Most CIOs I talk with are somewhere in the same stretch of road. The Copilot licences are deployed. The board is asking what the investment has returned. And at some point in the budget cycle, adoption became an internal responsibility, because the organisation already has an L&D function and an internal communications function.
There is real logic in that. L&D and internal communications own the LMS, the intranet, the manager cascade and, most importantly, the trust of the business. When adoption lands with them, it lands with the people employees already listen to.
What I want to share here is what happens next, based on what we see inside enterprise rollouts, and what those teams typically need in order to succeed at scale.
Where internal-only adoption tends to slow down
The obstacle is almost never the technology. It shows up in four places.
Capacity. L&D is usually mid-way through an LMS migration or a leadership program. Comms is running open enrolment or a reorg announcement. AI enablement becomes the thing that gets attention on Friday afternoons.
Context. Generic "here is Copilot in Word" content leaves finance, legal and operations to work out the translation themselves. Most people quietly decide it does not apply to their day.
Cadence. A launch email and a single webinar create a spike. Behaviour change needs reinforcement for weeks after the spike fades.
Baseline. When nobody captured week zero, nobody can defend week twelve. This is the one that costs CIOs the most, because it removes the ability to prove anything to the executive committee.
The number that follows this pattern is familiar. Licences deployed in the nineties, weekly active usage far below that, and a growing quiet reliance on unapproved AI tools because the sanctioned one never became obvious.
What actually helps these teams
Every organisation is different, and the sequence below is what we have seen work most consistently. Some of it you can build yourselves. Some of it is worth partnering on. The judgement call belongs to you.
Give L&D a curriculum instead of a curriculum project
Building AI learning content from scratch is where internal teams lose a quarter. What works better is starting from an existing role-based library and tailoring it.
- A short foundations module for everyone, so the baseline is consistent
- Role-based sessions that use real work, for example Copilot in Excel for analysts, meeting and delegation workflows for managers, document review workflows for legal and HR
- Your AI acceptable use policy converted into a brief eLearning module, reviewed by IT and published through the LMS
- Train-the-trainer, so delivery capability stays in the building once the program matures
Hand internal comms a campaign at the editing stage
The most useful thing you can give a communications team is a cadenced campaign already mapped to your channels, so their work is adapting rather than originating.
| Phase | What goes out |
|---|---|
| Weeks 1 to 2, launch | Executive sponsor message, manager nudge, Teams or Viva post, training invitation |
| Weeks 3 to 10, execution | Weekly prompts and tips in the flow of work, self-serve learning assignments, a challenge such as a Prompt-a-Thon, champion nudges |
| Weeks 11 to 12 and beyond | Monthly adoption report, sponsor sync, success stories, feedback loop into the next phase |
The reinforcement layer in the middle rows is the part most rollouts underinvest in, and it is where habit forms.
Build a champions network that runs on peer proof
Adoption spreads faster when the example comes from inside the function. A workable model is a branded network with a defined role and a monthly cadence of 60 to 90 minute sessions where employees share what has genuinely worked for them. Comms amplifies the stories, L&D feeds the group new material, IT stays close to the friction they surface.
Report in the language the executive committee uses
Monthly reporting drawn from Microsoft 365 admin and Viva Insights data, organised into eight categories:
- Copilot readiness, eligible versus licensed users
- Activation, monthly and daily active users, prompts per user
- Agent activation, agents created and actively used
- Breadth of adoption by app
- Impact, estimated time saved and financial impact
- Sentiment, survey-based confidence and perceived value
- Operating cost, helpdesk volume and external AI tool spend
- Training effectiveness, coverage, completion and perceived usefulness
Close every report with a short prioritised recommendation list. The value sits in what you change next month.
What this has looked like in practice
Three examples from programs we have run:
- A global maritime organisation ran a single hands-on Prompt-a-Thon. High-frequency Copilot users tripled and the share of participants reporting 15 or more minutes of daily time savings rose by 275 percent. Nearly all participants felt prepared to use Copilot afterwards, up from around half beforehand.
- A North American food and beverage manufacturer reached 96 percent active usage within a single quarter.
- A Canadian municipality moved weekly active Copilot users from 19 percent to 78 percent across 2,200 staff over a twenty-week program, governed by a sponsor coalition and tracked against licence value.
In each case, training, communications, champions and measurement ran as one connected adoption program with a single sponsor accountable for the outcome.
Thinking through build, buy or blend
If you are being asked to deliver this internally, it is worth being explicit with your executive team about the trade-offs.
| Model | Fits when | Trade-off |
|---|---|---|
| Fully internal | You have dedicated adoption headcount and existing AI curriculum | Longer time to value, and L&D or comms capacity comes out of other committed priorities |
| Fully outsourced | Internal capacity is genuinely unavailable | Weaker cultural fit, and capability leaves when the engagement ends |
| Blended | You have credible L&D and comms teams already running at capacity | Requires coordination discipline, and gives you speed while capability stays in house |
The decision worth surfacing to leadership: the real question is whether your internal teams receive a tested program to run or spend two quarters designing one while the licence clock runs.
A 90-day sequence you can adapt
- Weeks 1 to 2. Discovery with IT, L&D, comms and two priority business units. Baseline survey plus a current usage snapshot. Agree the single metric that defines success.
- Weeks 3 to 4. Executive sponsor message goes out. AI policy module published. Champions recruited and briefed.
- Weeks 5 to 10. Role-based learning assigned, live hands-on sessions on a repeating cadence, weekly tips campaign running, first challenge launched.
- Weeks 11 to 12. Follow-up survey, first monthly report to the sponsor, success stories published, next-quarter priorities set.
Ninety days gets you a defensible number. Sustained habit takes longer, which is why the reinforcement cadence continues well past week twelve.