Large rollouts are won in the readiness plan, long before go-live

Rollouts of Microsoft 365 and Windows 10 across tens of thousands of users, or change programmes spanning hundreds of banking institutions, teach one lesson clearly: the outcome is largely decided before go-live. The technology works on the day. Whether people use it well depends on what was done in the months before.

The readiness work is not glamorous. A stakeholder map shows who is affected and who influences them. An impact analysis lists what changes for each group, in terms of their own tasks. Readiness checks find out who is prepared and who is not. Training is sequenced so that people learn when it is relevant, not three months early. Local champions give colleagues someone to ask. Each is simple, and skipping any one shows up as a predictable problem later.

AI changes one thing about this. With earlier tools, adoption was visible: a login count or a feature report told you roughly where you stood. With AI, use is discretionary and mostly invisible. A person can hold a licence, open the tool twice and go back to the old way, or use it daily and carelessly, and the dashboard looks identical. The signals that matter are behavioural and psychological, such as trust, fear, habit and perceived permission, and they have to be measured directly.

That is the reasoning behind a diagnostic like the Organizational Adoption Profile, which scores a rollout across five adoption drivers before launch. A score tells a team where adoption is likely to leak, so the first effort goes to the weakest driver instead of to whichever activity is already scheduled.

Regulated environments add a further requirement. In a bank, staff need to know clearly what they are allowed to do with a tool, with which data, and who they can ask. Ambiguity there does not produce caution so much as silence, and silence looks like low interest in the metrics.

A test for any programme sponsor, asked a few weeks before launch: could you tell, from the data you have today, which teams are likely to adopt, which are likely to stall, and why? If not, the readiness plan is not finished.