December 19, 2025 · 5 min read
What a Client's Resistance to AI Adoption Actually Taught Me
A client's team resisted an AI tool rollout for months, and once I actually listened instead of pushing the business case, the resistance turned out to be a reasonable response to a real, unaddressed risk.
I spent the first two months of an AI rollout project assuming the client's marketing ops team resistance was a change-management problem — people being reflexively wary of new tools, needing reassurance and training. I pushed harder on the business case, built nicer demos, showed bigger ROI numbers. Adoption stayed flat. Eventually I just asked, directly and specifically, what was actually bothering them.
What they actually said
The concern wasn't about the tool's usefulness at all — it was that the team had no clear policy on what happened if the AI-generated content it produced turned out to contain an error that reached a customer. Who was accountable. Whether their jobs would be the ones scrutinized if something went wrong, with no defined process protecting them for using a tool leadership had told them to use. Nobody had actually answered that question, and the team wasn't going to adopt a tool whose failure mode exposed them personally with no cover.
What looked like resistance to AI was actually a completely reasonable refusal to take on undefined personal risk for a tool nobody had agreed to take responsibility for.
What actually fixed adoption
- A written policy, approved by leadership, clarifying that AI-assisted output went through the same review and accountability chain as any human-drafted work.
- An explicit statement that using the approved tool as directed wasn't something the team would be individually penalized for if an error slipped through the existing review process.
- A visible error-reporting path that fed back into tool improvement, so the team felt like reported problems actually went somewhere instead of disappearing.
Adoption climbed steadily once the policy existed, with no change to the tool itself. The lesson I keep relearning on these projects: resistance framed as 'people don't trust AI' is very often 'people correctly identified an unaddressed risk that AI's proponents hadn't bothered to resolve,' and the fix is organizational, not technical.