Agencies Running Zero AI Processes Contracted at 56% in 2025. Heavy Adopters Contracted at 31%.

Key takeaways:

• Agencies running AI in zero processes contracted at 56% in 2025. Heavy adopters (5+ processes) contracted at 31%, with every depth tier in between landing in order.
• Even at heavy-adoption depth, the most common reported outcome is a soft one: “enhanced candidate/recruiter experience” (31%), not a hard revenue or cost number. And 16% of heavy adopters still report no measurable impact at all.
• AI depth was one of three things separating growth agencies from the ones that contracted, not the only one. Operational discipline and an owned-channel sourcing mix moved with it. Growth agencies were ahead on all three at once.

Almost half of the agencies in our 2026 State of Staffing Report use AI in zero processes anywhere in the business. Asked how their 2025 went, 56% of that group lost revenue. Move up to light use, one or two processes, and the contraction rate drops to 41%. Moderate use, three or four processes, drops it to 34%. Heavy use, five or more, drops it to 31%.

The same step-down shows up on the growth side, not just contraction. Nearly 40% of heavy adopters and 38% of moderate adopters grew revenue 11% or more in 2025. Among agencies running no AI at all, only 17% did. It’s worth noting that agencies with more resources to grow may also have had more resources to deploy AI first, but the direction is the same at every tier across the full dataset.

What heavy adopters get instead of a revenue number

Job description generation, candidate qualification, and candidate matching are the three most-used AI processes in the industry right now. None of them show up directly in the financials. That aligns with what heavy adopters say they’re getting: their most commonly reported outcome, even at five or more processes deep, is “enhanced candidate/recruiter experience” (31%), a real result but not a financial one. Reduced time-to-fill (20%), improved candidate matching (16%), and lowered cost-per-hire (14%) all trail behind it. And 16% of heavy adopters still say they have no measurable impact to point to. Light adopters are further behind: 36% report no measurable impact, and another 10% aren’t sure.

Not every AI process carries the same weight 

Not every AI process carries the same weight. Our State of Staffing report ranks each one by how much more likely an agency using it is to land in the growth tier versus an agency that isn’t:

Sales outreach sits at the bottom despite being one of the more heavily marketed AI use cases right now. Reporting and analytics, one of the least glamorous entries on the list, ties job description generation for the top spot.

AI was one of three factors, not the only one 

AI depth was one of three things separating growth agencies from the ones that contracted in 2025, not the whole story. The other two: operational discipline (a seven-point scorecard covering habits like weekly KPI reviews and documented SOPs, where growth agencies averaged 4.56 and contracting agencies averaged 3.56) and an owned-channel sourcing mix (50% of growth agencies ran an all-owned top three sourcing mix, versus 27-31% of everyone else). Growth agencies were ahead on all three dimensions at once, not just the one everyone’s talking about.

Fifty-four percent of the agencies in our research run nine or fewer recruiters. None of the three factors above requires an enterprise budget. Wiring AI into two or three specific processes, running a weekly KPI review, and building a sourcing mix that isn’t dependent on job boards are decisions a 12-person desk can make and act on faster than an agency ten times its size can get through committee.

Two different questions, two different numbers 

A year ago, the adoption headline was 61% using AI somewhere in the business. This report asks a sharper question, how many recruiting processes deep, and rests on a different number: 46% of agencies are still at zero. The two figures aren’t measuring the same thing, so they don’t chart as a trend line. What they do show, together, is whether a firm uses AI at all matters far less than how much of the real workflow it touches, and this year’s data is the first StaffingHub survey with the growth and contraction numbers attached to prove it.