How do you choose a niche and write an offer clients say yes to?

Every freelancer hears “you need to niche down,” and most hate the advice. It sounds like turning away work you need. But the reason to narrow isn't to earn less. It's that a stranger decides in seconds whether you're for them, and “I do marketing” gives them nothing to decide with.

In the companies I ran, we never hired the generalist agency. We hired the one whose website seemed to describe our exact problem. They weren't more talented. They were just easier to recognize.

What does a niche actually mean?

A niche is a group of people with the same problem that you can reach in a repeatable way. It usually has two parts: who (bookkeepers, dental practices, newly funded startups, wedding photographers) and what problem (invoices going out late, no follow-up after enquiries, a launch with nobody to run it).

Either part alone is weak. “I work with coaches” is broad enough to mean nothing. “I help coaches who've just launched a group program keep up with the admin and the inbox” makes a particular person sit up.

How do you choose your niche?

Start from evidence instead of imagination. Four steps:

1. List your best past clients or projects. Include jobs, volunteer work and favors, not just paid freelance work. For each one, write down the kind of business, what you did and what changed for them.
2. Mark the ones you enjoyed and did well. You're looking for the overlap of “I was good at this” and “I'd happily do it again.”
3. Find the pattern. It might be an industry, a size of business, a moment (just hired their first employee, just launched, just lost an assistant), or a type of problem.
4. Check you can find more of them. Is there somewhere these people gather, a list you can build, a group, a conference, a search you can run? A niche you can't reach isn't a niche, it's a wish.

If you're brand new and have no past clients, use the work you've done inside jobs. The eight years you spent in a dental office is a niche. So is the department you ran, or the software you know inside out.

How narrow should a niche be?

Narrow enough that your ideal client thinks “that's me,” and wide enough that there are hundreds of them you can reach. For most solo businesses that's something like “small accounting firms” rather than “small businesses,” and rarely as narrow as “accounting firms in one town with exactly four staff.”

Remember what a niche is for: it decides your marketing, not your morals. If someone outside it shows up with a project you want, take it. The niche is simply who your website, your posts and your outreach are written for.

How do you name the problem in their words?

This is where most positioning goes wrong. We describe our service (“inbox management,” “revenue operations,” “brand strategy”) and the client describes their pain (“I'm answering emails at 11pm,” “we have no idea which deals are real,” “our stuff looks homemade”). Sell to the second one.

Where to find their words: review sites, forum posts and community threads, questions in Facebook or Slack groups, the exact phrases past clients used on your calls, and comments on posts about your topic. Collect ten real sentences. Those are your headlines.

What is a positioning sentence?

One line you can say on a call or put at the top of your website:

I help [who] [get this result] [without this pain, or in this timeframe].

Some examples:

• “I help small accounting firms get their client onboarding down from three weeks to three days.”
• “I help coaches running group programs hand over the inbox and calendar in their first two weeks.”
• “I help dental practices stop losing new-patient enquiries that come in after hours.”

Say it out loud. If you wince, it's too clever or too vague. The test is whether a stranger can repeat it back to someone else, because that's how referrals travel.

How do you turn a niche into an offer?

A niche tells people who you're for. An offer gives them something to buy. Give it four things:

1. A name that says what it does, like “Inbox and calendar handover” or “30-day content reset.”
2. A result the client can picture, not a list of tasks.
3. A timeline, so they know when life gets better.
4. A price, ideally fixed for a clear first step, as in how to price your services.

Start with one offer. One clear thing you're known for beats five services nobody can keep straight. You can add the next one once this one sells.

How do you test a niche before you commit?

Don't rebuild your website first. Test it in conversations, which is faster and cheaper:

• Ten conversations. Use your sentence on ten people who fit: in messages, on calls, in groups. Watch for “how do you do that?” versus a polite nod.
• Listen for their words. If they explain your own offer back to you in better language, use theirs.
• Count the questions about price. People who ask what it costs are telling you the problem is real.
• Give it a few weeks. Changing your niche every fortnight is the same as having none. Only move on if nobody recognizes the problem after a genuine run at it.

How can AI help you find your niche?

AI is useful here as a pattern-spotter and an editor, not an oracle. It doesn't know your market, but it's very good at finding the overlap in a messy list and turning it into plain sentences. Start by dumping your history into it:

Then sharpen the sentence, with hard limits so it stays plain:

And package the offer so there's something concrete to buy:

What mistakes keep people stuck?

• Picking a niche you've never worked in because it looks lucrative. You'll have no stories, no vocabulary and no network there.
• Describing the service instead of the result. Buyers don't want consulting, they want the thing consulting gives them.
• Waiting until it's perfect. Ten conversations teach you more than a month of thinking.
• Changing it every month. Nobody can refer you if they can't remember what you do.
• Thinking a niche turns work away. It only decides who you speak to first.

Frequently asked questions

Once you know who you're for, go and find them: start with people who already know you, then write first messages that get replies. Or grab the 15 free client prompts.