub-scores investors actually read
Ignore the top line and the page becomes genuinely useful. Four things on it are real research you would otherwise pay for.
- Search volume for the exact phrase. Not because volume sets price, but because zero volume on a supposedly commercial term is a red flag you want early.
- Advertiser competition and cost per click. The best cheap proxy for whether money moves in the category. Businesses that pay well per click have budgets and buy domains.
- Registration counts across extensions. Covered below, and the most underrated number on the page.
- The comparable sales list. Checkable evidence, unlike the valuation itself.
Read those four, form your own view, and treat the headline as decoration. That is how the tool fits into a manual appraisal rather than replacing one.
The registration count is the signal to watch
If a string is registered in fifteen extensions, fifteen people at some point thought it was worth owning. If it is registered in two, almost nobody did. This is a direct, unmediated measure of historical demand for that exact string, and it comes with none of the modelling assumptions that make the valuation shaky.
It cuts both ways, which is what makes it interesting. A high count on an available or expiring .com means you have found a name with proven appeal and you should move quickly. A high count also means competition when the name drops, so expect it in the auction rather than in the daily expired list as a quiet catch.
A count of one or two on a name you are excited about is worth sitting with. It is the market’s long running vote, and it usually knew something.
When to ignore the headline number entirely
Four situations, no exceptions worth making:
The name is invented or abstract, where the model has no inputs. The name is three or more words, where the keyword value is real but the buyer pool is not. The name is in an extension other than .com, where comparable data thins out fast. And when the valuation sits far above every comparable sale the tool itself is showing you, which is a straightforward sign of extrapolation.
That last check applies to every appraisal engine, and it is the fastest way to catch a number that has run ahead of the evidence.
Where it sits against the newer models
Sales trained appraisals, the current generation, learn from what actually changed hands and largely ignore search volume, because volume turned out to be a weak predictor of price. They are better at short liquid names and less prone to inventing value on long keyword strings.
They are not strictly better. They inherit their own bias, since they only ever learn from names that sold, and they are blind to the fact that a phrase carries real commercial weight. That is the substance of why appraisal tools disagree, and it is why the mature workflow runs a keyword model and a sales model side by side.
Agreement between them is a decent confidence signal. Disagreement is the more useful outcome, because it tells you which kind of asset you are holding and therefore which kind of buyer you need to find.
Key takeaways
- Estibot values a name mainly from keyword economics: search demand, advertiser competition, and how widely the same string is registered across extensions.
- That model systematically flatters exact match keyword domains, because it prices the phrase rather than the likelihood that anybody will buy the name.
- Brandable and invented names score near zero, since a made up word has no search volume and no advertiser bids for the model to read.
- Experienced investors use the sub-scores as research inputs and ignore the headline valuation, which is the least useful figure on the page.
- The registration count across extensions is the most underrated signal it gives you, because it is a direct measure of how many people wanted that string.
Frequently asked questions
Is Estibot accurate for domain valuation?
It is accurate about keyword economics and unreliable about price. The underlying data on search demand and advertiser competition is real research, but the leap from that to a dollar figure assumes a buyer exists who monetises the phrase. For exact match commercial names that assumption sometimes holds. For everything else it usually does not.
Why does Estibot value my brandable domain at almost nothing?
Because an invented word has no search volume, no advertiser bids and no registration footprint, which removes nearly every input the model uses. The valuation is not a judgement that your name is bad, it is the model reporting that it has nothing to measure. Price brandables against comparable brandable sales instead.
Is Estibot free to use?
There is a free tier with a limited number of appraisals, and paid plans for higher volume and bulk processing. Check their current pricing before committing, since the tiers change. For occasional manual research the free allowance is usually enough, and bulk plans only make sense if you are screening large lists regularly.
Should I use Estibot or a sales trained appraisal model?
Use both and read them as answers to different questions. Estibot tells you whether the phrase has commercial value. A sales trained model tells you what strings of that shape have actually sold for. When both agree the name is probably priced correctly. When they disagree sharply, that gap is telling you which type of asset you really hold.