It’s important to get feedback when you’ve got an idea for a business name. (Credit: Rawpixel on Unsplash)

When we started our company, we spent a while thinking about names. Actually, we agonized over names. Naming your startup is important. And it’s tough. It’s basically clean sheet creation wrapped in an optimization problem.

We thought we landed on a great name: KnowItOwl. Get it? It lends itself to a great owl mascot. It speaks to the fact that we’re taking a smarter approach to insurance. We got positive feedback when we tested it with people. And the .com domain was available!

But during our first few months of dealing with vendors and engaging our first users, it became clear that we would probably have to change the name. We were constantly spelling and explaining it (“it’s know it all, but with an owl”). We realized our clever name was too clever by half. Literally: Half the people loved the name and half the people didn’t. My guess is that our initial positive feedback was from too small a sample size. And we misjudged the practicality of a homophonic pun in a name.

So we set out to come up with a new, better name. We read up on all the best practices and pitfalls. And here’s what we learned and what we did.

Step 1: Generate the possibilities for your startup name

You have to come up with only one name that’s awesome. But to get there, you’ll have to come up with a lot of names that fall well short of that bar. [Actual names we generated: Safetynuts, Lifepit and Harmadillo. LOL]. Creative fatigue will tempt you to settle for options that merely aren’t terrible. But your startup is your baby and your baby deserves better.

So how do you avoid naming your startup “Safetynuts?” There are several options for generating a list of names. The key here is finding inspiration.

  • You can crowdsource it. There are several sites that use naming contests and the hive mind to generate company names (see Naming Force, Name Station, SquadHelp, Name Contests, Hatchwise, and WordLab). We didn’t do this but these guys did with good results.
  • You can do your own brainstorming. Force your own Archimedes-in-the-bathtub moment. This is what we did. This article summarizes a lot of the techniques to generate possible names (e.g., root words, word associations and permutations, foreign words, etc.). Our team kept a Google doc in which anyone could add ideas at any time, allowing inspiration to take its natural course. We also invited some creative friends to a structured brainstorming session that proved really fruitful.

Here’s the process we followed.

  • Give a brief overview of the business (e.g., target customers, our mission and values) to brainstorming participants. But not too much – you don’t want to constrain the thinking, just guide it.
  • Write a word on the whiteboard that’s related to the business.
  • Each person has 10 minutes to write 10 names on a sheet of paper. The word can be used in the name or it can just serve as inspiration.
  • After the 10 minutes are up, everyone passes their list to the person on their left. Then everyone has seven minutes to come up with seven new names, iterating on and/or drawing inspiration from their neighbor’s list. Do another 5-minute round.
  • You can repeat the exercise with a second word. Make sure it’s sufficiently different from the first word (e.g., if the first word was something concrete, the second word should be more abstract, like an emotion).
  • Voila, you have a list! (Our list included the eventual winner.)

Step 2: select the finalists for your startup name

By this point you have a Google doc with a long list of possible names. You have to narrow it down to your top five to ten to test. Here’s how we did it.

  • Get rid of the garbage. Subject each name to the “Really?” test. (E.g.: Safetynuts. Really? Can you imagine yourself talking to The Wall Street Journal about your company, Safetynuts? Really?)
  • Get rid of names that are too similar to other companies, especially competitors (even small-scale competitors). You don’t want to unintentionally drive search traffic to them.
  • Check for the unintentionally obscene or misleading. Look for words within the words or forced combos in a domain name. You don’t want to be the next Pen Island or Speed of Art.
  • Do a trademark search and eliminate anything that’s protected or could possibly infringe upon an existing trademark
  • Grade the remaining options with a nominology / onomastics test. Simple green/yellow/red heat mapping along these criteria is really insightful. More on how to do that in this great read.
  • Do a domain search and get rid of the ones that are truly off the table. This is the source of much angst, as the world wide web is really the wild, wild west. Domain squatting makes some otherwise good names prohibitively expensive. But everything’s negotiable and a taken domain might still be within reach. The first comment to this Fred Wilson post on finding and buying a domain name is required reading on the whole domain buying process.

Step 3: Test the finalists among your target consumers

Once you’ve narrowed down the possible names to a viable list of 5-10 candidates, run them through a final wringer. Test the finalists for memorability, spellability/hearability and emotional associations with a large random sample in your target demographic. We designed our test based on the process outlined in this great blog post. Some helpful nuts and bolts from the test we ran.

  • We tested eight names, including our original name (KnowItOwl).
  • We stopped testing at ~300 total responses (some of which were garbage responses) because the results at that point indicated a clear winner.
  • We paid, on average, $2 per response. The cost depends on the length of your survey and how tightly you define the demographics of the sample (i.e., the more specific your criteria, the more each response will cost). We used Mechanical Turk, SurveyMonkey and AYTM to field respondents to take our survey. We used SoundCloud to record the names to test hearability and spellability.

Here’s how we fared on the different survey platforms:

  • AYTM: We got the best quality responses through AYTM. Their user experience and analytic tools are also well-designed. Unfortunately, they don’t have the functionality to conduct the hearability & spellability tests (i.e., you can’t put audio files on the platform). So we did those tests exclusively on SurveyMonkey and Mechanical Turk.
  • SurveyMonkey: By far the most flexible tool to design and run a survey. When it came to using their survey service, we found them more expensive ($3/response) and the quality of responses lower than AYTM responses.
  • Mechanical Turk: For those on a tight budget, this is a great platform to field respondents. Our average cost was $0.50/response. But you will need to design your survey with another service (e.g., SurveyMonkey). Also, getting responses took longer compared to AYTM and SurveyMonkey.

And the winner is…

Policygenius. It was the clear winner (sorry, Safetynuts).

KnowItOwl is cute, but bombed the hearability and spellability test. Turns out homophonic puns are funny but don’t make the best business names.

So KnowItOwl is now Policygenius. And we’re really excited to continue this journey with our new name.

P.S. Last time we checked, safetynuts.com is available for $350, if you’re in the market.

A version of this story originally appeared on Policygenius.