Recruiting on Meta Ads: 3,848 applications at $3.90 in five months

The problem

The client is an employer who needs people in an office. Paid social was already running: no complaints about it, but no sense that the channel was finished either. So they brought in a second buyer — approaches differ, and results can move in either direction.

Then the niche does what the niche does. Only a minority of people who submit an application reach a final interview, and fewer than that start work. Which means the brief is not "get leads cheaper". It is "keep the top of the funnel full enough that the hiring team closes its plan".

Two conditions we took on:

• steady volume, without the gaps between weeks that leave recruiters idle;
• a cost per application low enough that after every drop-off stage, the cost of an interview held still lands where the client considers it normal.

What we did

The funnel

We built on a lead form — in this niche it usually performs best. We tested that rather than assumed it: a landing page and direct contact with a manager ran alongside. In this particular setup both did worse, so the budget went to the lead form.

A lead-form application is colder than one from a landing page where the person has read the terms. In practice, with decent handling on the client's side, more cold applications beat fewer warm ones at a higher price. Their team could take the volume.

Campaign structure

• CBO, 1/5/1 and 1/3/1 — one creative per ad set, expensive ad sets switched off as the data comes in.
• Broad audience, 20–45, no language narrowing: narrowing by language cuts out everyone whose device is set to something other than their own language, and those are real candidates.
• Placements — Instagram and Facebook.
• 1/10/1 and 1/3/2 were tested too. Here they performed worse, so we stayed on 1/5/1 and 1/3/1.

Creative

UGC, static and video all went into testing. Static produced the cheapest conversions — and burned out fastest, so creative was replaced often. That is not a launch-week job but a standing line of work: while one set is running, the next one is already being made.

What happened

Five months, $15,000 spent, 3,848 applications. Average cost per application $3.90; an interview held cost up to $12.

Cost per hire was still being counted on the client's side when this was written, and we will add it when it lands. Putting a flattering number here that we do not have would defeat the point of publishing the case.

What carries into the next hiring project: the funnel is chosen by a test, not by habit; broad beats hand-narrowed targeting in this niche; static wins on price but only if someone is replacing it every week.