Your target audience isn't something you guess. It's something you reverse-engineer from the problem you solve, and from the exact words real people already use to describe that problem. Demographics come last, not first.
"I struggled in the beginning with my audience. If you take a look back it's kind of all over the place. I didn't know who I was trying to reach. When you start it will all feel strange." A marketer describing the exact starting point this guide is for.
This wasn't a one-off complaint. When we asked Hugo to research founder and marketer frustration directly (294 TikTok posts found, 64 validated as genuinely relevant), not knowing who to talk to came back as the most foundational, most recurring pain point in the entire set, ahead of anything about tools or budget. See the full run in we asked Hugo what founders hate about market research.
Start from language, not demographics
"Women 25 to 40 who care about wellness" describes millions of people with almost nothing in common. "People who've tried three different electrolyte brands and still feel bloated by 3pm" describes a specific person you could write an ad headline for, one they'd recognize as being about them.
To find that language, go to where people already complain about your category: reviews of competing products, forum threads, comment sections under category-relevant content. Look for the phrase that shows up again and again, in the customer's own words, not a marketer's paraphrase of it.
If you have no customers yet
You're not starting from zero. You're starting from the closest adjacent product's customers. If you're launching a new hydration brand, read reviews for the three or four brands closest to yours. The people leaving three-star reviews, the ones with a specific complaint instead of a generic rating, are your best early signal. They already want something like what you're building. They're just not fully satisfied yet.
Behavior beats demographics
Two people with identical age, income, and location can want completely different things from the same product category. One wants convenience, the other wants status, a third wants to solve a specific health concern. Targeting by age and income averages across all three and speaks clearly to none of them.
Behavior and stated need, what someone has already tried, what specifically didn't work, what they explicitly said they wanted next, is a sharper signal. It's also harder to gather by hand, which is exactly why most brands default to the easier, weaker demographic version.
Turn it into something usable
A target audience isn't done when you have a paragraph describing them. It's done when you can use it: to write copy that sounds like something they'd say themselves, to pick channels where they're already having the conversation, and to brief a creator on exactly who they're talking to. If your target audience description doesn't change how you'd write an ad, it's too vague to be useful yet.
This is also not a one-time exercise. Real language and real behavior shift, sometimes within a single product cycle. Treat your target audience description the way you'd treat a live dashboard, not a document you write once and file away. Hugo reads reviews, forums, and social conversation continuously, so this description can update as your actual audience does, instead of going stale six months after you wrote it.
A target audience is the exact words a specific, real group of people would use to describe their own problem. Everything else is a guess wearing a demographic costume.