how domain valuation works.

It isn't a marketplace, a registrar, an auction, or a brokerage. Nothing here is for sale, there are no accounts, and the application does not maintain personal search histories. Anonymous, aggregate page-view measurement is provided by Vercel Web Analytics. Links to registrars are plain links, not affiliate links.

Registration data — creation and expiry dates, registrar, status codes, nameservers and DNSSEC state — comes from the registry's own RDAP service, the official successor to WHOIS. Responses are cached briefly to avoid hammering registries. Registrant contact details are never shown.

Availability follows the same rule. When the authoritative registry has no record for a name, we report it as available. When a registrar API is configured, it takes priority and can supply real registration pricing. When neither can answer, the page says availability is not verified rather than guessing.

If the RDAP bootstrap has no usable service for an extension, or a registry times out, the domain shows “registration data unavailable” — an honest gap instead of filled-in numbers.

Registration prices come from Porkbun's public price list, which is a real, current registrar price. It is a list price for the extension, not a quote for one specific name: registry-premium domains are priced individually and cost more, and the interface says so wherever a price appears.

Market calibration comes from publicly reported sales, published aggregate transaction benchmarks and, when configured, GoDaddy GoValue. GoValue contributes retail, wholesale, suggested-list and sell-probability signals from its aftermarket model. Those outputs remain model evidence — they are not silently relabelled as completed transactions.

Discovery generates candidate names locally, screens the full batch over DNS, confirms the strongest openings against registry RDAP and checks finalists against the Internet Archive. A registered but undelegated domain can look absent over DNS, so the interface distinguishes a fast screen from a registry-confirmed result.

Read the primary references: the ICANN RDAP documentation, the GoValue API reference and Escrow.com's q2 2025 domain investment index.

Discovery generates hundreds of candidate names around a seed word: semantic compounds, prefixes, pronounceable coinages and dedicated three- or four-letter forms. The generator is deterministic and runs without a third-party language model or API key.

DNS screens the full batch quickly, but a registered and undelegated name can look absent. The deepest shortlist is therefore checked against registry RDAP and known registered names are removed. Confirmed rows carry a check mark; where RDAP cannot answer, the interface says “screened” instead of claiming certainty.

Investor ranking combines name score, modeled liquidity and real registrar carrying cost. The resulting shortlist still requires human judgment; availability, estimated value and the chance of eventually finding a buyer are separate questions.

The Domain Score is a weighted blend of seven heuristics — brandability, length, TLD quality, memorability, commercial intent, keyword quality and buyer reach — computed from the name itself. Each one explains its own reasoning on the analysis page.

These are rules, not measurements. They are not externally verified metrics, and no part of the score reflects traffic, revenue, or anything happening on the domain today.

The valuation begins with the extension and name structure: length, language, commercial category, hyphens or digits, domain score and age. It then blends whatever real market evidence applies in logarithmic price space: a reported sale of the same domain, relevant reported comparables, a published market-class benchmark, and an optional live GoValue appraisal. Similar names receive deliberately less weight than an exact-domain sale.

Every domain gets a number. Above roughly $1M the raw output is compressed — an uncompressed multiplicative model happily prints $47M for a two-letter name — and past $1.5M the estimate is marked as a wide range, because at that tier the price is set by one buyer's strategic need rather than by anything visible in the name. A wide, honest range is more useful than a blank.

The output separates an investor-to-investor wholesale exit, a patient end-user range, a suggested listing price and a 0–100 liquidity score. Liquidity is a modeled relative signal, not a measured sell-through rate. Provider sale probability is only shown when the live provider actually returns one.

One more guardrail: a name nobody has registered is discounted hard. The market has already passed over it at registration cost, so the figure shown is potential, not resale value.

• Reported sale price

  What a public source reported for a past transaction. Historical evidence, never a current offer.

• Current asking price

  What a seller is asking today, from a marketplace listing. A fact about an offer, not about value.

• Model estimate

  A blended range from name quality and the market evidence available for this domain. An opinion, not a fact.

They are labelled differently everywhere they appear. Curated public figures are explicitly marked as reported and not independently verified; provider appraisals have their own block; missing data stays missing rather than being invented.

Score: investor-score-v1.0
Valuation: market-consensus-v1.0

Both live in isolated modules. Evidence labels and version tags remain visible so a result can be traced when either model is recalibrated.