Name Your Product (And Live With It)

Naming is one of the most over-stressed and under-thought decisions early-stage founders make. Founders spend weeks deliberating, ask 30 friends what they think, change the name twice in the first six months, and end up with a result no better than what they would have picked in two days of disciplined work.

Bad names hurt. Generic names disappear in search. Hard-to-spell names die in word-of-mouth. Names that conflict with bigger companies invite trademark trouble. But the most common failure mode is not picking a bad name — it is being indecisive long enough that the lack of a name blocks every other launch decision.

This is the playbook to land on a name that survives the next five years, do it in a defined window, and stop second-guessing.

Why Most Naming Processes Fail

Three patterns produce the typical bad outcome:

Choosing in committee. Crowd-sourced naming votes converge on names everyone tolerates and no one loves. The name no one objects to is rarely the name that earns attention.
Optimizing for the wrong constraints. "Available .com domain" is a real constraint, but most founders inflate it into a search criterion that filters out 90% of plausible names. In 2026, a .ai, .io, or .co is a perfectly fine answer; insisting on .com is a self-imposed handicap.
Treating naming as a marketing decision. Marketing decisions are reversible (pivot the messaging); naming is functionally permanent (you live with it for years). The right framing: name as a contract with future-you.

The version that works is constrained: a 5-day decision window, a clear rubric, a small decision group (1-3 people), and an explicit acceptance of "good enough" rather than perfect.

What You Are Building This Week

By the end of this work:

A clear set of constraints (the rubric you'll use to evaluate names)
A list of 30-50 candidate names generated against the rubric
A shortlist of 5-8 names that pass the rubric
A final choice, with the reasoning documented for future-you
The .domain registered, the @social-handles claimed, the basic trademark check completed
Permission to stop thinking about it

This guide pairs with ICP (the audience whose ear the name has to register on), Value Proposition (the name has to live next to the value prop without contradicting it), and Brand Voice (the name's tone is part of the brand voice).

1. Pick the Naming Pattern

Five naming patterns dominate AI / SaaS in 2026. Each has tradeoffs. Pick one upfront — narrows the search space dramatically.

I'm naming [product description]. My ICP is [audience]. My main competitor names are [list 3-5].

Help me pick a naming pattern from these five:

1. **Descriptive** — name describes what the product does (Mailchimp, FastWrite, GitHub Copilot)
   - Pros: instantly clear, easier SEO, shorter sales cycle on first explanation
   - Cons: less distinctive, harder to expand the product into new categories
   - Best for: focused single-feature products, especially in commodity categories

2. **Coined / made-up** — invented word, no inherent meaning (Stripe, Notion, Anthropic)
   - Pros: maximally distinctive, trademarkable, escapes search competition
   - Cons: requires more brand-building to load with meaning; harder to remember on first hear
   - Best for: products that aspire to define a category or expand beyond their initial scope

3. **Metaphorical** — borrows from a familiar object or concept (Slack, Sequoia, Tower)
   - Pros: memorable, evocative, easy to talk about, often easy to spell
   - Cons: must clear trademark + .com / .ai / .io conflicts; risk of stale or culture-bound metaphors
   - Best for: products with a single strong attribute the metaphor amplifies

4. **Acronym / initialism** — letters that stand for something (IBM, AWS, GPT)
   - Pros: short, professional, often available as a domain
   - Cons: dry, hard to differentiate, often forgettable
   - Best for: B2B-only products, technical infrastructure, where a less-marketable name still works

5. **Founder name + thing** — your name + a category word (Mailgun's founder pattern, Linear / Cursor / Lovable as compressed founder-feel names)
   - Pros: personal, trustworthy, easy to register
   - Cons: limits future products that don't fit the personal brand
   - Best for: founder-led products in early stage

For my specific product:
- Recommend ONE pattern with rationale
- For each pattern I rejected, name the specific reason (e.g., "Acronym is wrong because my buyer is non-technical")
- Tell me what naming patterns my competitors use — and whether I should match or contrast

Default if no strong reason: coined or metaphorical for products that may grow beyond a single feature; descriptive for hyper-focused single-feature tools.

The single most underrated rule: pick the pattern, then generate names within the pattern. Generating freely across all patterns produces a chaotic shortlist and decision paralysis. Constraint enables creativity.

2. Build the Rubric First

Before generating names, write the criteria you'll evaluate them against. The rubric prevents the late-stage emotional whiplash where one sounds great Tuesday and feels wrong Wednesday.

Build a 7-criterion rubric for evaluating my product name candidates.

For each criterion, define a 1-5 score:

1. **Clarity** — Does the name help a stranger guess what the product does within 5 seconds of hearing it? (5 = "yes, instantly"; 1 = "no, confusing")

2. **Memorability** — Could I tell a friend the name once and have them remember it 24h later? Test by saying the candidate name out loud to someone unfamiliar.

3. **Spell-ability** — Can someone hear the name and spell it correctly without seeing it written? (Lyft, Airbnb fail here; Stripe, Notion pass.) Critical for word-of-mouth distribution.

4. **Pronounceability** — Can someone read the name and pronounce it out loud without hesitation? Names that need correction in conversation lose momentum.

5. **Differentiation** — When I search "[name]" in Google, do I get a sea of unrelated results that bury me, or a clear signal? Names that already conflict with established products in adjacent categories tank long-term SEO.

6. **Domain + handle availability** — Is the .com / .ai / .io / .co available at reasonable cost (under $1k for a domain in 2026)? Are @[name] available on X / LinkedIn / GitHub / Discord?

7. **Trademark cleanliness** — Quick USPTO Trademark Search and EUIPO check: are there registered marks in the same class (typically International Class 9 for software / Class 42 for SaaS)? A "no" here is a hard cut, not a tradeoff.

Apply weights:
- Trademark cleanliness: hard cut (anything that fails this is out, regardless of other scores)
- Domain availability: 25% weight
- Memorability + Spell-ability + Pronounceability: 35% combined
- Clarity + Differentiation: 25% combined
- Plus 15% for the gut-feel "this fits the brand" check

Output: the rubric as a single 1-page doc I can score every candidate name against.

The trademark check matters more than founders think. A name that conflicts with an existing registered trademark in your class is a future cease-and-desist letter waiting. The check is free (USPTO TESS, EUIPO eSearch) and takes 10 minutes per candidate.

3. Generate 30-50 Candidates

The right way to generate is volume first, quality filter after. Most founders generate 5 candidates, fall in love with #2, and never run a real comparison. Force the volume.

Generate 30-50 candidate names for my product using my chosen [naming pattern from Section 1].

Generation prompts to use, run in batches:

Round 1: literal descriptive names (10-15 candidates)
- Combinations of the verb + object that captures my product's core action
- Examples for an AI content tool: WriteFast, ContentForge, DraftLab, PromptDesk, etc.

Round 2: pattern variants (10-15 candidates)
- For metaphorical: nouns from one specific domain (animals, tools, geography, mythology)
- For coined: 5-8 letter inventions, typically vowel-rich for pronounceability
- For founder + thing: my name combined with category words

Round 3: cross-language and -reference (5-10 candidates)
- Words from Latin, Greek, Japanese, Italian that capture a quality (e.g., "Linear" — clean, simple)
- Modified spellings of existing words (e.g., "Lyft," "Doordash")

Round 4: AI generation (5-10 candidates)
- Prompt Claude / GPT: "Generate 10 [naming pattern] names for [product description] that match these characteristics: [criteria from rubric]. Avoid: generic, hyphenated, longer than 8 letters, hard to spell."
- Cross-check AI suggestions against trademark and domain availability — AI tends to hallucinate availability.

For each candidate output:
- The name
- The naming pattern (descriptive / coined / metaphorical / acronym / founder)
- A 1-sentence rationale for why I included it

Don't pre-filter at this stage. Volume + diversity > quality. The rubric in Section 2 does the filtering after.

Output: a single list of 30-50 candidates, each on its own line, ready for evaluation.

The cross-language move is underused. Many of the best 2020s SaaS names ("Linear," "Notion," "Vercel") are mildly archaic English or pulled from another language. Adding a Latin-Greek-Italian-Japanese pass widens the search beyond the obvious English combinations everyone else has tried.

4. Filter Through the Rubric

Score every candidate against the rubric. Most will fail one criterion or another; the goal is to surface the ones that pass all criteria, not to pick a "winner."

Score every candidate from Section 3 against my 7-criterion rubric.

For each candidate, output:
- Name
- Trademark check status (pending / clear / conflict — quick search results)
- Domain availability + cost (rough check via Namecheap / Porkbun)
- @social-handle availability (X / LinkedIn / GitHub / Discord — quick check)
- Score on each of the 7 rubric criteria
- Composite score (with trademark as hard cut)

Surface:
- The 5-8 highest-scoring candidates (the shortlist)
- Reasons each lower-scoring candidate dropped
- Any candidate where I overrode the rubric (e.g., low memorability but a name I love anyway) — track this for honesty later

Important: trademark conflicts are HARD CUTS. A name that conflicts with a registered mark in my class is removed regardless of how much I like it. Founders who push past this lose 6-12 months of brand-building when the cease-and-desist arrives.

Output: the shortlist of 5-8 candidates with full scoring, ready for the next step.

The single most-skipped step: registering the actual social handles to confirm availability. Domain availability checks miss cases where someone owns @yourname on X but never built anything there — the handle becomes effectively unrecoverable, and you ship with @yourproductHQ awkwardness forever.

5. Run the Three Tests Real Buyers Use

A name that scores well on the rubric still has to survive contact with reality. Three tests every shortlisted name must pass:

For each shortlisted candidate, run these three tests:

**Test 1: The radio test (memorability)**
- Say the name out loud to 5 people unfamiliar with my product
- Wait 24 hours
- Ask them to recall the name
- Names that get remembered correctly by 4+ of 5 testers pass; below that fails
- This single test eliminates a surprising share of shortlist candidates

**Test 2: The category-positioning test**
- Tell 5 different ICP-matching strangers: "There's a new product called [name]. What do you think it does?"
- The guesses should cluster in the right category (e.g., "an email tool" for an email product)
- If guesses are wildly varied or wildly wrong, the name has clarity problems
- Run this online via Twitter polls, Reddit, or DM your existing customer base

**Test 3: The five-year test**
- Imagine my product has 100,000 customers, has expanded into adjacent categories, and is being acquired or going public in 5 years
- Does the name still make sense? Is it not embarrassing? Does it scale beyond the initial scope?
- Names that are too tightly scoped (e.g., a "PromptHQ" that needs to expand beyond prompt management) fail this test

For each shortlisted candidate, output:
- Test 1 result (recall rate %)
- Test 2 result (category-clarity rate %)
- Test 3 result (yes / no / depends — with the "depends" reasoning)
- Final pass / fail / reconsider

After running tests:
- Names that pass all three become finalists
- Names that fail any one are usually out (unless I have a very strong gut reason to override — track that override)

The radio test is the most predictive. Names that fail the recall test in casual conversation will fail in word-of-mouth distribution every time, regardless of how clever they look written down.

6. Pick (Decisively)

After tests, you'll typically have 2-3 finalists. The temptation is to keep narrowing, ask more people, run more tests. Resist. The marginal cost of more deliberation outweighs the marginal benefit.

Pick the final name from my finalists.

Decision rule:

1. If one finalist clearly outscores the others on the rubric AND passes all three real-buyer tests, that's the name. Move forward.

2. If two or more finalists are tied:
   - Which name am I most excited to write on a business card / type into Slack 50 times a day?
   - Which name will I be embarrassed to say at a conference five years from now?
   - When I write the candidate at the top of a draft pitch deck, which one feels like it belongs there?

3. If I genuinely can't decide between two:
   - Pick the one with better domain availability and cleaner trademark
   - Domain.com economics: a $5,000 domain that's perfect is worth it; a $50,000 domain that's perfect is rarely worth it; a $1,000 domain that's good enough beats a $20 domain that's bad

4. If none feel right:
   - I'm probably looking for a name that doesn't exist
   - Re-examine the naming pattern (Section 1) — maybe coined when I was searching descriptive
   - Run one more generation round with the new pattern, NOT another month of analysis

Document the decision:
- Name picked
- Why (one paragraph)
- The runner-up (in case I need to revisit)
- Date locked in (commit publicly to the team to prevent second-guessing)
- The trademark / domain registrations to complete this week

Output: a single-page "naming decision" doc that future-me can refer to in 18 months when someone says "should we have called it X?"

The decision-by-default approach: if you cannot decide between two names in 5 days of consideration, you do not have a meaningful difference between them. Pick the one with cleaner availability and move on.

7. Lock It In Within 7 Days

Once decided, register everything within a week. Delays are how good names get stolen by other people.

Within 7 days of picking the name:

1. **Register the domain** — primary (.com / .ai / .io / .co), plus the obvious typo-domains and the .com if you didn't get it (typos can redirect to the primary)
2. **Claim the social handles** — X, LinkedIn, GitHub, Discord, Reddit (even if you won't use it now). Name-squatters hit 24-48h after Product Hunt launches.
3. **Trademark filing** — for products you're confident in, file an Intent-to-Use trademark (~$250-$350 per class with the USPTO direct, $1,500 with a lawyer). Defensive, not optional past 100 customers.
4. **Email infrastructure** — set up @[name].com email + privacy@ + support@ + security@ aliases. Per [Onboarding Email Sequence](../../../VibeWeek/6-grow/onboarding-email-sequence-chat.md), proper deliverability requires the right sender domain.
5. **Update everywhere** — pitch deck, GitHub repo name, Stripe product name, internal docs. Tedious but compounds: every reference to the old working name is friction later.

If any of these is blocked (e.g., a critical handle is taken), now is the time to know — not 60 days into the launch when you've already committed.

Failure-mode drill:
- **Critical handle taken**: try +HQ, +app, +ai, get[name], use[name]. Note: founders sometimes recover a squatted handle by emailing the squatter — works occasionally for ~$500.
- **.com taken at unreasonable price**: take the .ai or .io. The 2010s "must have .com" rule is functionally retired in 2026 for tech products.
- **Trademark conflict surfaces during filing**: don't bury this. Get a lawyer involved immediately; sometimes the conflict is in a different class and resolves easily, sometimes it's a hard rebrand.

Output: a checklist I can complete in 1-2 working days max.

The 7-day lock-in window is the rule that prevents permanent indecision. Founders who give themselves "until launch to finalize the name" rebrand 3 times in 6 months. Founders who lock in 7 days after deciding rebrand zero times.

What Makes a Name Worth Rebranding

Rebrands cost weeks of engineering, marketing, and customer communication. Reasons that justify the cost:

Trademark conflict — cease-and-desist letter or risk of one. Nonnegotiable; rebrand fast.
Massive product shift — pivoted from B2C to B2B, current name doesn't fit. Rebrand is part of the pivot.
Hard-to-spell name causing measurable distribution loss — if word-of-mouth is half what it should be because nobody can spell the URL. Run an actual A/B before committing.

Reasons that do not justify a rebrand, even though founders often think they do:

You found a name you like better. No, you've spent a year staring at the current one and would have the same feeling about any other name in 12 months.
Some customers said the name is "weird." A small percentage of customers will say any name is weird. The metric is whether prospective customers can find, remember, and pronounce it — not whether they prefer something different.
A bigger company launched a product with a similar name. Annoying, but not a rebrand reason unless trademark conflicts are real (different classes are usually fine).

If you do rebrand: see Reduce Churn for the customer-communication discipline. Existing customers care more about the rebrand than new prospects do; the migration plan matters more than the new name.

Common Failure Modes

"I've been deliberating for 8 weeks." Naming is not a hard problem; you have made it one. Force a 5-day window. Pick a "good enough" finalist and ship. The name's value compounds; the indecision's cost compounds faster.

"Everyone I've polled likes [generic name] but I don't love it." Polled names converge on tolerable, not exceptional. Trust the rubric + tests over the popularity vote.

"I picked but the .com is $30k." Take the .ai / .io / .co. In 2026 the .com obsession is mostly nostalgia. The .ai is genuinely fine for AI products and often preferred.

"I rebranded twice in 9 months." The first name was probably fine; you re-litigated a closed decision. Lock in earlier with shorter deliberation windows.

"Trademark conflict surfaced after filing." Always run the USPTO TESS check (and EUIPO if Europe-bound) before locking in. Free and 10 minutes per candidate.

"My name has SEO problems — there's an existing brand using it." Should have caught this in Section 5's category-positioning test. Either rebrand now (cheaper than later) or accept the SEO penalty and lean harder on direct distribution channels.

Deliverable

A locked-in product name with the reasoning documented
A primary domain (.com, .ai, .io, or .co) registered
All major social handles claimed (X, LinkedIn, GitHub, Discord)
Trademark filing initiated (Intent-to-Use minimum) for the international class your product falls under
Email infrastructure set up at the new domain
A "naming decision" doc archived for future reference

What's Next

With the name locked in, move to Craft Your Value Proposition — the value prop now has a stable name to anchor on. Then Define Your Brand Voice — voice has to align with the name's tone. Then Write Your Landing Page Copy — every reference to the product carries the name forward.

Don't revisit the naming decision in Day 2 onward. The name is locked. Energy goes into the rest of the launch.