Your vacancy pages get traffic, and most of that traffic leaves without applying. You already paid for those visitors: through ads, through your site, through a job board. Retargeting is the only way to get a second shot at them. Below is how to set it up: which audiences, which windows, which message per touch, and what tracking sits underneath.

The arithmetic that justifies retargeting

Say your vacancy pages get 2,000 visitors a month between them and 3 percent reach the thank-you page. That is 60 applications and 1,940 people who looked at your vacancy, were interested, and did nothing. Those 1,940 are the warmest audience you will ever get: they live in your region, they were looking for this work, they clicked through.

Now say you reach that group four times over two weeks at an 8 euro CPM. That is 7,760 impressions, so roughly 62 euros of media budget. If 1 percent apply after all, you add 19 applications at just over 3 euros each. Even at 0.3 percent that is still six extra applications.

Note: these are assumptions used to show the arithmetic, not industry figures. Plug in your own numbers. The shape stays the same: the remainder is always much larger than the conversion, and reaching that remainder costs less than buying new cold traffic.

The five audiences to build separately

One audience called 'all website visitors' is not retargeting, it is a bucket. Build these five separately, because each one deserves a different message and a different window.

  1. Vacancy page viewers who never reached the thank-you page. Rule: URL contains /vacancy, minus everyone with the application event. Your base audience and usually your largest.
  2. Form starters who did not finish. The warmest group you have and often the smallest. You need a separate event for this, for example when step 1 is completed.
  3. Video viewers past a threshold. Use 25 percent or 15 seconds as your floor. Below that you are looking at scrolling behaviour, not interest.
  4. Past applicants who were never placed. Export them from your ATS as a customer list and use them at job family level, not at vacancy level.
  5. Placed flexworkers whose assignment is ending. Four to six weeks before the end date they are back on the market. Highest placement odds, lowest cost, and almost nobody builds it.

Windows: why 180 days is almost always wrong

The default is often 180 days. For a specific vacancy that is wrong: you keep advertising a job that was filled long ago, on intent that is five months old. Rule of thumb: your window is never longer than the average time your vacancy stays open. If a vacancy runs three weeks, 30 days is your ceiling.

  • Viewers of a specific vacancy page: 14 to 30 days.
  • Form starters: 7 to 14 days. This intent decays fastest.
  • Video viewers: 30 to 60 days. Softer signal, longer shelf life.
  • Past applicants from your ATS: 180 to 365 days, but at job family level and with a different message.
  • Ending assignments: not a window but a date: end date minus four weeks.

What your second and third ad should say differently

The most common mistake is showing the same ad again. If the first impression did not convince, the sixth impression of that same creative will not either. All you buy is irritation. Think in three touches that each do something different.

  1. Touch 1: the job itself. Role, location, hours, salary indication. The goal is the click through to the vacancy page.
  2. Touch 2, day 1 to 5: remove one objection. People drop off for a reason: uncertainty about pay, shift work, commute, contract type or the process itself. Pick one and answer it literally. 'Applying takes two minutes and you do not need a CV' is an ad, not a subclause.
  3. Touch 3, day 6 to 14: proof and a face. Your recruiter on camera, a day on the shop floor, or a lower bar: ask your question over WhatsApp, request a call back.

Every touch changes three things: the objection, the format and the bar in your call to action. If touch 3 still points at the same form people already abandoned at touch 1, the problem is not your ad. Go over your vacancy landing page first.

Frequency caps: the point where you become annoying

With a national audience, frequency spreads itself out. With a regional audience of 1,500 people it does not: there your daily budget decides directly how often someone sees your face.

Do the maths. Say: 1,500 people in the audience, an 8 euro CPM, and three impressions per week feels acceptable. That means 4,500 impressions a week, about 36 euros a week, so a little over 5 euros a day. Put 25 euros a day against that same audience and you buy nearly 15 impressions per person per week. That is the point where people in a small town start talking about you, and not favourably.

Formula: maximum daily budget = (audience size x desired impressions per week x CPM / 1,000) / 7. These are assumptions too, but the logic holds: with a small audience you derive your budget from your frequency, not the other way round. Set a cap in the platform as well, for example two impressions per seven days.

Audience too small: the regional agency problem

Ad platforms need a minimum audience size before they deliver at all. One vacancy page, one region, one month of traffic often does not clear that floor. You end up with an audience that never delivers and you conclude that retargeting does not work. The answer is to aggregate.

  • Merge vacancies from the same job family into one audience: all engineering, all logistics, all healthcare. The message then speaks to the family, not to vacancy number 4412. 'Engineering work in the Eindhoven region' works fine for someone who looked at a welder vacancy last week.
  • Extend the window for that bundled audience to 60 to 90 days. That is fine, because you are no longer advertising one vacancy.
  • Combine sources: site visitors, video viewers and people who engaged with your social profiles in one warm audience.
  • If you still do not clear the floor, stop running retargeting as a separate campaign and put that budget into your cold campaign and your landing page.

Work out what you currently pay per candidate before you switch on a third campaign.

Exclusions: stop advertising to people who already said yes

Little undermines your credibility faster than an ad for a vacancy someone applied to last week. Exclusions are not a detail, they are half of the setup.

  • Anyone with a thank-you page event: excluded from all retargeting on that vacancy for at least 90 days.
  • Placed candidates: as a customer list from your ATS, excluded from all candidate campaigns for as long as they are working.
  • Your own staff and recruiters: excluded, otherwise they distort your reach and your cost figures.
  • Candidates who withdrew or were rejected on hard criteria: excluded from that job family, not from everything.

These lists only work if they are refreshed weekly. Done by hand, that lasts three weeks and then stops. Connect your ATS to your campaigns so the export runs on its own.

The tracking that has to sit underneath

Without reliable signals you are building audiences out of noise. Four things need to be right before you put a euro into retargeting.

  1. Pixel plus server-side Conversions API. Browser-side only, you lose events to ad blockers, browser settings and in-app webviews. Send the same event server-side as well, with a shared event ID so it does not count twice.
  2. A real thank-you page on its own URL. No pop-up and no JavaScript state change. That URL fires one clear event, and that event is both your conversion and your exclusion trigger.
  3. A separate form-start event. Without it you cannot build the warmest audience in this article.
  4. Consent handled properly. Your audiences only fill up with visitors who consented to marketing cookies, so they are structurally smaller than your visitor numbers in analytics. Decide with your own privacy officer how you set this up and how long you keep the lists.

Conclusion

Retargeting in recruitment is not an extra channel, it is clearing up waste you already paid for. Build five audiences instead of one, keep your windows short, give touches 2 and 3 a different message from touch 1, derive your budget from your frequency, and exclude everyone who already said yes.

Start small: the thank-you page event and the first audience are an afternoon of work. If you want this running without spending months on it yourself, take a look at what we do for staffing and secondment agencies.