Modeling Your Recruitment Funnel Backwards: From 12 Placements to a Weekly Budget

Almost every agency budgets forward: there's money, it goes into campaigns, and afterwards you look at what it produced. Working backwards is better. You start from the number of placements you have to hit and work back to the weekly budget and capacity required. Below: the KPIs that matter, a full worked example, and the two traps that cost the most.

Working backwards instead of budgeting forward

Budgeting forward gives an outcome you can't steer on: you only know afterwards whether it was enough, and by then the quarter is over. Working backwards gives a weekly standard per step. If you need fifty applications a week, you can see in week three that you're behind - and where.

Second benefit: your assumptions become visible. A model forces a number onto every conversion. Those numbers start as estimates - fine, as long as you treat them that way and replace them with measured values after a quarter.

The steps your funnel actually has

Most agencies model three steps: ad, application, placement. Too coarse, because everything that goes wrong sits in the middle. Work with the steps you can measure:

1. Impression to click
2. Click to application
3. Application to qualified candidate
4. Qualified candidate to scheduled intake
5. Scheduled intake to completed intake (your no-show)
6. Completed intake to submission to the client
7. Submission to placement

On the client side a second chain runs: prospect, meeting, assignment, fillable assignment. Those two have to stay in proportion, or you're recruiting for vacancies you don't have.

The KPIs that matter, and what they hide

Every KPI below is useful and misleading at once. Never read them in isolation.

• Cost per application - drops as soon as you broaden targeting or shorten your form: worse candidates for less money. Read it together with your qualification rate.
• Cost per qualified candidate - hangs on your definition of 'qualified'. If that's a recruiter's judgment, it drifts with the pressure on the pipeline. Put the criteria in writing.
• Cost per placement - commercially the only number that counts, and also the slowest. At low volumes one placement moves it by tens of percent. Report it quarterly, not weekly.
• Application-to-intake ratio - read as candidate quality, but mostly it measures your own follow-up speed.
• No-show ratio - largely a function of response speed, confirmation and expectation setting, not of candidates.
• Time-to-first-contact - your best leading indicator: measurable daily and predictive. Measure it in hours, on the median; the average gets wrecked by outliers.
• Time-to-fill - incomparable until you fix when the clock starts: at assignment intake, campaign launch or first submission. Pick one definition.
• Candidate-side versus client-side ratio - per job family: how many placeable candidates against how many fillable assignments? If the shortage is on the assignment side, a signal-driven prospect list beats more ad budget.

Worked example: from 12 placements back to a weekly budget

All the numbers below are assumptions used to show the method, not benchmarks. Replace them with your own measured values.

Say: 12 placements this quarter, campaign live for 12 weeks with one week of lead-in. Your assumptions per step:

• Submission to placement: 40%
• Completed intake to submission: 50%
• No-show on intakes: 20% (so 80% show up)
• Qualified candidate to scheduled intake: 50%
• Application to qualified: 25%
• Click to application: 8%
• Cost per click: €0.80

Working backwards:

1. 12 placements ÷ 0.40 = 30 submissions
2. 30 ÷ 0.50 = 60 completed intakes
3. 60 ÷ 0.80 = 75 scheduled intakes
4. 75 ÷ 0.50 = 150 qualified candidates
5. 150 ÷ 0.25 = 600 applications
6. 600 ÷ 0.08 = 7,500 clicks
7. 7,500 × €0.80 = €6,000 media budget

€6,000 over 12 weeks is €500 a week. That gives you your derived KPIs: €10 per application (6,000 ÷ 600), €40 per qualified candidate (6,000 ÷ 150) and €500 in media cost per placement (6,000 ÷ 12). That last one excludes salaries and overhead: media cost, not cost price.

And the capacity that comes with it

The budget is the easy half. 600 applications over 12 weeks is 50 a week someone has to screen; figure eight minutes each and you're at just over six and a half hours. On top of that 75 scheduled intakes, just over six a week, of which five go ahead: with preparation, roughly six hours. Together thirteen hours a week, without any client contact. A plan that asks for €500 a week and a day and a half of recruiter capacity is a different plan than it looks.

Sensitivity: one ratio in the middle

If your qualification rate slips from 25% to 20%, you need 150 ÷ 0.20 = 750 applications, 750 ÷ 0.08 = 9,375 clicks and 9,375 × €0.80 = €7,500. That's €625 a week: 25% more budget for exactly the same goal, plus over 62 screenings a week instead of 50.

If instead you halve your no-show from 20% to 10%, you need 60 ÷ 0.90 = 67 scheduled intakes, 133 qualified candidates, 533 applications, 6,667 clicks and €5,333 of budget: €444 a week, over 11% less.

Without a model, 'qualification rate' and 'no-show' both look like soft process things. With a model you can see that five percentage points on one costs €1,500 a quarter, and ten percentage points on the other saves €667.

Two structural traps

Optimizing a step that isn't the bottleneck

Say you push your cost per click down 20%, from €0.80 to €0.64: €1,200 less over the quarter. But if your bottleneck is screening capacity, you're buying applications nobody calls back in time. You lower your cost per application and raise your cost per placement at once.

So ask it per step: if this ratio improves by 10%, how many placements does that produce? The step with the biggest answer is your bottleneck - often not the advertising side but the follow-up. The levers on the campaign side are covered in the article on cost per candidate.

Averaging across sectors and vacancies

Say that €6,000 and those 600 applications come from four campaigns: logistics 400 applications for €1,600, engineering 100 for €3,000, healthcare 60 for €900, construction 40 for €500. An average of €10 per application - exactly the number from your model.

Underneath that average sits something else: logistics costs €4 per application, construction €12.50, healthcare €15, engineering €30. If your margin is in engineering, the average steers you the wrong way: one good campaign hides four bad ones.

The fix: report every KPI per job family, never at total level only. And agree a minimum volume below which you don't read a segment - at twelve applications a ratio is coincidence.

Reporting cadence: week, month, quarter

• Weekly: spend against plan, applications, time-to-first-contact, scheduled intakes and no-shows - leading indicators you can correct immediately.
• Monthly: cost per qualified candidate per job family, application-to-intake, no-show ratio, and time-to-fill for vacancies closed that month.
• Quarterly: cost per placement, the candidate-side versus client-side ratio, the sector mix, and whether your assumptions held. Now you replace estimates with measured values.

What you may not conclude from one week of data: that a campaign is expensive or cheap per placement, that a creative wins, or that a job family doesn't work. At roughly one placement a week, one lucky placement moves your number by a hundred percent. A week shows that something has stalled, never that something works.

Instrumentation: the minimum

None of the above is possible without recording it. The floor:

• One mandatory source field per candidate, in your ATS - not in a loose spreadsheet.
• A fixed UTM convention, with the job family in the campaign name.
• A deduplicated conversion event on your thank-you page, so click-to-application is measurable.
• Timestamps at six moments: application received, first contact, intake scheduled, intake completed, submitted, placed.
• A job family code on every candidate record and every campaign, so you can join them.
• A weekly spend export per campaign on that same code.

With that in place, the modeling itself is half an hour of spreadsheet work. Without it you're calculating on feel, and every model is an opinion. Inside JAM we put that measurement setup in place before the first euro goes into ads.

Conclusion

Working backwards turns your quarterly plan from a hope into a standard. You know what has to come in each week at every step, you can see within three weeks where it's stalling, and you can justify whether an extra thousand euros produces a placement. The model doesn't have to be precise, just explicit.

Start with your own version of the sum, even if you have to estimate half the ratios. Want to run it on your own numbers? Book a call and we'll lay your funnel and assumptions side by side.