How do I know if my AI tools are actually driving revenue?

Tie each AI tool to a revenue metric, not an activity metric. If your sourcing tool doubled candidate outreach, track whether submittals per placement dropped and whether time-to-fill shrank. If those numbers held flat, the activity went up but nothing moved to the P&L. Activity that does not shorten the path to a placement is cost, not return.

What metrics should a PE ops leader put in front of the board?

Give the board four numbers per firm: cost per placement, time-to-fill, gross margin per recruiter, and revenue per recruiter hour. Then show the trend line before and after the AI rollout. Boards do not care that call time hit 286 minutes a week. They care whether that time turned into billable placements at the same or better margin.

Our activity is up but revenue is flat. Is that the AI's fault or the market?

It is usually both, and you cannot fix either until you separate them. Pull your data from before the market softened and compare conversion rates, not raw volume. If your submittal-to-hire rate held steady while volume climbed, the market is the drag. If conversion dropped, your AI is generating busywork. The scorecard tells you which lever to pull.

Build an AI ROI Scorecard for Your Staffing Firm

ASA says recruiter interactions jumped 60% and call time hit 286 minutes a week. The market is still soft. Here is how to tell AI activity from AI revenue, and how to build a scorecard your board will trust.

Recruiter interactions jumped 60% year over year, and weekly call time hit an all-time high of 286 minutes. That is ASA's Q1 2026 Staffing Productivity Report. Your recruiters have never been busier.

The staffing market is still soft. Revenue is flat or down at most firms I talk to. So you have a room full of people working harder than ever, and a P&L that does not show it.

That gap is the whole story. AI made your team more active. It has not yet made your team more profitable. And your board is going to ask you to prove the difference sooner than you think.

Activity Is Not Revenue. Stop Reporting It Like It Is.

Most AI dashboards measure motion. Outreach sent. Calls dialed. Candidates screened. Messages personalized. These numbers go up and to the right, and they feel like progress.

They are not progress. They are input. A recruiter can send 400 personalized messages a week with an AI tool and place fewer people than the recruiter who sent 80 by hand last year. Volume without conversion is just faster spending.

Here is the trap PE-backed portfolios keep falling into. The vendor demo shows a 3x lift in recruiter activity. The ops leader buys it. Six months later activity is up, seat cost is up, and gross margin per recruiter has not moved. Now you are explaining to the deal team why the tech spend did not land.

The fix is simple to say and hard to do. Every activity metric needs a revenue metric sitting next to it. If the two do not move together, the activity is noise.

The Four Numbers That Actually Belong on the Scorecard

Forget the 30-metric dashboard your software gives you. Boards do not read those, and neither should you. Build your AI ROI scorecard around four numbers that connect straight to the P&L.

Revenue per recruiter hour. Take billable revenue and divide it by total recruiter hours worked. If your team put in 286 minutes a week on calls and this number is flat, the extra time is not converting. This is the single most honest measure of whether AI bought you leverage.
Cost per placement. Add your fully loaded AI tool spend into the cost of each placement. If the tools were supposed to make placements cheaper to produce and this number climbed, you are paying more to do the same work.
Time-to-fill. AI's clearest win is speed. Sourcing and screening should shorten the path from open req to filled req. If time-to-fill did not drop after your rollout, the AI is not doing the job you bought it for.
Gross margin per recruiter. This catches the thing volume metrics hide. A recruiter can place more people at worse margins because they are chasing easy fills the AI surfaced. Margin per head keeps them honest.

Notice what is missing. No message counts. No dial counts. No screening volume. Those live one layer down as diagnostics. They explain the four numbers. They do not replace them.

How to Separate the Market From the Machine

The hardest question you will get is this one. Is revenue flat because of the market or because the AI is generating busywork? You cannot answer that with volume data. You answer it with conversion rates.

Pull two windows of data. One before the market softened and before your AI rollout. One from the last full quarter. Then compare conversion at each stage, not raw counts.

Submittal-to-interview rate
Interview-to-offer rate
Offer-to-placement rate

If those conversion rates held steady while volume climbed, the market is your drag. Your team is doing good work on more at-bats, and the demand just is not there. That is a sales and market problem, not a tech problem.

If those conversion rates dropped while volume climbed, your AI is manufacturing activity that does not convert. More submittals, worse fit. More outreach, colder responses. That is a tech and workflow problem, and no amount of extra call time fixes it.

This one comparison tells you which lever to pull. Most leaders skip it because the raw volume numbers look great and nobody wants to ruin a good dashboard. Do the comparison anyway.

Build the Scorecard Before the Board Asks

PE ops leaders, this is the part that matters for you. The board conversation about AI is shifting from "are we using it" to "what did it return." That shift is happening in 2026. You want your scorecard built before the question lands, not scrambled together the week of the board meeting.

A board-ready AI scorecard is one page per firm. Four numbers. Trend lines showing before and after the rollout. One line of plain English on what the trend means. No vendor logos. No activity vanity metrics.

When you can show that time-to-fill dropped 18% and revenue per recruiter hour climbed 12% while the market fell, you have proof the AI is working. When the numbers went the other way, you have an early warning you can act on instead of a surprise you have to explain.

The firms that win the next 18 months are not the ones with the most AI tools. They are the ones who can tell activity from return, and who cut the tools that only produce motion.

Your Move This Week

Pick one AI tool you have running right now. Sourcing, screening, engagement, whatever is live. Pull its activity number and put your matching revenue number next to it. Time-to-fill, cost per placement, or revenue per recruiter hour. Compare the two trend lines from before the rollout to today.

If they move together, you found a winner. Fund it. If activity climbed and revenue did not, you found a leak. Fix the workflow or cut the tool. Either way, you now have one honest data point, and that is where the real scorecard starts.

Your Recruiters Are 60% More Active. Is Revenue Following?