How to Name a Startup: A Phoneme-First Guide to the Complete Process

Most startup names are chosen the same way: a whiteboard session, a shortlist of five to ten options, a vote, and a gut-feel consensus. The name that wins is the one that creates the least friction in the room, which is a very different thing from the name that will work hardest in the market.

The problem is not creativity. Founders who run whiteboard sessions are not bad at thinking of names. The problem is evaluation. Without a framework for assessing why one name is stronger than another, the process defaults to aesthetic preference -- and aesthetic preference in a room of three people tells you nothing about how a name will land on the 10,000 people who encounter it in the first year.

This guide covers the complete naming process: from strategic brief to ranked shortlist, with the methodology behind each step. The process is what top naming consultancies charge $75,000+ to run. What follows is a structured version you can apply to your own project -- or hand to a naming engine that can run it computationally in under two hours.

Why the name is decided before the pitch

A name is processed in approximately 150 milliseconds -- before the listener has had time to consciously evaluate it. In that window, the brain extracts four things from the phoneme sequence: energy level (is this name active or still?), scale (does this feel large or intimate?), trustworthiness (does this feel familiar or foreign?), and category fit (what does this name remind me of?).

These impressions are not formed from meaning. They are formed from the acoustic properties of the sounds themselves. The hard "k" in Kroll signals precision and authority before anyone knows what Kroll does. The open "o" in Notion signals spaciousness and calm before anyone opens the app. The sibilant "s" in Stripe signals speed and fluidity before the first transaction.

This is the core insight that separates phoneme-first naming from brainstorming: the impression a name creates is encoded in its sounds, not its definition. Two names can have identical dictionary meanings and radically different brand effects. A naming process that does not evaluate phoneme properties is evaluating the wrong thing.

The five mistakes that sink most startup names

These are the patterns that appear most frequently in names that fail -- not immediately, but gradually, as the company grows and the name becomes a liability rather than an asset.

Category description instead of brand. "CloudSync," "DataStream," "QuickPay." These names communicate what the product does, which feels safe at the beginning. The problem is they cannot own the description. Every competitor can use the same words. The name will not differentiate in a crowded category, will not survive a pivot, and will age badly as the category matures and the terms become generic.

Cleverness that requires explanation. A name that is a pun, an acronym, or a forced portmanteau that requires decoding is a name that creates friction at every introduction. "So how do you spell that? And what does it mean?" is not the conversation you want to have at a conference. Names should be immediately statable, immediately writable, and immediately memorable. Cleverness that survives contact with a stranger who has no context is not the same as cleverness that delights the founding team.

Phoneme profiles that conflict with the brand promise. A company promising speed with a name that phonetically reads as slow. A company promising security with a name that sounds casual. A company targeting enterprise buyers with a name that sounds like a consumer app. The phoneme profile creates an expectation before the product is described. If that expectation is wrong, the product has to overcome it at every sales interaction.

No tension -- forgettable by design. The best brand names sit in a specific zone: they are more surprising than their category average (which makes them memorable) but not so surprising that they feel wrong for the category (which makes them feel risky). David Placek of Lexicon Branding calls this the tension zone. A name with no tension blends into the category. A name with too much tension feels like a liability. Most whiteboard names have too little tension -- they were optimized for consensus, which systematically selects against the unexpected.

Cross-language problems discovered late. A name that works in English may create significant problems in the markets where you will raise funding, hire, or sell within three years. The time to discover that your name has an unintended meaning in Mandarin or German is before you file the trademark, not after you have printed 10,000 business cards.

Before you generate a single name: the strategic brief

The most common reason naming projects fail is not that they cannot find a good name. It is that they cannot recognize a good name when they see one, because they have not defined what "good" means for their specific situation.

The strategic brief answers this before the first candidate is written. David Placek developed this methodology at Lexicon Branding -- the agency behind Pentium, Febreze, and PowerBook -- and it has been applied to over 3,000 naming engagements in the last 30 years. Four questions, in order:

How do you define winning? Not your business goal -- the specific reaction you want a first-time encounter to produce. A journalist writing about your funding round should feel what, exactly? A potential customer seeing your name in an ad should feel what? "Impressed by the technology" and "safe enough to try" are very different answers, and they produce very different phoneme targets.

What do you have to win? Your real competitive advantages right now: technology, team, data, relationships, regulatory clearance, timing. The name does not describe these directly -- it hints at them. A company with a proprietary data moat and a company with first-mover timing need names that carry different kinds of authority.

What do you need to win? The gap between where you are and where you need to be. If you need enterprise trust, the name needs to carry more institutional weight. If you need consumer adoption, it needs less friction. If you need to displace an incumbent, you need more tension. The name helps close the gap before the product gets the chance.

What do you need to say? The single idea the name must encode without explanation. One sentence a listener should feel before they know what the company does. This becomes the semantic anchor against which every candidate is evaluated. If you cannot state this in one sentence, the name brief is not ready.

These four answers, together, define the performance criteria for the name. Generation and scoring can then be calibrated against them rather than against vague preferences.

Step 1: Generate with adversarial volume

Step 2: Score against 14 psychoacoustic dimensions

The output is a composite score per candidate, ranked from highest to lowest. The top 20 are retained for the shortlist. Every name below rank 20 has a lower composite score than every name above it -- not a lower aesthetic score, a lower phoneme performance score across the dimensions that matter for this specific brief.

Step 3: Classify Brand Archetype per finalist

Step 4: Apply the five filters

• 01
  Phonetic fit test

  Say the name aloud in three contexts: introducing your company in a conference hallway, saying it over a bad phone connection, and saying it to someone who does not speak English as a first language. A name that passes all three is phonetically robust. A name that requires spelling, clarification, or repetition at any of these touchpoints has a real-world friction problem that composite score does not fully surface.
• 02
  Context performance test

  Place the name in three sentences: a WSJ headline announcing a funding round, a TechCrunch or vertical-press launch post, and a dotcom URL format (yourname.com, yournamehq.com). Names that work in all three contexts have broad context robustness. Names that work in one but look wrong in another reveal a positioning tension that should be resolved before registration.
• 03
  Cross-language screen

  Run the name through phoneme equivalents in the major markets you will target within three years. At minimum: Mandarin, Spanish, German, and Japanese. Each language maps the English phoneme sequence to its closest equivalent and evaluates for unintended meaning, pronunciation barrier, or cultural association problems. This screen catches the majority of international launch problems before they become launch problems.
• 04
  Tension zone test

  A name should feel more surprising than the category average without feeling wrong for the category. Score each finalist against the phoneme landscape of the 15 to 20 best-known names in your competitive set. The tension score measures how far the name sits from the category centroid. Below 0.3 is invisible -- it blends in. Above 0.75 is the optimal zone. Above 0.9 may be too dissonant for the category. Names that cluster with competitors fail this test even if they score well in isolation.
• 05
  Domain and IP baseline

  Before committing to any finalist, verify: (1) the exact-match .com is available or acquirable, (2) no exact-match trademark exists in your primary USPTO class, (3) the social handle is available on the two platforms most relevant to your go-to-market. This is not a legal clearance -- it is a baseline that eliminates names with obvious blockers before you engage a trademark attorney for the serious candidates.

Step 5: Name Construction analysis

Step 6: Name in Context rendering

The complete deliverable

When these six steps are run against 300 to 1,500+ candidates, the output is a ranked PDF proposal containing everything a founding team needs to make a defensible naming decision.

Flash or Studio: how to choose

If you are uncertain, Flash is the lower-risk starting point. The brief form gives you the same output structure as Studio -- just at 300 candidates with category-default scoring weights rather than Placek-calibrated weights. Many clients use Flash to orient on the phoneme landscape of their category, then return for Studio when they are ready to commit.

Going deeper by vertical

The core process above applies across all startup naming projects. But each industry has specific phoneme conventions, naming traps, and evaluation criteria that a generic brief will not surface. The following guides cover the vertical-specific layer in detail.

What the process cannot tell you

A rigorous naming process narrows the field from 300+ candidates to a shortlist of 20 with high phoneme performance, strong archetype alignment, cross-language safety, IP baseline clearance, and context robustness. It eliminates the names that would have failed and surfaces the ones that can succeed.

What it does not tell you is which finalist to choose. That final decision requires judgment about which name your founding team can say with conviction, which one belongs to the company you intend to build rather than the company you are today, and which one carries the story you want to tell for the next decade.

The process eliminates the bad options. Judgment selects the right one. Neither works without the other -- which is why naming projects that start with judgment and skip the process almost always produce names that feel right at the time and become problems later.