Most naming projects fail for a boring reason. The team falls in love with a word before anyone agrees what the name has to do. Six weeks later there are two hundred options, three favourites, and no way to choose between them. A process prevents that, and it also produces a name that still makes sense when the business is twice the size.

Write the naming brief first

One page, written before anyone suggests a word. It should state what you sell, who buys it, the three markets you expect to enter, and the words you must avoid. Add the practical limits: a domain you can afford, a name that works in the countries you plan to trade in, and whether a trademark search is realistic.

The brief also needs a decision rule. Who chooses? How many rounds? What happens if nobody loves any option? Naming goes wrong most often at the approval stage, not the idea stage. Agree in advance that the final choice will be made by one named person, using the criteria, not by consensus.

Decide which kind of name you need

There are only a few families, and each carries a different cost.

  • Descriptive: says what the company does. Easy to understand, hard to protect, and it dates when the offer changes.
  • Evocative: suggests a feeling or image. Flexible, but needs marketing to attach meaning.
  • Invented: a made-up word. Distinctive and easy to trademark, but every customer needs the story.
  • Founder or place: warm and credible at small scale, awkward when the founder leaves or the company expands.
  • Compound: two real words joined. Memorable, though many good ones are already taken.

Pick two families to explore, not five. If you plan to sell the business one day, weight protectability higher. If you sell locally and rely on referrals, a plain descriptive name may earn more trust than a clever one.

Generate in large batches, badly

The first fifty ideas will be obvious and mostly unusable. That is the point: get the obvious ones out of the way. Work in batches of twenty with a time limit, and write everything down without judging. Use prompts to force range: a word from your customer's daily vocabulary, a verb, a place, a sound, a translation of a simple idea.

Run separate sessions with sales, support, and one customer. The vocabulary people use when they describe the problem is often better than the vocabulary the industry uses.

Shortlist against criteria, not taste

Take twenty candidates and score them on five things: says something true, easy to spell after hearing it once, easy to say aloud, works as a verb or adjective if needed, and legally plausible. Score each from one to five and resist weighting.

Then say each name in a sentence you will actually use. Call us when the lift breaks, ask for the manager at Bexley and Co. Write the domain and the email address. If the email looks like a typo waiting to happen, that is information.

A good name survives being said badly on a phone call with a bad signal.

Test with strangers, not friends

Friends are polite. Ask ten people who have never heard your pitch to write down what they think the company does after hearing the name once. Then ask them to spell it. Count the misses. A name that needs a sentence of explanation is not necessarily disqualified, but you need to know the cost before you commit.

Avoid testing the logo at the same time, or half the feedback will be about the font.

Before you fall in love, check the trademark register in your main market, the company register, and the domain. Search the name plus the word reviews to see what customers already think of it. Check that the name does not mean something unfortunate in your second language market, and that it does not collide with a competitor's product line.

This stage kills options, so run it on three names rather than one. A cheap search now is cheaper than a rebrand in year three.

Plan for the decade, not the launch

Ask how the name behaves when you add services. A name built around one product becomes a constraint the moment you sell two. Ask how it looks as a small logo, as a Twitter handle, and as a line on an invoice. Ask whether you would be comfortable saying it in a serious meeting about a data breach.

  • One-page brief approved by the decision maker
  • Two naming families chosen, not five
  • At least 150 raw ideas recorded
  • Shortlist of twenty scored on five criteria
  • Ten stranger tests with spelling checks
  • Trademark, company, and domain check on three finalists
  • Future-fit test against your three-year plan
  • Written rationale for the winner, kept on file

The rationale matters more than most teams expect. In two years someone will ask why the name was chosen, and a paragraph written now will save an afternoon of speculation, or a pointless rebrand.