Beyond Job Board
Dependency

A single firm-wide cost-per-hire number is the kind of metric that lets a problem grow undetected. It averages a $1,340 all-occupations median against verticals that cost far more, and buries the trend underneath.

The labor market softened in 2025. Fewer workers planned to search (43% heading into 2026 compared to 93% the prior year), and apply rates climbed as those who did search applied more broadly to a smaller pool of openings. With 6.9 million openings drawing 5.6 million hires, platforms had no shortage of activity. Costs rose regardless.

This doesn't mean job boards are the enemy. They're the right call for an urgent, high-volume fill where speed is the point. The trouble starts when every search runs through the board by default, no one attaches a number to it, and the firm loses track of which placements belonged on a channel it already owns. The 2026 State of Staffing report found that 43% of agencies above $50 million in revenue spend $10,000 or more a month on job boards. At that scale, the board is a six-figure line at a unit price the firm does not set.

There is a deeper cost than the unit price. A job board hire is a transaction. The firm pays, the candidate starts, the assignment ends, and when the next req opens, the firm pays the platform again to find someone new. The board did its job. The firm rented a placement instead of keeping a candidate. What changes the economics is whether the firm turns that paid placement into someone it can place again, without paying to find them twice.

No channel does even a quarter of the work. Job boards carry 23.9% of placements, direct outreach 23.4%, ATS rediscovery 16.4%, and the rest splits among referrals, the career site, and AI tools. A growing firm runs that spread on purpose. It knows which placement belongs on which channel, and at what volume, because someone decided rather than defaulted.

Agencies already know which channels convert. But almost no one runs them as a system. Asked to name their single best-converting source, about 76% of agencies pointed to owned and earned channels: referrals, direct sourcing, the internal database, the career page. Job boards drew 19%.

So what did separate them? Operating discipline. Agencies that grew revenue 25% or more scored 4.47 on a seven-point operating-maturity scale. Agencies that shrank scored 3.56. AI moved with that discipline rather than standing in for it: 78% of high-growth firms run AI inside the ATS against 64% of all firms.

Job board spend, meanwhile, tracks how big a firm already is, not where it is headed. 43% of firms above $50 million spend $10,000 or more a month, and spend crossed with growth shows no pattern at all.

Lastly, contracting agencies are nearly twice as likely as non-contractors to run no formal referral program at all, 26% against 14%. The firms that built an owned channel into a system skewed toward the ones that grew.

Judge a channel on what it returns over a candidate's working life, not on what one placement cost. A job board hire is a transaction. The fee buys one start; when that assignment ends the candidate is gone, and the next req starts the meter again. A networked hire, someone who came in through a referral or a relationship the firm already had, works longer, comes back, and brings others. Cost per placement is too small a number to see any of that.

Start with the asset every firm already owns. 91% of agencies re-source at least some placements from their ATS or CRM. Only 32% run it as a primary source. A candidate already in the database was sourced and screened once, on a bill the firm already paid. Find that same person again through a paid job board, and the firm has paid the acquisition cost twice for someone it already had. The database is the cheapest channel in the building, and most firms barely open it.

Tenure is the first place the asset shows. Median lifetime days worked, by source: in light industrial, referred candidates worked 64 days against 36 for job board hires; in healthcare, 180 against 99; in travel nursing, 174 against 116. A 50 to 82% gap, and it holds in every vertical.

For those planning to lean harder on their own career site, it's worth noting that career-site hires tenure within three days of job board hires. What makes a candidate stay is the relationship behind the application, not the web page it came through.

Tenure runs long because the candidate comes back. Referred candidates redeploy at 51% against 38% for job board candidates in healthcare, and 39% against 31% in light industrial. And each one brings others. Counting the placements their referrals produce, the network multiplier runs 2.70 in healthcare and 2.06 in light industrial.

A job board hire is a fee for a start. A networked hire is that start, plus the assignments the candidate comes back for, plus the people they refer. It multiplies. It belongs on the asset side of the page.

Referrals carry the longest tenure of any channel measured here and still drive only 9.1% of placements. That's a reason to measure them, not to rebuild the whole budget around them.

Three capabilities can be measured on both sides: plan and practice.

1. AI deployment: 82% of firms plan AI-driven sourcing, 22% run AI in five or more processes.
2. ATS rediscovery: 57% plan to expand it, 32% run it as a primary source.
3. Referrals: 39% name them the top converter, under 5% make them a primary source.

Two more (candidate communication and tenure-by-source measurement) have intent on the record and no execution number anywhere, because firms do not track them.

Referrals are the sharpest case. Agencies name them the top converter, but 19% have no program, 38% run a basic or manual program, 31% have a structured one but still manual, and only 11% have a fully automated program. More than half of all firms sit at none or basic. The capability they rank first for conversion is the one they have least often turned into a system.