Why Off-the-Shelf ATS Platforms Fail Multi-Desk Agencies
One desk, one workflow, one commission model — that's what most recruiting software assumes. Add a second vertical and the assumptions start costing you.
An off-the-shelf ATS is usually the right answer for a single-desk agency. It's cheaper, it works on day one, and the workflow it imposes is probably fine.
The trouble starts at the second desk. Not because the software breaks, but because it quietly assumes there's one way your agency works, and there are now two.
What "multi-desk" actually breaks
Different processes per vertical. Perm placement in professional services is take the brief, source, submit, interview, offer, placed, invoice on start. Light industrial temp is a pool, a shift request, fill it, timesheet, invoice weekly. Executive search is a retained process with stages, stakeholders and reports that has almost nothing in common with either.
Force all three through one pipeline and two of them are using a process built for someone else. You'll see it as low system adoption on those desks, and the actual cause is that the tool doesn't fit their day.
Commission that varies. Perm on a percentage of salary. Contract on margin over the assignment. Split credit between sourcer and account owner. Different rates per desk, per seniority, sometimes per client. Most platforms model one scheme, so the rest lives in a spreadsheet and gets argued about monthly.
Margin visibility. Perm margin is straightforward. Contract margin is bill rate minus pay rate minus burden, per hour, across the assignment, and it changes when rates change mid-contract. Agencies routinely can't produce true margin per contract placement, which means they can't tell you which desk is actually profitable.
Candidates who belong to more than one desk. A finance candidate who'd also take an operations role. Two desks, one person, and in most systems either a duplicate record or a territorial argument.
Reporting that has to work both ways. Perm desk on placements and fee. Contract desk on contractors on assignment and weekly margin. Neither is wrong; one report can't serve both, and a platform that gives you a fixed dashboard gives you one.
The workarounds and what they cost
Agencies don't hit a wall. They adapt, and the adaptations have a price.
A spreadsheet appears. Usually commission first, then margin, then the contract desk's real pipeline. Every spreadsheet is a place where the system's numbers and reality diverge.
Fields get repurposed. "Source" becomes desk. A custom field becomes assignment end date. It works, and then a new hire misuses it, and now your data is wrong in a way nobody notices for months.
Two systems. One for perm, one for contract. Now the candidate database is split, which is the one thing you cannot afford to split.
People opt out. The contract desk keeps its real pipeline in a shared doc because the system's stages don't match reality. Your reporting now covers half the business.
That last one is the expensive failure, because it's invisible from management. The dashboard looks fine. It's just describing a fraction of what's happening.
When to stop adapting
Reasonable signals that the platform has stopped fitting:
- More than one desk keeps its real pipeline outside the system
- Commission is calculated in a spreadsheet every month, and disputed
- You can't produce margin per contract placement without manual work
- Onboarding a recruiter includes explaining which fields "actually mean" something else
- You've turned down a vertical because the system wouldn't handle it
- Per-seat cost has become a factor in whether to hire
None of those alone justifies building something. Three or four together usually means the workarounds cost more than the alternative.
What to keep in mind if you do build
Keep one candidate database. Everything else can vary by desk. This cannot split, ever.
Model the submission as a first-class record, not a candidate status. It's the join between the two pipelines and every useful question is asked of it.
Make stages configurable per desk. Same underlying records, different workflow. Perm and contract shouldn't share a pipeline definition.
Compute margin from the placement record. Bill rate, pay rate, burden, hours. If it can't be calculated, it will be estimated, and estimates favor whoever is estimating.
Attribute both sides. Sourcer and account owner on every placement, separately. This makes commission calculable rather than negotiable.
Assume the process changes. You'll add a vertical. Build so that's a configuration change, not a project.
The honest caveat
Custom is not automatically better. A well-run single-desk agency on a good off-the-shelf platform will beat a badly-run agency on a bespoke system every time, and building something to escape a tool you haven't properly configured is an expensive way to avoid a week of setup.
The case for building shows up when the workarounds have become the process — when the real business runs in spreadsheets and the platform is where you file things afterwards. That's the point where you're already paying for custom software. You just don't own it.
We build multi-desk recruiting systems for US staffing agencies, with pipelines that differ by desk and one database underneath. See how it fits together, or tell us where the spreadsheets are.