How do I know if my AI tools are actually making money or just sitting there?

Pull two numbers. How many people log into each tool every week, and what happened to your time-to-fill or submit-to-interview ratio since you bought it. If logins are low and the metrics did not move, the tool is a cost, not an asset. Most firms have never run this check. Do it this month.

We bought an AI sourcing tool and recruiters say they are too busy to use it. What now?

That is not a tool problem. That is a workflow problem. Recruiters skip tools that add a step instead of removing one. Sit with one recruiter for an hour and watch how they actually work. Then redesign the tool to live inside that flow, not beside it. If it cannot fit the flow, you bought the wrong tool.

Should we wait for the AI market to settle before buying more?

Stop buying for now, but do not wait to act. You almost certainly own tools you are not using fully. Fix the adoption gap on what you already have before you spend another dollar. Most firms can find 80% of the value they wanted in software already sitting in their stack.

The AI Productivity Trap in Staffing Firms

Staffing firms bought more AI in 2025 than ever, and 42% still lost revenue. The tools are not the problem. The gap between installation and impact is. Here is how to close it.

Staffing firms spent more on AI in 2025 than any year before it, and 42% of them still lost revenue. That is not a coincidence. That is a pattern.

Gartner reported that 88% of HR leaders see no significant business value from AI yet. Read that again. Almost nine out of ten leaders bought the tools, ran the rollout, sat through the demos, and got nothing they can point to on a P&L.

The story everyone keeps telling is about adoption. Who is using AI and who is behind. That is the wrong story. The real story in 2026 is the gap between installing AI and getting impact from it. Those are two different things, and most firms confuse one for the other.

Installing Is Not Using

Here is what I see when I walk into a staffing firm that bought five AI tools last year. The contract is signed. The tool is connected to the ATS. The vendor sent a nice onboarding deck. The leadership team checked the box that says "we are an AI-forward firm."

Then I ask one question. How many recruiters logged into that tool last week? The room goes quiet.

Installation is an event. Impact is a habit. You can install a tool in an afternoon. Building the habit that turns it into revenue takes weeks of deliberate work, and almost nobody budgets for that part.

This is the trap. Buying the tool feels like progress. It is measurable, it is fast, and you can announce it. Changing how 40 recruiters actually work every day is slow, invisible, and hard. So firms keep buying and keep skipping the hard part. Then they wonder why the revenue line is flat.

Why the Tools Sit Unused

I have watched dozens of these rollouts. The tools fail for the same three reasons almost every time.

The tool adds a step instead of removing one. A recruiter has a way they work. If your AI sourcing tool asks them to leave their flow, log into a new screen, and copy data back and forth, they will skip it under pressure. And recruiters are always under pressure.
Nobody owns the outcome. The owner who signed the contract moved on to the next fire. No one is responsible for whether the tool actually changes a number. So it drifts.
There was no before number. If you never measured your submit-to-interview ratio before you bought the screening tool, you cannot prove the tool did anything. So you cannot defend it, scale it, or fix it.

None of these are technology problems. Every one is an operations problem. That is why buying another tool never fixes it. You are treating a workflow disease with a software prescription.

What Impact Actually Looks Like

Let me show you the difference with a real scenario.

A 60-person IT staffing firm buys an AI screening tool. The first firm installs it, tells recruiters it exists, and walks away. Six months later, login data shows 11 of 38 recruiters ever used it. Time-to-fill did not move. The tool costs $4,000 a month and produced nothing.

The second firm does it differently. Before they buy, they measure one thing: their team spends an average of 90 minutes per req on first-pass resume review. They pick the tool to attack that exact number. They train three recruiters first, not all 38. They watch those three for two weeks and fix the friction. Then they roll it to the rest with a real workflow, not a slide deck.

Ninety days later, first-pass review drops to 30 minutes per req. That is one hour back per req, across hundreds of reqs a month. That hour goes into more submittals and more client calls. The revenue line moves because the tool changed a specific behavior that was tied to a specific number.

Same tool. Same price. Wildly different result. The difference was never the software. It was the work around the software.

The Build, Change, Adopt Lens

This is exactly why I work in three stages, and why most firms only do the first one.

Build is the easy part. Pick the tool. Connect it. Configure it. Most firms are fine here. Vendors are good at this stage because it is where they get paid.

Change is where the money lives and where firms quit. This is redesigning the actual workflow so the tool removes steps instead of adding them. It is naming an owner. It is setting the before number so you can prove the after number.

Adopt is the part with no shortcut. It is training small groups, watching real usage, fixing friction, and holding the line until the new way becomes the only way. This takes weeks, not an afternoon. It is the part the trap convinces you to skip.

When a firm tells me AI is not working for them, I almost never find a tool problem. I find a firm that did Build and called it done.

Stop Buying. Start Counting.

Here is your one action this week. Do not buy anything. Do not schedule a new demo.

Make a list of every AI tool you pay for. Next to each one, write two numbers. How many people used it last week, and what business metric it was supposed to change. If you do not know either number, that tool is a candidate to cut or to fix.

For the one tool with the most potential, find its before number. Then sit with one recruiter for an hour and watch them try to use it inside their real work. You will see the friction in the first ten minutes. That friction is the whole problem, and now you can fix it.

The firms that win in 2026 will not be the ones with the most AI. They will be the ones that turned three tools they already own into three numbers they can prove. That is a choice you can make this week.

The AI Productivity Trap: Why Staffing Firms Are Adding Tools But Not Adding Revenue