Engineering, healthcare, logistics, office support: you place all of it, and every new vacancy starts from scratch. New audience, new ads, new story for the client. Specialising promises to end that, but mostly it feels like giving away revenue. Below is a decision framework: what a niche buys you, what it costs, how to score candidate niches and how to test one for two quarters without risking your agency.

What specialising actually buys you

The case for a niche is not a marketing story, it is a cost story. When all your vacancies serve the same audience, you pay once to learn that audience instead of paying again every month.

  • Cheaper candidate acquisition. One audience, one set of creative, one landing page structure. What you learn on vacancy four makes vacancy twelve cheaper.
  • A pool that keeps working. Whoever you did not place on vacancy A is usable for vacancy B, because the profile is the same. At a generalist agency most of your response disappears unused into the archive.
  • Higher fees. You are not negotiating an hourly rate, you are selling access to a profile the client cannot find on their own.
  • Shorter sales cycles. A client does not have to judge whether you can do it; your proof is about exactly their role.
  • Referrals that compound. In a defined market people know each other; every good placement gets passed along.

Worked example: what one audience does to your budget

Say you have 2,000 euros of ad budget per month. As a generalist you split it across four audiences, so 500 euros each. Every campaign stays small and produces too little data to test anything. Put the same amount behind one audience and four times the volume runs through the same ads, which lets you seriously test opening line, imagery and form. Note: these are assumptions to show the mechanism, not industry figures.

What it costs you

Anyone who lists only the upside is selling you something. A niche brings three real risks and you have to price them in advance.

  1. Revenue concentration. One sector slowing down hits your whole agency. Agree a limit up front, for example: no single client accounts for more than a quarter of revenue, not even inside the niche.
  2. Cyclical exposure. Construction, transport and industry move differently from healthcare and education. Look at your own numbers from the past two years and see how demand behaved when things went against you.
  3. Saying no to work you can do. Every request outside the niche you take anyway costs you exactly the attention that makes the niche profitable.

Specialising costs revenue in the short term and produces margin in the medium term: work out how many months you can carry that dip.

Three axes: job family, sector, geography

You can narrow along three axes, and which one you pick matters a great deal.

  • Job family: electrical fitters, nurses, finance professionals, machine builders.
  • Sector: food, commercial construction, semiconductors, local government.
  • Geography: one province, the region around a single city, a forty kilometre radius.

The mistake many agencies make is narrowing on all three at once. Electrical fitters, in food manufacturing, in one province is not a niche but a handful of vacancies a year. Pick one axis sharply and stay broad on the other two. Electrical fitters, nationwide, in any sector: that is something you can scale in and where the same ads keep working.

Run a check before you choose: count how many vacancies are open right now inside your definition. Rule of thumb: if you are structurally below a hundred live vacancies, the definition is too narrow for a campaign engine.

Score your candidate niches on seven criteria

Put three or four candidate niches side by side in a spreadsheet and give each criterion a score from 1 to 5.

  1. Existing placements. How many did you make here in the past twelve months? Below five you have no proof, only an idea.
  2. Margin per placement. Not revenue but margin, after campaign costs and recruiter hours.
  3. Recurrence of demand. Does the same client ask for this every quarter, or is it a one-off fill? Recurring demand is what makes your pool valuable.
  4. How reachable the candidate is. Can you reach these people on paid channels by job title, interest or behaviour, and is the volume enough to get out of the learning phase?
  5. How reachable the buyer is. Do you know who signs (operations manager, HR, owner) and can you find and approach that person?
  6. Competitive density. How many agencies already position explicitly on this profile? Two is fine, twenty means you need to differentiate on something else.
  7. Cyclical sensitivity. What happens to this demand if interest rates rise, construction stalls or a funding pot runs dry?

Two knockouts: without at least five placements in twelve months and without a reachable candidate audience, you do not start. For the rest the total decides, not your preference. Our case studies show which definitions translate into volume.

Test for two quarters without betting the company

You do not have to switch over in order to learn. Set up a ring-fenced experiment alongside your existing revenue.

  1. Assign one recruiter and give that person at least half the week for the niche. Two people who also do a bit of it produces nothing.
  2. Set aside a fixed ad budget for six months and accept that the first six weeks are tuition.
  3. Build one landing page and one set of creative for the niche, not four.
  4. During that period, approach only clients inside the niche, with a proposition that names the role literally.

Measure four things separately from your regular revenue: cost per applicant, applicant to placement ratio, the time from first conversation to signed assignment, and how many candidates from the pool you reuse for a second vacancy. If you cannot see those four in isolation, your test is not measurable.

When to continue and when to stop

Define the outcomes in advance, otherwise you will talk your results into a success afterwards.

  • Continue: your cost per applicant falls, you place at least two candidates from the same pool on different vacancies and you close two new clients inside the niche.
  • Adjust: candidates come in, clients do not. Then the problem sits in your proposition or in how you defined the buyer, not in the niche.
  • Stop: after two quarters your cost per applicant has not fallen, your pool produces no reuse and conversations take just as long as before. Then you are not buying an advantage, only a smaller market.

What to do with clients outside the niche

Nobody cancels their revenue in one go. Sort your current clients into three groups.

  1. Keep. Profitable, low hassle, recurring demand. They stay, even outside the niche. They fund your transition.
  2. Wind down. Low margin, many hours, occasional requests. Do not renew agreements, do not take new requests and be straight about it.
  3. Refer out. Work you no longer do goes to a friendly agency, preferably with an agreement about referrals coming back.

Sequence: in the first quarter you cancel nothing and only get more selective about new requests. In the second quarter you stop selling actively outside the niche. Only once the niche has delivered revenue in two consecutive quarters do you really wind down group two.

Your positioning changes, and that is the point

  • Website: your homepage names the role, not flexible staffing solutions. A visitor should see within a couple of seconds that you solve their problem.
  • Outreach: your first sentence is about the role, not about your agency. We place electrical fitters in commercial construction is an opening line someone can say yes or no to.
  • Job ads: you write about the work itself, with details only someone from that world knows. That is the difference between ads people scroll past and ads they read.

The biggest gain is what you can do with signals. A generalist who sees a company post three vacancies still does not know whether that is relevant. With a sharp niche you judge every signal within a minute and answer it with a concrete offer. How to recognise those signals and turn them into a conversation is covered in the article on hiring intent signals.

Conclusion

Specialising is not a matter of belief but a calculation: pick one axis sharply, score your candidate niches on seven criteria, test for two quarters with stop rules fixed in advance and sort your clients into three groups so the transition funds itself. You leave work on the table that you could have done. What you buy back is cheaper candidate acquisition, a pool that keeps producing and shorter conversations.

Want to test a niche without losing a quarter to setup work? See how we build candidate and client acquisition for agencies, or book a call and we will work through three candidate niches with you.