Why Your Team Likes Using AI and the P&L Hasn't Noticed

What is a satisfaction score measuring?

Think about how the question is usually asked. "How satisfied are you with the AI you have access to?" Or, "Has AI made your work easier?" Those are fair questions. They measure whether AI is pleasant and useful to the person using it.

In the chain we use to trace AI spend through to return, that sits at Usage. Money was invested, people have access, and people are using AI and like using it. Those first two links are the ones every organization can see, because the finance system records the first and the vendor dashboard and the pulse survey both report on the second.

What the survey does not ask is where the saved time went, whether anyone decided which work AI should be pointed at, and whether people have the skill to use it on the work that carries weight. Those are the links in the middle: Direction, Skill and Reinvestment. Nobody measures them, and they are where a return gets made or lost.

So a high satisfaction score is good news about Usage. It tells you nothing either way about the rest.

Where does the time saved by AI go?

It is the question the survey leaves out.

Say a marketing coordinator used to spend most of Tuesday pulling together the weekly performance report. Now an AI assistant drafts it in under an hour, and she checks it and sends it. She is right to say she likes it. The question is what happened to the rest of Tuesday.

Often, nothing happened to it on purpose. Some of it went to email. Some of it went to the next report, which also got done faster. Some of it went into a little breathing room, which after the last few years most people have earned. None of that is a problem in itself. It does not show up anywhere a finance team can see.

Time saved turns into return only when somebody decides what it should be spent on and sets the work up so it can be. That is Reinvestment, and a manager has to make that decision, because AI will not make it for them. When nobody makes it, the saving spreads thin across the week and disappears. There is more on that decision in AI Saved Your Team Hours. Where Did You Send Them?

Why doesn't the survey catch the skill problem?

The other thing a satisfaction survey tends to miss is who is answering it.

The people who use AI most are usually the ones who were already comfortable with it. They found the good uses early, they got faster, and they rate it highly. The people who tried it twice, got something unhelpful back, and went back to the old way often skip the survey, or answer it politely.

So the score can climb while skill across the organization stays uneven. A handful of people get much better at their work, and the team as a whole produces about what it did before. That is a Skill problem, and an average hides it well.

The same goes for Direction. If nobody has said which workflows AI is meant to change, everyone applies it to whatever they happened to try first. People are satisfied with the result, because they chose the task. The business gets a scatter of small improvements with no connection to what the leadership team said the investment was for.

What should leaders measure alongside satisfaction?

Keep the survey. It tells you whether people are using AI and whether it is any good, and you need to know both. Stop presenting it as evidence of return, because the first person who asks what moved will find the space between the two.

Alongside it, a few plainer questions get closer to the middle of the chain:

• Direction. Can the leadership team name the handful of workflows the investment was meant to change? If each leader gives a different list, that is the finding.
• Skill. Beyond the enthusiasts, what share of the team uses AI on the work that matters, and can they show you the before and after?
• Reinvestment. For the time the survey says was saved, where was it deliberately sent? Which piece of work exists now that did not before?

None of these need a new platform. They need someone to ask them in the room, and a team willing to give an honest answer. The second part is usually the harder one.

Back to the vice president and her slide. The strongest version of it keeps the survey result and puts one more line underneath: here is where the saved time went, and here is what it produced. If she cannot write that line yet, the room should hear that too, because it tells everyone where the next quarter's work is.

Where to start

If your own numbers look like this, with happy users and a flat P&L, the useful first step is working out which link in the chain is holding the return. Take the free AI Profit Ready Assessment: eight questions, about two minutes, and it shows on screen where the profit is leaking. If you want to work through the result with your leadership team in a room, the AI Profit Workshop is a half-day for up to six people, and its fee is credited in full toward the AI Profit Sprint if you start one within six months.