Last quarter I watched a company celebrate rolling out an AI assistant to 4,000 people. Thirty days later, active usage sat at 6%. The licenses were bought. The value was not. This is the AI adoption gap, and it is the single most expensive mistake in enterprise transformation right now.
Here is the uncomfortable truth I keep repeating to leadership teams: buying AI is a procurement event. Using AI is a behavior change. These are not the same project, and treating them as one is why so much budget quietly evaporates.
The gap is real, and it is measurable
I have seen the pattern across ERP programs and AI rollouts alike. Tools get deployed with a launch email, a slide deck, and a webinar. Then nothing. Six weeks in, a handful of enthusiasts are power-users, and everyone else has gone back to the old way.
The reason is simple. People do not change their daily habits because a tool exists. They change when the new way is easier, safer, and visibly rewarded inside the work they already do. If the AI lives in a separate tab that nobody is asked to open, it will stay closed.
A tool that requires people to remember to use it has already lost. Adoption happens where the work happens.
The playbook I actually use
None of this is theoretical. This is the sequence I run, and it is deliberately unglamorous.
1. Start with one painful workflow, not one shiny tool
Do not pilot "AI." Pilot a specific, annoying task that a real team does every week: drafting supplier emails, reconciling reports, answering the same 40 support questions. Pick something with a clear before-and-after. A good pilot is 20 to 40 people, one workflow, 6 to 8 weeks. Small enough to move, big enough to prove.
2. Recruit champions before you recruit users
For every 15 to 20 people, you need one champion: a respected peer, not a manager, who uses the tool daily and helps colleagues in the flow. Champions cut your support load and, more importantly, they make adoption a social fact instead of a top-down order. When your best analyst says "this saves me an hour," that beats any executive memo.
3. Integrate into the workflow, not alongside it
This is where most rollouts die. The AI has to appear where people already work: inside the ERP screen, the CRM, the email client, the ticketing tool. Every extra click, login, or context switch cuts adoption. Your target is zero new tabs. If people have to leave their environment to get value, they won't.
4. Measure usage and outcomes, weekly
What you don't measure, you can't manage. From day one, track two layers:
- Adoption signals: weekly active users, tasks completed with AI, repeat usage over 30 days.
- Outcome signals: time saved per task, error rate, cycle time, and the honest one — user satisfaction.
Publish these numbers every week to the pilot team. Visibility creates momentum, and it tells you fast whether you have a product problem, a training problem, or a workflow problem. They require very different fixes.
5. Close the loop and remove friction
Collect what confuses people, then kill those frictions one by one: a confusing prompt, a missing permission, an output people don't trust. Trust is the real currency of AI adoption. Every fixed friction is a new group of daily users. Every ignored one is a quiet exit.
The mindset shift for leaders
The hardest part is not the technology. It is accepting that adoption is the product, and change management is the delivery mechanism. The model is a commodity. Whether your people actually use it is your competitive advantage.
So stop asking "which AI tool should we buy?" Ask instead: which workflow will we change, who will champion it, and how will we know it worked? That is a different conversation, and it is the one that separates companies getting real returns from those paying for shelfware.
Where to start Monday
Pick one workflow. Name three champions. Set two metrics. Run an eight-week pilot with a real before-and-after. Do not roll out to 4,000 people until you can prove value with 40.
The companies winning with AI are not the ones with the biggest license count. They are the ones who treated adoption as a discipline, not an afterthought. The gap between buying and using is exactly where the value lives — and closing it is a choice you can make this quarter.
Originally published on cedricbignet.com. I'm Cédric Bignet — AI & ERP Change Management expert at Novartis and founder of AInspire.
Top comments (0)