Your Company Paid for Microsoft Copilot. 64% of Your Team Isn't Using It.
- May 24
- 5 min read
Hong Kong organisations are buying Copilot licences faster than their employees can figure out what to do with them. The problem isn't the tool. It's the missing training layer that Microsoft doesn't provide.
Somewhere in your organisation right now, there is a Microsoft 365 Copilot licence sitting idle.
Your IT team deployed it. Your CEO mentioned it in a town hall. Someone sent an all-staff email with a link to a Microsoft tutorial video. And then — nothing. People went back to doing things the way they always have.
This is not a Hong Kong problem. It is not even a Copilot problem. But it is happening at an accelerating rate across the city, and it is costing organisations real money every single month.
35.8% The Microsoft 365 Copilot workplace adoption rate globally — meaning fewer than 4 in 10 employees with paid access actively use it. - Avantiico Copilot Adoption Guide, 2026
Compare that to ChatGPT, which has an 83.1% workplace conversion rate among users who have access to it. The gap is striking — and it is not because Copilot is a worse product. Copilot is embedded directly in the tools your team already uses every day: Outlook, Teams, Word, Excel, PowerPoint. ChatGPT requires an extra tab.
The real difference is familiarity, confidence, and — critically — training.
What Is Actually Happening in Hong Kong
The local data makes the picture sharper. According to the HKPC AI Readiness in Workplace Survey 2025, which gathered responses from 800 Hong Kong companies, talent shortage — not technology access — is the single biggest barrier to AI adoption. The tools exist. The licences are being purchased. The human capability to use them is not keeping pace.
Hays Asia's 2025 salary research confirms this precisely: 53% of Hong Kong professionals report using AI tools at work, but only 33% have received any formal training to do so. Nearly half avoid AI tools entirely because they do not feel adequately supported.
The uncomfortable arithmetic: If your organisation pays HK$250 per user per month for Copilot licences and 64% of your team is not using them, you are burning a substantial portion of that budget on shelfware — while the adoption gap quietly widens.
KPMG's March 2026 Hong Kong Employment Outlook found that AI literacy is now a priority capability for 47% of local employers — up from just 20% a year earlier. The urgency is real and it is accelerating. But most organisations are still trying to close a human gap with a technology purchase.
The Three Reasons Copilot Adoption Stalls
The pattern is consistent across industries and company sizes. Copilot gets deployed. Adoption plateaus within weeks. The organisation quietly writes it off as a "phase one pilot." Here is why this keeps happening:
The announcement is mistaken for training
An email, a town hall, or a link to Microsoft's own tutorial library is not a training programme. It is information delivery. Behaviour change requires practice, repetition, and feedback — none of which a PDF can provide.
Employees do not know what Copilot is actually for
Most staff encounter Copilot as a generic AI chat panel. Without clear, role-specific use cases — "here is how you, in your specific job, use this today" — the tool feels abstract. Abstract tools get ignored.
Fear of looking incompetent in front of colleagues
Generative AI produces visible, shareable output. Employees who are unsure of their prompting skills are afraid to produce bad results where others can see them. This fear is rarely acknowledged — and it kills adoption faster than any technical barrier.
The "Silicon Ceiling" Your Managers Cannot See
BCG's 2025 global AI at Work survey identified what it calls a silent adoption divide: while more than three-quarters of leaders and managers use generative AI several times a week, regular use among frontline employees has stalled at 51%. BCG named this the "silicon ceiling" — and it is particularly visible in markets like Hong Kong where AI tool deployment has outpaced structured enablement.
The trap is that managers who use Copilot daily have lost the ability to perceive what it feels like to be intimidated by it. They see a simple, helpful tool. Their team sees an unfamiliar interface where mistakes are public. Both groups are right. Neither group is being served by the current approach.
PwC's January 2026 Hong Kong Workforce Hopes and Fears Survey put a number on the consequence: while 61% of Hong Kong employees now use some form of AI at work, PwC described the overall situation as AI being "underutilised" — with structured upskilling and proactive change management identified as the essential missing components.
What Actually Works: The Evidence
The productivity case for Copilot — when properly adopted — is well-established. A UK government pilot with 20,000 users found employees saved an average of 26 minutes per day. Microsoft's own legal team completed tasks 32% faster with a 20% accuracy improvement. PwC's 2025 Global Workforce Hopes and Fears Survey found that daily generative AI users are 34 percentage points more likely to report productivity gains than infrequent users.
The word "daily" matters. Occasional use does not generate compound returns. Daily use — the habit — is where the productivity gains actually live. And habits are built through structured, repeated, socially reinforced practice. Not through self-directed exploration of a product your IT team deployed three months ago.
DataCamp's 2026 State of Data and AI Literacy report delivered the clearest evidence yet: organisations with a mature, enterprise-wide AI upskilling programme report nearly double the rate of significant positive ROI compared to those without one. The training is not a nice-to-have. It is the primary driver of return.
What a Proper Copilot Enablement Programme Looks Like
There is no shortage of generic AI courses available online. The reason they do not close the adoption gap is structural: they are passive, individual, and disconnected from real workflows. What works is different in every dimension.
Start with role-specific use cases, not tool features.Finance teams need to see Copilot summarise board papers in Outlook. HR teams need to see it draft job descriptions and analyse survey data. Marketing teams need content briefs generated in seconds. Generic prompting tutorials produce generic engagement.
Train in teams, not as individuals.When an entire team learns together, shared language emerges. Colleagues can help each other, compare outputs, and normalise AI use as a group behaviour rather than a solo experiment. This is the single biggest driver of sustained daily adoption.
Address the fear directly.Name it in the room. "Some of you are worried about looking incompetent in front of colleagues. That is normal. Here is a safe space to produce bad outputs, learn why they were bad, and try again." Organisations that skip this step see adoption plateau after the initial novelty wears off.
Connect to measurable workflow outcomes."How much time did this save you on that task?" is a question that transforms Copilot from a novelty into a tool. Post-session measurement — even informally — creates the feedback loop that builds habits.
Follow through with manager reinforcement.The half-life of a one-off training session without management reinforcement is approximately two weeks. Managers who model Copilot use in meetings and check-ins are the single most powerful adoption accelerator available.
The Copilot Question Hong Kong Companies Need to Ask Now
Microsoft's March 2026 licensing updates — including expanded bundling of Sales, Service, and Finance Copilots into the core offering — make this conversation more urgent, not less. The tool is becoming cheaper and more pervasive. The competitive advantage will not come from having access to it. It will come from the teams that know how to use it confidently, daily, and in ways that directly accelerate their work.
If you have purchased Copilot licences and you are not seeing measurable adoption, the question is not "is this tool worth it?" The question is: "Did we give our people a real reason, a real skill, and a real safe space to use it?"
For most Hong Kong organisations, the honest answer is no. And that is a solvable problem — but not with another email.
