Naming is mostly saying no

People imagine naming as a moment of inspiration: a word arrives and everyone in the room just knows. In our experience it is much closer to editing. A typical naming project at Orvel generates between 300 and 500 candidates and ends with one, which means that more than 99 per cent of the work is saying no. Doing that well, quickly and without bruising anyone depends on deciding how you will say no before you start.

Agree the filters first

Before anyone suggests a single name, we agree a set of filters with the client and write them on one page. Every candidate passes through them in the same order, cheapest first, so the expensive checks are spent on the fewest names. Most candidates fail the first two, which cost nothing but an honest minute. The language check removes more good names than any other step: on one project last year a shortlist of nine lost four in a single afternoon to our Portuguese, Polish and Turkish readers.

• Sound: can you spell it after hearing it once on a bad phone line, and say it without explaining it?
• Sense: does it belong to one of the territories we agreed, or is it simply a nice word?
• Sentence: does it survive being used, as in “I booked it through…” or “send it to…”?
• Languages: does it mean anything unfortunate in the markets on the plan? Native speakers, never a translation app.
• Availability: a usable domain or a sensible variant, and the main social handles.
• Law: a knock-out trademark search in the classes that matter, then a full search on the final few.

Separate making from judging

The commonest mistake is judging names while you are still generating them. Someone offers a word, someone else winces, and from then on the room plays safe. We split the two completely. For four or five days, three of us generate names separately and without comment, aiming for quantity: compound words, borrowed words, invented words, words from the client’s own history and archive. Only at the end of that week do we put the lists together and start cutting, and then only with the filters, never with “I just don’t like it”. If a name feels wrong, the job is to find which filter it fails. Surprisingly often it fails none, and the dislike turns out to be about a person or a product it reminds somebody of.

Generating names is easy. Agreeing why you are rejecting them is the real work.

Saying no to the client, kindly

Clients usually arrive with names of their own: a founder’s working title, a word a board member likes, the street where the company started. We put every one of them through the same filters as ours, in the same table, with no special treatment and no special hostility. Most do not survive, and watching a name fail on spelling or trademark grounds is far easier to accept than hearing a studio say it prefers its own ideas. Occasionally a client’s name survives everything and wins. We are always pleased when that happens, because it means the name already had a history before we arrived.

Know when to stop

The last no is to the search itself. There is always another round to run and another hundred words to try, and past a certain point more candidates make the decision worse, not better. We stop when we have between three and six names that pass every filter. Then we set each one in the places it will actually live, a shop sign, an app icon, a line on an invoice, a sentence said down the phone, and ask the decision makers to choose among names that would all work. Choosing between good options is a much calmer conversation than defending one, and the name that wins tends to be chosen for how it behaves rather than for how it looked on a slide.