Help & FAQ
How Domain Name Finder works, start to finish.
Domain Name Finder helps you discover available domain names by combining your own lists of words and checking each combination for availability. The typical flow is:
- Keywords — build lists of words (prefixes, suffixes, brand terms, industry terms, etc.).
- Builder — pick two lists and a set of TLDs
(
.com,.io, …). The engine generates every combination (the "Cartesian product") and checks each one for availability. - Searches — each run is saved so you can revisit its results, re-check, or export them.
- Tools — other generators (typos, hyphenated, pronounceable, thesaurus, etc.) for when a straight combination isn't what you want.
- Favorites — bookmark the names you like.
Tools take one word (or list) and transform it a different way each — typos, mangling, a thesaurus swap, and more. Looking to cross two lists instead? That's the Domain Builder.
Type in a word you're considering and see how it stacks up — live-checks availability across today's top-weighted TLDs and shows the same Domain Score, broken down point by point, that you'd see anywhere else in the app.
Set on one name and want it somewhere? Type the word, pick a whole family of extensions — Tech, Business, Country, whichever — and every combination gets checked in one go.
Paste a list of words or domains and check them all at once. Each line becomes one name — full domains like example.com are trimmed to their stem and crossed with whichever TLDs you tick, so one paste can test .com, .net and .io in a single run. Example: paste 50 name ideas and get an availability verdict on every one, with filters and export like any other search.
Generate every possible short letter combination of a fixed length (3–6 characters). This is exhaustive, not random — the full a–z space in order, which grows fast (3 letters = 17,576 stems; 6 letters = 308 million, generated in capped batches). Example: length 3 walks aaa, aab, aac… across your TLDs — the classic way to hunt for a free three-letter name.
Generate random, pronounceable made-up names using vowel/consonant patterns. Templates like CVC or CVCV are filled with random letters (awkward consonants such as q and x are left out), so results sound like brandable words instead of letter soup — every run produces a fresh set. Example: the CVCV pattern can yield names like zavo, redu or malo.
Turn a word into leetspeak-style variants (to→2, o→0, i→1, …). Substitution rules — letters-to-numbers, phonetic respellings, vowel-dropping — are applied one at a time and chained up to three deep, ranked by how readable the result stays. Example: cloudbase → cl0udbase, kloudbase, cloudb4se, or the chained cl0udb4se.
Swap out each letter of a word for every other letter, one position at a time. Single-position swaps give word-length × 25 variants, and the optional two-position mode combines pairs of positions for much deeper (capped) coverage. Example: brand → arand, crand … brend, brind — every single-letter respelling of the word.
Generate common keyboard-adjacency typos for a word. Covers the four classic fat-finger categories: adjacent-key substitution on a QWERTY layout, swapped neighbouring characters, a dropped character, and a doubled character. Example: shop → ahop (adjacent key), shpo (swap), sho (dropped), shoop (doubled).
Insert hyphens between word pieces in every combination. With N pieces there are N−1 gaps, and every hyphen/no-hyphen pattern across them is generated — 2^(N−1) variants, with no leading, trailing or doubled hyphens possible by construction. Example: blue + water + host → bluewaterhost, blue-waterhost, bluewater-host and blue-water-host.
Expand up to 3 words into their synonyms and combine them. Each word can be expanded or kept as-is; expanded words pull synonyms from the built-in thesaurus (no external API) and the resulting variant lists are crossed in your word order. Example: fast + host with fast expanded → fasthost, quickhost, rapidhost, swifthost…
Permute a word set, or mix-and-match words from multiple-choice slots. Permutation mode tries every ordering of one word set; multiple-choice mode picks one word from each slot in fixed slot order (a cross-product). Example: slots [get, my] + [cloud] + [app, hub] → getcloudapp, getcloudhub, mycloudapp, mycloudhub.
Combines 2 to 6 typed-in words: generates every SUBSET of them (pairs, triples, etc.) in every order, not just the full set.
Combine a core word with a set of modifier words. Pick a built-in theme (cool, business, premium, tech or startup) whose curated modifiers attach before or after your word — or use any of your own word lists, with every list word tried on both sides. Example: core word anchor + the tech theme → getanchor, anchorhub, anchorify.
Build domains using starts-with, ends-with, contains, excludes, length, or ?-wildcard pattern rules. Set any mix of constraints and every name satisfying all of them is generated, capped at 10,000 per search — estimates shown up front are an upper bound. Example: pattern pay?? → every five-letter name starting with pay (payaa … payzz).
What happens to what you search for. Short version: nothing is published unless you press Share.
How often each registry actually gives us a definitive answer, from our own lookups. Also available as JSON.
Public DNS resolvers — addresses, filtering policy and DoH support. Download as CSV, JSON or XML.
How names are scored, and how to change the rules.
The document with legal force. Privacy above is a plain-language statement of practice, not a substitute.
A named collection of words you control — for
example a brand list (aarons,
cloud) and a suffix list
(hub, hq, labs). The
Builder pairs two lists together.
It takes the left list and the right list and produces every left+right combination, once per selected TLD. Two 100-word lists across 3 TLDs = 100 × 100 × 3 = 30,000 names to check. The estimated count is shown before you run, so you don't accidentally generate millions.
Each generated name is then checked for availability via DNS or RDAP.
No. Your searches are yours. Nothing you search is published, sold, or shown to another user unless you press Share on that search. We don't register domains, so we have nothing to gain from the names you find.
Sharing is opt-in per search, and it is a deliberate click — a search is not reachable by its ID until you make it so. Each search's own results page shows whether it has been shared.
Two things are public on purpose, and it is better to say so here than to let you discover them later:
- A search you Share gets a public, search-engine-indexable link — that is what the button is for. It publishes the recipe too: the search's name and which Favorite Keywords lists it combined, so someone can reproduce it.
- Globally shared Favorite Keywords lists are public and browsable by design. Favorite Keywords lists you create are private to you unless they are shared globally.
No. A search is a snapshot taken when it runs. Adding a word to a list afterward does not change, break, or appear in any existing search.
To include newly added words, use Run Again with These Settings on that search (or start a brand-new search) — either one regenerates combinations from the lists as they stand right now.
Recheck All re-tests the availability of the names already in that search — useful because a domain can be registered or dropped over time. It does not regenerate combinations and does not pick up newly added words.
Run Again with These Settings queues a brand-new search using the same recipe, regenerating the whole combination set from the lists as they are right now — so it does pick up anything you've added since.
A "header row" is a first line naming the column (like
word) rather than data. The toggle is on both the
Import and Export panels:
- Import — tick "First row is a header" only if your file's first line is a column name. If your CSV starts straight into data (no column name), leave it off so the first line is imported as a word.
- Export — tick "Include header row" to write
a
word/domainfirst line; otherwise the file is pure data. - Domain columns — 1 or 2 — on domain
exports, choose how the domain is laid out:
- 1 —
cloudfast.com, the whole domain in one cell. Best for pasting straight into a registrar's bulk-availability box. - 2 —
cloudfast,com, the name and the extension split apart. Best for sorting or grouping by TLD in a spreadsheet.
- 1 —
Your last choice is remembered as your default — for the header row and the column layout alike, across search exports and Favorites. Invisible byte-order-marks (which Excel sometimes adds) are stripped automatically on import.
On a list's edit page you can: type one word and
press Enter; click "Paste many words at once" and paste a
whole column or comma-separated list; or use Import
to upload a .txt/.csv file. Duplicates
and invalid characters are handled for you in every case.
DNS and RDAP are the two you can pick from for a search. WHOIS runs behind the scenes, not as a search-wide option — see below for why.
RDAP (recommended, and the default) is the official successor to WHOIS — it queries the registry directly. More accurate on average, but slower (~300ms per domain) and rate-limited by some TLDs. It is not infallible: a small number of registries have been caught answering differently depending on which server asks, well-formed but wrong. This app keeps a per-TLD trust list built from real evidence and quietly relies on DNS instead for any TLD where RDAP has been proven unreliable, rather than trusting every RDAP answer equally.
DNS Check is faster and has no rate limit. It checks whether the domain resolves, which is a reliable proxy for "registered" — but technically a domain can be registered without DNS records, and a handful of registries answer every unregistered name identically (a parking placeholder) rather than with a real "no such domain." This app detects and corrects for that per TLD too. Best for large jobs.
WHOIS is the original protocol both of the above effectively replace — plain text, not structured data, and the slowest and most commonly IP-blocked of the three. It is used as a secondary check (the Advanced Availability Report's diagnostic pass, and a fallback when DNS and RDAP disagree) rather than offered for bulk searches, where its blocking and speed would make large jobs unreliable.