“Should we reclaim the unused licences?”
It sounds like a procurement question. A portion of our AI licences were being barely used, the renewal was approaching, and the responsible thing, on paper, was to harvest the idle ones and shrink the bill. Straightforward cost discipline.
I’m glad we paused before treating it that way, because the question was never really about money.
An unused AI licence isn’t like an unused software licence. When someone doesn’t use a project management tool, it usually means their work doesn’t need it. When someone doesn’t use AI for months, in this era, it’s a signal about something else: they don’t see the relevance, they haven’t had the learning support, they’re anxious about it, or their leader has silently signalled it doesn’t matter here. The licence isn’t idle inventory. It’s a diagnostic reading, telling you exactly where your adoption program isn’t reaching.
Reclaiming the licences would have destroyed the diagnostic and confirmed the anxiety. Imagine the message: we gave you access to the most significant workplace technology in a generation, you hesitated, so we took it back. Every hesitant person learns that hesitation is self-fulfilling. And the next time we ask the organisation to lean into change, that memory votes.
So instead of harvesting licences, we followed them. Where was usage low? The patterns were not random. Whole teams sat dark together, which almost always pointed to the leader, not the individuals. Some areas lacked obvious use cases and needed help translating the technology into their work. Some people, given an hour of side-by-side support, lit up and never looked back.
The licences were the cheapest part of the whole equation. The information they generated about our organisation was worth far more than the saving.
When your renewal conversation comes, and it will, resist letting it become a spreadsheet exercise. Ask instead: what is our usage data telling us about where adoption needs help? Then spend accordingly.