Skip to main content

When to Push and When to Wait in AI Rollouts

The hardest judgement call in enterprise AI: knowing when resistance is a signal to slow down, and when it's a barrier to push through.
12 March 2025·5 min read
Tim Hatherley-Greene
Tim Hatherley-Greene
Chief Operating Officer
This is the question I get asked most often by AI programme leaders. "We're getting pushback from the operations team. Do we push forward or do we wait?" There's no universal answer. But after enough rollouts, I've developed a decision framework that's wrong less often than my instinct alone.

The Decision Framework

Resistance during an AI rollout is always information. The question is what it's telling you. Sometimes it's telling you the team isn't ready, and pushing will cause damage. Sometimes it's telling you the team is uncomfortable with something new, and waiting will just delay the discomfort without resolving it.

Push When:

The resistance is fear-based, not evidence-based. "I'm worried this will replace me" or "I don't trust AI" are fears that don't resolve by waiting. They resolve through experience. Controlled, supported exposure to the AI system with easy override capability builds trust faster than any amount of reassurance.
The pilot data is clear. If you have evidence from a pilot that the system works and delivers value, the case for deployment is stronger than the case for waiting. Resistance from a team that hasn't experienced the system is less informed than evidence from a team that has.
Leadership is aligned and committed. If the executive sponsor is firm, the business case is approved, and the resources are committed, waiting without cause creates doubt. "If leadership believed in this, why did they pause?" A delay that's not clearly explained reads as uncertainty.
The resistance is concentrated in a small group. If 80% of the team is open and 20% is resistant, pushing forward with strong support for the 20% is usually more effective than waiting for unanimous buy-in that may never come.

Wait When:

The resistance is evidence-based. "The system gets our document types wrong 30% of the time" or "this doesn't integrate with our workflow" are technical objections that need solving, not overcoming. Pushing through evidence-based resistance produces workarounds, not adoption.
The team is already overwhelmed. Change fatigue is real. If the team is simultaneously dealing with a restructure, a compliance audit, and quarter-end pressure, adding an AI rollout creates a breaking point. Wait for capacity.
The data isn't ready. If the AI system will perform poorly with the team's actual data (not pilot data, actual production data), deployment will confirm the team's scepticism. First impressions matter. A poor first experience is harder to recover from than a delayed start.
Leadership is divided. If the executive sponsor supports it but the team's direct manager is sceptical, pushing creates a credibility gap. The team follows their direct manager's cues, not the distant executive. Get the manager on board first.
The difference between productive discomfort and destructive pressure is whether the team has the support to get through it. Push with support. Never push without it.
Tim Hatherley-Greene
Chief Operating Officer

The Grey Zone

Most situations aren't clearly push or wait. They're somewhere in between. For those:
Deploy to willing volunteers first. Let the enthusiastic minority go first. Their success creates evidence that shifts the wider team from resistant to curious.
Set a time-bound experiment. "We're going to try this for four weeks with full support. At the end, we'll evaluate together." Time-bounded experiments reduce the perceived risk. "If it doesn't work, we'll stop" is a powerful de-escalator.
Increase support, not pressure. If the team isn't adopting, the instinct is to apply pressure. More deadlines, more mandates, more reporting. This almost always backfires. Instead, increase support: more hands-on help, more customisation to their workflow, more listening to their concerns.
Check your own bias. AI programme leaders want the rollout to succeed. That desire can read resistance as irrational when it's actually informed. Before pushing, ask genuinely: "Is there something this team knows about their work that I'm missing?"

There's no formula that works every time. But the pattern I've seen is consistent: pushing with strong support and clear evidence works. Pushing without support creates resentment. Waiting without a plan creates drift. And the hardest but best option is usually a middle path: move forward with those who are willing, support those who aren't, and build evidence that makes the case for the cautious.