The licence nobody opens
Nadia runs customer service for eleven people at a family-owned wholesaler, and she has an AI licence she has opened twice. Both times were in the training session. Her manager bought the seats after a good demo, sent an email with a link, and moved on to the next thing. The seats renew every month.
Ask her why the tab stays closed and she gives you a practical answer. She has no idea whether she is allowed to paste a customer's order history into it, nobody told her which of her tasks it is supposed to take over, and the tool does not know a thing about the products she sells. So she does what she has always done, and the invoice keeps arriving.
The tab she never opens
Three quarters of the AI licences that organisations pay for stay unused. Nadia is not an exception in her company. She is the pattern.
What makes her case worth studying is that the software works. The model is capable. The vendor delivered. The part that failed sits around the tool: nobody decided what she should stop doing, nobody connected it to her order data, nobody told her what is safe to type into it. The licence was the easy purchase. Everything that would have made it pay off was left for later, and later never got an owner.
Six blocks that multiply
A working AI setup rests on six blocks: strategy, data, way of working, governance, technology, and culture and skills. They multiply. One zero makes the whole thing zero.
For Nadia that maths is concrete:
- Strategy: nobody said which customer question she is meant to answer faster.
- Data: the tool cannot see her order history or her delivery times.
- Way of working: her scripts, her handover to the warehouse and her call targets stayed the same.
- Governance: no answer on what she may share with the model.
- Technology: the one block that was fine.
- Culture and skills: two hours of training, six months ago.
Five zeroes and one strong block. The product of that is zero, and the invoice keeps coming.
You cannot buy your way past the five blocks you skipped.
Where the zero usually sits
In small companies the missing block is almost always the way of working. There is no data team to blame and no architecture committee to wait for. There is a team leader with a full week and a new tool on top of it.
Nadia's week has 140 calls, a stock issue every Tuesday, and two people on holiday. Add a tool to that week and it becomes extra work. Redesign a piece of that week around the tool and it becomes lighter. The difference is whether someone sat down with her for an afternoon.
What she actually uses
Real use of self-chosen tools runs two to three times higher than use of the licences companies hand out. Nadia belongs there too. She drafts her difficult replies in a free tool on her own phone, pastes the text back into the ticket, and tells nobody.
That is shadow AI, and it is the most useful signal in the building. It shows you exactly which task is heavy enough that people solve it themselves. Ask your team which tool they already reach for and why, and you have your first use case, validated by the people who will use it.
Make her the design team
The people who already found a workaround know where the value is. Give Nadia one afternoon a month, a written rule on what she may share, and permission to change her own process. That covers four of the six blocks in a single move.
What you do on Monday
- Pull the licence report and count how many seats were opened last week.
- Ask three people which AI tool they use on their own phone, and for which task.
- Write one page: what may be shared with a model, what may not. One page is enough.
- Pick one task from that list and redesign the week around it with the person who does it.
- Set a date in six weeks to look at the licence report again.
Chapter 7 of CX+AI works through the six blocks with the questions to ask per block. The CX+AI Maturity Check gives you a score per block, so you can see which zero is holding back the rest before you buy anything else. Bekijk de materialen