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

Pick one metric that touches money and track it before and after the tool goes live. Time-to-fill, submit-to-interview ratio, or gross margin per recruiter. If the tool cannot move one of those in 90 days, it is a cost, not an investment. Activity numbers like call time and emails sent do not count. Those tell you people are busy, not that you are winning more deals.

Our recruiters love the new tools. Isn't adoption a good sign?

Adoption is a good sign that people will use the tool. It is not a sign the tool makes money. Those are two different things. Recruiters can happily use a sourcing AI every day while your fill rate stays flat. Love for a tool measures comfort. Revenue measures results. Track both, but never confuse one for the other.

What is the single fastest way to spot wasted AI spend?

Compare your tool invoice list against your outcome metrics from the last two quarters. For every tool, ask one question: what number did this move? If nobody can answer with a specific figure, that tool is a candidate to cut. Most firms find at least one tool that does nothing but generate activity nobody asked for.

The AI Productivity Paradox in Staffing Firms

Recruiter call time doubled year-over-year and AI tool usage climbed. But McKinsey says only 6% of AI-adopting firms see measurable EBITDA impact. Here is the framework to connect the activity to the money.

Recruiter call time doubled year over year. A June 2026 ASA report built on Prodoscore data put a number on it. AI tool usage across staffing desks climbed at the same time. On paper, that looks like a win.

Then McKinsey published a number that ruins the party. Only 6% of organizations adopting AI generate measurable EBITDA impact from it. Six percent. Everyone else is spending money and generating motion.

That gap is the AI productivity paradox. Your recruiters are talking more, sourcing faster, and touching more tools than ever. Your P&L does not care about any of that unless it turns into placements and margin. Right now, for most firms, it does not.

Busy Is Not the Same as Profitable

Here is what happened on a lot of desks over the past year. A recruiter got an AI sourcing tool. They got an AI note-taker. Maybe an AI email writer. Each tool made one task faster. So the recruiter did more of that task.

More sourcing. More calls. More outreach. The activity dashboard lit up green. Leadership saw the green and assumed the money would follow.

It did not follow, because nobody connected the faster task to a revenue outcome. Doubling call time only matters if those calls close more reqs or fill them faster. If your time-to-fill did not drop and your margin did not rise, the extra calls were just noise with a phone attached.

This is the trap staffing VPs keep falling into. Activity metrics are easy to measure. Revenue outcomes are harder. So people measure what is easy and hope it maps to what is hard.

It rarely does on its own.

The Three Numbers That Actually Move

You do not need a data science team to fix this. You need to stop tracking activity as if it were the goal and start tracking three outcome numbers instead.

Time-to-fill. If AI is doing its job in sourcing and screening, this drops. Measure it per desk, per month, before and after each tool. If it holds flat, the tool is not helping you win.
Gross margin per recruiter. This is the number PE ops leaders actually care about. More activity should let a recruiter carry more reqs or higher-value reqs. If margin per head is flat while tool spend rises, you are subsidizing busywork.
Submit-to-hire ratio. AI screening is supposed to raise the quality of who you submit. If you are submitting the same volume of candidates to land the same number of hires, the screening tool is decorative.

Pick these three. Put them on one page next to your list of AI tools. Now you can see which tools earn their keep and which ones just make people feel modern.

Why the 6% Win and Everyone Else Doesn't

The firms in McKinsey's 6% share one habit. They tie every tool to a business outcome before they buy it, and they kill the tool if the outcome does not show up.

The other 94% buy tools to keep up. A competitor announced an AI feature, so they matched it. A vendor gave a great demo, so they signed. Nobody wrote down what number the tool was supposed to move. So when the number did not move, nobody noticed.

This is a Build, Change, Adopt problem, and most firms only do the first part. They build the tech stack. They never manage the change or measure the adoption against revenue.

Building the stack is the easy 20%. Getting recruiters to change how they work, and proving that change made money, is the hard 80%. That is where the 6% separate from the pack.

Here is the uncomfortable part. Some of your AI tools are making recruiters faster at the wrong things. A recruiter who sends 200 AI-written emails a day is not more valuable than one who sends 40 that land better candidates. Speed applied to a low-value activity just gives you more low-value activity, faster.

How to Audit Your Stack This Week

You can start closing the paradox in your own firm without a big project. Do this in one afternoon.

Pull two lists. List one is every AI tool you pay for and the monthly cost. List two is your outcome numbers for the last two quarters: time-to-fill, gross margin per recruiter, submit-to-hire.

Now line them up. For each tool, write down the one outcome number it was supposed to improve. Then write down whether that number actually moved.

You will land in one of three buckets for every tool.

Moved a number. Keep it and double down.
Moved nothing but people like it. Put it on a 60-day watch. If it does not produce, cut it.
Moved nothing and nobody can explain why you bought it. Cut it now.

Most firms find they can cut two or three tools and lose nothing but the bill. That freed-up budget goes toward the one or two tools that actually move margin.

The recruiters who talk more are not the problem. The problem is that nobody built the bridge between the talking and the revenue. That bridge is your job as a VP or ops leader, not the vendor's.

Your Next Step

This week, build the two-list audit. One page: every AI tool and its cost on the left, your three outcome numbers on the right. Draw a line from each tool to the number it was supposed to move. The tools with no line are your answer.

Do that before your next tool renewal comes up. It will change what you renew, and it will change how you buy the next one.

The AI Productivity Paradox: Your Recruiters Are Busier Than Ever. Is Your Revenue?