Naming in an AI Age

AI Naming: The Right Way to Use AI for Brand and Product Names

• The NameStormers • Season 3 • Episode 14

Share episode

In this episode of Naming in the AI Age, Mike Carr explains how AI can enhance naming when used strategically. He warns against basic prompts and generic testing, noting that AI needs brand-specific context. Carr introduces "context engineering"—feeding AI detailed inputs like brand strategy and audience insights—to generate better names. He covers prompt structuring, using advanced models like GPT-4o, and the continued importance of human intuition and critical thinking. While AI accelerates idea generation, naming experts remain key for legal checks, real-world testing, and team alignment.

Mike Carr (00:04):

Welcome back to another episode of Naming in the AI Age. And so today we're going to talk about why you should use AI or naming, which most naming consultants want to talk you out of. But I think there's some great reasons to use ai. I think there's some challenges with using ai. And so I want to take you through how we use it and some of the things that I think you should do if you're going to be using AI to name things, and I'm looking at some of my notes as I go just to remind me of all the topics that I want to cover. So anyway, one of the things that you need to be sure and think about is, well, AI isn't necessarily going to know our branding strategy or brand persona, and so is it really going to come up with some good names?

(00:49):

And then what about checking for legal availability, right? It comes up with some interesting names, but it's legally not available or linguistically it has a problem. Some language or I'm going to test the name with synthetic respondents, you cannot test the name with synthetic respondents. Synthetic respondents, to me, having come from Nielsen, the marketing research company many years ago is a disconnect. The whole point of testing is to test it with whom your consumers are, whom your target is, whom your customers are. And synthetic respondents are not real human beings. So please, whatever you do, don't use AI to test names with, but there are some other reasons to use it. Okay? So what I want to talk about is the right way, as I said, and what I believe is there are two types of prompting and there's been a lot of discussion, especially for folks who are new to ai, that you just enter in a simple prompt and you can get some pretty good results and you can't.

(01:41):

A simple prompt generally is not going to generate stuff that's very good. So here's an example. I need a name for a new non-alcoholic beverage. Let's say each beverage is formulated to taste as good as the alcoholic drink it mimics. So you're giving it a little bit of context there, whether that is a Tom Collins or a Tequila Sunrise. I want a name that conveys incredible authentic flavor that is almost indistinguishable from its alcoholic counterpart. Please give me 20 name candidates. So that's your prompt, right? You thought about it for maybe 10 seconds, right? Yeah. I'm talking about a name for this thing. Describe what it is and I want a name that sort of does this. Now come up with some names and sure enough, when you run it, it doesn't take AI very long to come back to you with a bunch of names that are sort of, nah, they just don't excite.

(02:33):

The examples that I came up with using chat GPT were names like Virtu Sip, simu Spirits, and my favorite bad name flavor Fidelity. I mean, I can't imagine going in and asking anyone for a flavor fidelity when you're in a busy bar and even on pack that name is going to take up a lot of real estate. It's quite a mouthful. There are many, many issues with that. So what is the answer here? And there's a new type of engineering that's being talked about, not prompt engineering, but context engineering. And this is what we do for clients. This is what you pay naming consultants to do is we spend a lot of time upfront with a brief and some folks poo poo the importance of a creative brief. It is essential. It really is important. And the reason it's important is it sort of sets the entire framework for the naming exercise.

(03:28):

And if you feed this thing into ai, it provides the context that is needed. So AI sort of understands all the things that you understand as a human being, as director of product innovation at your company as the insight manager company that often has to do research on new products as the brand manager, whatever your role is, sharing all that knowledge with AI is super important if you want to come up with a better name or a better set of names. So I'm going to talk to you a little bit about the prompt to use to sort of set the stage to set the context to engineer the context. So AI is going to do better. So there's three things. There's the basic stuff that are the nice to haves and then there's the creative guidance. So the basic stuff you need to include sort of the overall objective, short and succinct, what is it you're trying to do?

(04:21):

And of course the product description. What is the product? If you're naming your company, tell 'em a little bit about the company. What's key features and benefits? Talk about the target audience more than just demographics, more than just their age and their gender and actually more than just psychographics, but pain points, aspirations, what you want the name to do for them. What's their mindset? The more you give AI with respect to aspiration, thematic direction to go, the better it's going to get. I can't overemphasize that enough. We give it competitors on all our briefs and we ask it to go out and scan the competitors' websites and all the materials out there on the internet and it does a really good job. Now the reasoning models of the models that we use here, and this has come up in the last six months, these didn't use to exist, but they're unbelievable.

(05:14):

So like chat GPT zero three mini, and there's another version that's like mini or max or something, Gemini 2.5 Pro Quad four, Opus and Sonnet, whatever AI flavor you like, if you use the reasoning model and you ask it to do the deep research, it is going to give you a much better set of names. It also is going to digest all the information you give it like I've been talking about. So the overall objective, the product description, the target audience, the competitors, and ask it to go out and research the competitors and do an analysis and all that. The brand persona. So for this fictitious non-alcoholic beverage, you want a name and you want the brand to sort of reflect something that's hip, sort of contemporary with it. It's fun, it's playful and it's real, right? It's honest, it's trusted. And then brand strategy.

(06:04):

Is the name going to work alone? Do you have a master brand that needs to work with? Are there some sub-brands that you wanted to work alongside? Is it a family name, is it specific line item extension, all that kind of stuff. Then the nice to haves, the length and the style of the name. So shorter is generally better, but of course there's a trade off there. The shorter the name, the harder it is to get through legal. So often one to two words, three syllable max is sort of what we all give ai. Easy to pronounce in English, doesn't mean anything offensive in Spanish. The tone playful but not silly, right? So avoid puns, avoid something that might be misspelled or mispronounced. Anything that's too boring or anything that's too edgy. Or maybe you want something that's very edgy that's very out there. So try put all that stuff in there, right?

(06:48):

Legal and linguistic guidance screen for trademark conflicts in classes 32 and 31 or whatever the classes are that conflicting trademarks might be found in. And of course you can ask AI this, you describe the product and it'll come back and tell you the relevant international trademark classes to search in and then avoid the slang with different languages that you may be going for. If you're selling up into the Quebec, you would want Canadian French on that list. If in the US of course Spanish would be important. Creative guidance, key dimensions. So we want something that sounds authentic, it doesn't contain alcohol, so maybe that should be in the name that it's not a Tom Collins, it's not a sangria sunrise. It's the equivalent of that in terms of taste. Something that's going to taste just as good, just as flavorful as that. So all the fun without the hangover.

(07:36):

If you have some taglines in mind, throw them in there. If you have some brand positioning statements in mind, throw them in there. And then you want names that are metaphorical. What's the analog superheroes? If you see a few names where it positions this product line as a superhero because you get all that benefit of being able to go out and enjoy a good night out on the town with your friends. None of the problems with drinking and alcohol. It's your best friend, right? Something that you trust that you know isn't going to cause you to get any problems. And then examples of what you like and why the name Curious Elixirs sort of grabs my attention. I might have some stopping power or I really dislike a name like Phony Negroni. Phony Negroni, which was one of the names AI came up with by the way, when I prompted it earlier with the short prompt.

(08:21):

I don't think phony in the name for a product that's going to try to mimic the real taste of an alcoholic drink is of course appropriate. You probably don't either. And then what's your success criteria? What do you want AI to sort of rank itself on? What's the stopping power of the name? What names are going to generate that instant reaction to have that hook to stop the consumer? The perspective customer. How memorable is the name? That's sort of the golden rule of naming, right? They can't recall it. You've lost the battle. If it is memorable, if it's inherently memorable, you're off to the races. Now whether AI can really do a true assessment of that, it's still subject to debate, but that's something that certainly I would put in the context engineering that you give ai, is it unique and does it resonate emotionally?

(09:04):

Now, if you prompt AI with all of this and then you ask it in the prompt at the end might be something like generate the 20 names and generate the 30 names and rank 'em for distinctiveness. Okay? So it's going out there and checking on what's already out there for you and provide rationale for each name, what's your thinking behind that name? And maybe even a tagline, give me a distinctive tagline that sort of relates to that name and stays true to our brand persona. And then I want you to list the names in ranked order. I want you to list the names that are the most distinctive or they have the most emotional resonance at the top, and then list the rest of them in descending order and just show me the names in a table. And I want the following columns. I want the name, I want the rationale, your reasoning for the name.

(09:47):

I want your suggested tagline and maybe the overall rank. And then if names do have problems, if you think they're probably unavailable, list those separately. You may still want to see those, right? You may still want to see the names that AI comes up with, but you don't want them thrown in with everything else. And then you can repeat this process. Now, I just did this for another client earlier today and took me an hour maybe to go through all this and iterate through. And what I found found a couple of things I thought was really interesting. It came up with maybe two names out of the 50 or so it generated that I didn't have to change much and I thought were interesting enough to put on the list for our trademark screening. More importantly though, it came up with some words, some directions that our team hadn't considered.

(10:30):

And so then I had to turn AI off and really think through it. And it's that critical thinking, right? It's that human intuition that I had to bring to bear and came up with some more really good ideas. And that might be done solo, that might be done in conversation with others on your team, but it's this idea that AI can get you part of the way there, especially if you'd use some context engineering procedures and then you take it over the finish line yourself or with your team and this iterative process, you repeat this several times, you refine it, you better and better prompting, it gets better and better. So the more I use ai, we use it almost every day. Well actually we use it every day, but for different things, the more I realize you still need to have that human element in there, right?

(11:12):

It's not a replacement for that, but it gets you where you want to go quicker. It opens up a broader range of possibilities. It's not the end all solution. I don't think it's ever going to be replacement for real people and testing names for real people, but it's a great partner. It's a great creative aid. And I think the more you use ai, the better you get at it. That's one of the things I've noticed with some of our team members. They're folks now on our team that are far more proficient at doing this kind of stuff than I am because they're living and breath this and I just don't have time, right? I've got other things to do as you probably do too. So if you've got some folks on your team that are the AI gurus, I would just turn 'em loose. And now you can also be sure and set AI so that all the data you enter is private.

(11:55):

I know every AI LLM that we use, we set those settings to make sure that none of that information is shared. So if we're working on a confidential client project, we don't have to worry about any of that getting out. Then the last thing I want to leave you with is my sales pitch for naming consultants. Since I am a naming consultant, since name Stormers is a naming consultant, even if you're using AI and you feel pretty darn good with the results, the value that we can bring to the table is the final sort of, Hey, we've been doing this for 40 years. I personally work on thousands of these projects. Let me take a look and just give you an assessment of any issues. We can help you with the final legal due diligence before you hand it off to your attorney. We can do the linguistic checking, we can do the name testing with real people, real targeted customers and prospects, not synthetic respondents.

(12:44):

I have no idea what a synthetic respondent is or how it works, but I would be very hesitant to use that to try to get a clean read on what your market might say. So I think there are a lot. And then all the selling of the name internally, which is often one of the most discounted parts of what we bring to the table, but I think many of our clients, the end engagement field, probably the greatest value we provide is that we can help them push the name over the finish line and get buy-in from the top levels all the way down in a very positive, proactive way where everyone's pumped up and excited and enthusiastic versus a drawn out battle where people are screaming and getting upset at one another and everything else. And so the outcome might still be a name that you feel pretty good about, but you've alienated half the people on your team trying to get there. So we can help you do that too. So if you decide not to use it, I hope you find this podcast of interest, but if you do decide to use this, please give me a call on my mobile and that number is 512-917-6923. Have a great rest of your week. See you now. Bye.