39 applicants a week instead of 4, each 87% cheaper

Introduction

The sales network of a financial services company was looking for new advisers in Budapest and in a large regional city. The goal was not as many clicks as possible, but a working recruitment system that brings in people who fit the team, actually join, and stay for the long term.

So we measured every level of the funnel: what an applicant costs, whether they pick up the phone, whether they fit the team, whether they turn up to the interview, and whether it ends in a mutual yes.

Challenge

We started from scratch. There was no careers page, no tracking, no active ad account, no applicant handling and no recruitment content. The accounts and tracking were ready by the end of August, and the ads went live on 1 September.

The results of the first three weeks were not acceptable: an applicant cost HUF 48,050, and on average 0.4 applications came in a day. The data pointed to four causes:

• Generic message: the ads leaned on the training programme and said little about the actual role.
• Wrong searches: about 47% of the Google spend went on existing customers searching for account admin.
• Broken mobile form: 90% of applicants came from a phone, and the form did not fit on the screen.
• Stuck applicants: the alert system misfired and statuses were not updated.

On top of that, Meta puts job ads in a separate, stricter category: you cannot target by age, gender or district, and lookalike audiences are off the table. In recruitment, it is therefore the message and the image, not the targeting, that decide who applies.

Solution

In week four we changed direction: new messages, a new creative system and a weekly rhythm, with owners and deadlines. The daily budget stayed around HUF 25,000; what changed was the quality of the ads and the page.

Creative, one question at a time. Over seven rounds we tested 72 creative concepts with about 140 images. Each round answered one question: typography alone does not stop the scroll, an image without people feels impersonal, but a job-ad card showing the specific role works. Four approved directions remained, each clearly showing the role, the offer and the type of engagement.

Two ways to apply, measured separately. The Meta campaign sending people to the website brought pre-qualified, higher-quality applicants; the Meta lead-form campaign brought applicants faster and more cheaply. Budget goes where applicants make it to interview, not just where they are cheaper. On Google a separate recruitment campaign runs on clear job-seeker keywords only, with a 6% click-through rate.

A careers page built for mobile. We rewrote the copy over six rounds, introduced the managers, fixed the mobile form, and added a second form halfway down the page, because attention fades faster on a phone.

No applicant gets lost. An automated applicant sheet: instant assignment to managers, a 48-hour alert for applicants without a status, and duplicate filtering.

Owned channels alongside the ads. Three career articles in the knowledge base (about 4,900 words, based on an analysis of 11 competitors), two posts a week per platform, structured data, and a site that AI assistants can read easily too.

Channels and tools

Summary

In recruitment, the ad is only the first step. The improvement did not come from a bigger budget, but from everything working as one system, from the careers page to the manager alert, and from learning every week from real applicants. The client extended the pilot; the next goal is more applicants at today's cost, and the first new hires.