You are paying for twenty seats. Six people log in. Two use it properly.
The instinct is to push harder. Another training session, a reminder in the all hands, a gentle word about the cost.
That rarely works. The reason people avoid it has almost nothing to do with knowing it exists.
Five reasons this happens.
1. It is slower than the old way, and it always was
For most tasks the AI version gets faster once you know how to use it, and stays slower for the first two weeks while you learn.
Your team is busy. Two weeks of being slower is a real cost they pay personally, for a benefit that lands later and mostly on the business.
So they use the old way. Rationally.
How to spot it. People call the tool "fine" and never open it. No complaints, no engagement.
What to do. Give them the two weeks back. Protect the time explicitly, or accept that people keep choosing whatever gets today's work done.
2. They do not know when to trust it
This is the big one, and it hides well.
Someone uses AI for a task. The output looks plausible. Now they choose... send it, or check it.
Checking takes as long as doing the work themselves. So they check. Every time. The tool saves them nothing, so they stop.
How to spot it. People use it and then rewrite the output completely. Or they use it for trivial things and never for real work.
What to do. Teach the edges. Where does this tool perform well? Where does it drift? What does a bad answer look like in your business?
Run a session where the team tries to break it on purpose. People who have watched a tool fail in a safe setting trust it far more than people who have only seen it succeed in a demo.
3. It sits outside where the work happens
Your team works in email, in your CRM, in a spreadsheet. The tool is a separate tab.
Every switch is a decision point. Every decision point leaks people back to the familiar path.
How to spot it. Usage spikes right after training, then decays over three weeks. Classic shape.
What to do. Move the AI to where the work already is, or accept low usage. Willpower loses to friction over months.
4. Nobody senior visibly uses it
People read what leaders do and discount what leaders say.
If the founder still forwards things to an assistant to write, the message is clear whatever was said at the workshop.
How to spot it. Ask three people who in the leadership team uses it. Watch how long it takes them to answer.
What to do. Use it yourself, in front of people. Share a prompt that worked. Share one that failed. Being visibly mid learning does more for adoption than any mandate.
5. They think it is coming for their job
Rarely said out loud. Often the real thing.
The person best at the task you just automated has the most to lose and the most influence over whether it succeeds. They will be polite about it and it will not get used.
How to spot it. Enthusiastic agreement in meetings, no behavior change. Quiet resistance in the details. Reasons the tool will not work for their particular case.
What to do. Say it directly. What happens to this role when the admin disappears? If the honest answer is that the job gets more interesting, say so and mean it. If the honest answer is harder, people already suspect it, and pretending costs you more than the truth.
Find out which one you have
Stop guessing. Three conversations, twenty minutes each, with people who are not using it.
Ask one question, then be quiet.
"Walk me through the last time you thought about using it and decided not to."
Do not defend the tool. Do not explain the feature they missed. Let the silence work. The answer you get in minute four is the real one.
Three conversations usually surface the same reason twice. That is your problem.
The thirty day reset
Week 1. Three conversations. Name the actual reason. Cut seats for anyone who has not opened it in sixty days, because you are paying for a number that flatters you.
Week 2. Pick one workflow, one team, five people. Move the AI into where that work already happens.
Week 3. Run a session on the edges. Where it works, where it fails, what bad output looks like. Break it together on purpose.
Week 4. Show the result to the wider business. Real numbers from real work by people they know.
Five people using it properly beats twenty people with a login. Adoption spreads sideways from a working example far better than it spreads down from an announcement.
When to walk away
Sometimes you bought the wrong tool for work that was fine already.
If three conversations show nobody has a problem the tool solves, cancel it. You save the subscription, and you stop teaching your team that AI arrives, gets ignored, and quietly disappears.
That lesson is expensive. It makes the next attempt much harder.
Check whether a build will stick.
The Build and Embed Readiness tool scores your team on the six markers that predict adoption, and shows you which one is holding you back.