You type your idea into a search box. Taken. You add a word. Taken. You try the shorter version, the plural, the one with the clever spelling. Taken, taken, parked since 2011 with a "make offer" button. Twenty minutes in, the obvious conclusion arrives: every good domain is gone.
It isn't. What's gone is the obvious corner of the namespace — and you have been searching almost exclusively inside it. The single most effective change you can make is not searching harder or later at night; it's changing the shape of the names you feed into the search. Availability is less a matter of luck than of construction: certain ways of building a name collide with millions of other people's brainstorms, and certain ways almost never do. This guide covers why the exhausted feeling is real but misleading, the patterns that put you in open territory, and how to check a batch of candidates properly so you stop falling in love with names that were never free.
Why it feels like everything is taken
Three mechanisms are stacked on top of each other, and only one of them is about scarcity.
You are searching a tiny slice. The .com zone is enormous, but the region you're probing is minuscule: common English dictionary words, two-word combinations of industry nouns, and short letter strings. Dictionary words in .com were effectively claimed decades ago. Every three-letter combination is registered — there are only 17,576 of them, and that ran out long before you had the idea. Four-letter strings went the same way. These aren't "good domains being taken faster than ever"; they're finite sets that were exhausted once, permanently.
Everyone brainstorms with the same inputs. Ask any group of founders in the same industry for name ideas and you get startlingly overlapping lists, because they're drawing on the same vocabulary of appealing words — bright, flow, forge, nova, spark, lift, north, ledger — and gluing them to the same category nouns. If a name occurred to you inside thirty seconds of thinking about your category, that's evidence it occurred to a few thousand other people too. The speed with which a name arrives is inversely correlated with its availability.
Registration is cheap and renewal is passive. A .com costs roughly the price of a takeaway meal per year, so a name can sit locked up for a decade because someone registered it for a project they abandoned in month one. Most of what blocks you isn't a competitor; it's inertia on somebody's auto-renew — which is also why "nothing loads at that address" tells you nothing about whether it's registered.
The useful reframe: you're not fishing in an empty lake. You're fishing in the ten metres of shoreline where everybody stands.
Search where the supply actually is
Available names are concentrated in constructions that don't arise from ordinary brainstorming. Here are the patterns, ordered by how reliably they yield registerable .com names — with what each one costs you, because none of them are free.
1. Unexpected compounds
Two concrete, ordinary words that have no business being next to each other: Ridgequill, Tallowfin, Grovemint. This is the highest-yield pattern by a wide margin, because the combinatorial space is vast and human brainstorming barely touches it — people naturally pair related words, which is exactly the exhausted region.
Trade-off: the name says nothing about what you do, so your positioning has to be carried by the tagline and the product. That's a real marketing cost, though a smaller one than founders fear — most memorable brand names are semantically arbitrary.
2. Morpheme blends and clipped roots
Take a root that carries a feeling and truncate or fuse it: lumen → Lumeo, sever/severin → Severic. You inherit a whisper of meaning without using a word anyone can register a claim on.
Trade-off: these can drift into the generic startup-sound if the root is fashionable (endless names built from nova, aura, zen). Pick roots that are unfashionable, and say the result out loud before committing.
3. Fully invented phonetic names
Built from syllables chosen purely for sound: Vandiro, Kaleff, Obrium. Availability here is the highest of any pattern, since you're generating strings that have never existed.
Trade-off: the heaviest one. A name with no semantic hook has to be taught to every customer, which costs marketing spend or time — worth it for a company built to last a decade, overkill for a side project. Keep them short and phonetically obvious, or you'll spend your life spelling them out loud.
4. Verb-led and phrase names
Sendbloom, Getfathom, Trythicket. Prefixing a verb pulls you out of the noun-collision zone and reads as a product action.
Trade-off: get- and try- prefixes have been used enough to read as a stopgap, and a verb prefix on a taken name signals that you accepted second place. Better when the verb is genuinely part of the brand than when it's a workaround.
5. Vocabulary from adjacent fields
Sailing, geology, bookbinding, falconry, baking — trades with rich concrete nouns nobody in your industry is mining. Coppice, marlin, fathom type words.
Trade-off: the bare single word is nearly always taken, so budget for compounding it with a second word from the same register.
Two patterns to be careful with. Deliberate misspellings (Kwik, Flexx, Lyfe) do free up availability, but every misspelling is a permanent tax: you will dictate that spelling forever, and people who hear the name on a podcast will type the correct spelling and land somewhere else. Hyphens and inserted filler words solve availability on paper and fail in the real world, where nobody remembers where the hyphen went.
Check in batches, not one at a time
The reason the search feels demoralizing is usually procedural: people check one name, feel the sting, and go back to thinking. That loop maximizes attachment and minimizes throughput.
Invert it. Generate thirty to forty candidates across at least three of the patterns above without checking anything. Then check them all in one pass, recording four columns: name, result state, renewal price, date checked. Only then look at what survived.
This matters because "available" is not one answer — a check can come back available at standard price, available but premium-priced, taken but parked, taken and listed for sale, taken and genuinely in use, or simply unknown when a lookup fails. Reading all six as a binary is how founders overpay or abandon winnable names. The mechanics of that — WHOIS versus RDAP, why a registry lookup is authoritative and a blank webpage means nothing — are covered in our guide to checking domain availability. Run that knowledge over a batch, not over one name you already love.
A generator with live availability attached to every candidate collapses this loop entirely: the names arrive already resolved against the registry, so the check stops being a separate chore and becomes a property of each idea. That's the specific problem Whelex was built for.
Disclosure: Whelex earns a registrar affiliate commission when readers register a domain through its links. That never changes which names we suggest or what availability we show.
What an "available" result does and doesn't buy you
Two limits worth stating plainly before you get attached to a survivor.
Availability is point-in-time. A result is a snapshot of the instant you checked, and nothing about checking reserves anything. A name that is free this morning can belong to a stranger this afternoon — short, pleasing constructions occur to more than one person. So: don't rush the deciding, but don't dawdle after you've decided. Registering your top two finalists for a year while you make the final call is a cheap hedge, not extravagance.
Availability is not permission. A domain can register cleanly while the name collides with an existing trademark in your sector. A registration receipt is not clearance. Search the official trademark registries for your markets for signals, and get a professional check before you invest seriously in a name — no domain tool, this one included, can tell you a name is legally safe, and none of this is legal advice.
A pass you can run this afternoon
- Write down the feeling you want the name to carry — three adjectives, not a description of the product.
- Generate ten candidates each from unexpected compounds, morpheme blends, and invented phonetics. Do not evaluate while generating.
- Cut anything you cannot say once and have someone spell back correctly.
- Check the survivors in one batch. Record state, renewal price, and checked-at date.
- Take the names that came back available at standard price and only now judge them on brandability.
- Run a trademark-signal search on your top two, then register promptly.
If step 5 leaves you with nothing, the problem is almost always step 2: not enough candidates, or all of them from the same pattern. Go wider before you go looking for a different TLD.
FAQ
How do I find a domain name that isn't taken?
Change the construction, not the effort. Feed the search unexpected two-word compounds, clipped or blended roots, and invented phonetic names rather than dictionary words and industry noun pairs — those regions of the namespace were exhausted long ago. Generate a batch of thirty or so first, then check them all in one pass.
Are all the good .com domains really gone?
The obvious ones are: single dictionary words, all three- and four-letter strings, and most common two-word industry combinations. The namespace as a whole is nowhere near exhausted, because compound and invented names are combinatorially vast. What is scarce is the exact category-descriptive name you first thought of.
How many names should I check before I settle?
As a rule of thumb, generate thirty to forty candidates before checking any of them, and expect a healthy majority to come back taken even with good construction. Attachment before checking is what makes the process painful; volume before judgment is what makes it fast.
Does searching for a domain make it more likely someone takes it?
Registry-level lookups are the low-risk way to check, registrar policies prohibit front-running, and documented cases are old. The practical advice holds regardless of the theory: once you've genuinely decided on a name, register it rather than checking it repeatedly and hoping.
Stop searching the crowded shoreline
The founders who find good available domains aren't luckier — they're generating in a different part of the namespace and checking in batches instead of one heartbreak at a time. You can run that loop by hand with a notebook and a registrar tab, and it works.
Or skip to the end of it: describe your idea to Whelex, get ranked brandable candidates built from exactly these patterns, and see the live .com state on every one before you get attached to any of them. Generate business name ideas with live availability at whelex.com.
