Methodology

How DnRatings values a domain name

DnRatings answers two questions about any domain: what would it sell for, and is anyone likely to want it. Here is how each answer is made, and where its limits are.

1. A price model trained on real sales

The estimate comes from a gradient-boosted tree model trained on a large set of real domain aftermarket sales across venues such as GoDaddy, Sedo, Dynadot, Namecheap and DropCatch. It reads the name itself: length, extension, dictionary words and how common they are, brandability (how it sounds and how easy it is to spell), commercial keywords, numbers and hyphens. Synthetic low-value examples keep it from over-pricing random strings, a common failure of automated appraisals.

2. A value range, not a single number

Every valuation shows a low, mid and high estimate and a confidence score. The range is calibrated on sales the model never saw during training, so a wide range honestly means "uncertain". Short, clean .com names get tighter ranges than long or unusual ones.

3. Buyer demand from real auctions

Price alone doesn't tell you whether a name will sell. DnRatings also scores buyer demand: the chance a name gets at least one bid if it goes to an expired-domain auction. That model is trained on the outcomes of real GoDaddy expired-domain auctions, where under 1% of names get a bid. On auctions it hadn't seen, it ranked names that drew bids above those that didn't about 94% of the time (AUC 0.94). The valuation shows the chance next to the average, for example "12% against 1%".

4. Comparable sales and context

Valuations list the most similar past sales. They also show domain age, search interest and estimated cost-per-click when they're available, plus a plain-language summary of strengths and weaknesses.

What the numbers are not

Tools around the valuation

Questions about the method: see the FAQ or write to [email protected].

Value a domain now

Free account, 5 valuations a day, and 7 days of Pro to start.

Get started free