Successify Media

Two in three Copilot licences sit unused

About 36 % of people with a Microsoft 365 Copilot licence use it regularly, and small businesses are far below that. The blocker is usually permissions and use cases, not training.

Manish Kumar, founder ·

Microsoft 365 Copilot Business lists at about AU $31.40 per user per month ex GST in Australia. Buying it is a five-minute decision. Getting people to use it is the part nobody budgets for, and the numbers on that are not flattering.

35.8 %

of employees with Copilot access use it regularly. The other two in three are paid for and idle.

Source: Recon Analytics, January 2026

That figure holds up across the wider 2026 survey work, which puts median active use among licensed users in the mid-30 percent range. First-year figures below 40 % are normal before anyone intervenes deliberately. This is not a Microsoft problem or a product problem. It is what happens when a licence is treated as the rollout.

Smaller firms are further behind, not further ahead

The intuition is that a small business should adopt faster: fewer people, less politics, no change board. The data says the opposite.

The smaller the firm, the less Copilot gets used

Organisations with an active Copilot deployment, by company size, 2026

  • Small business, under 500 staff12 %
  • Mid-market, 500 to 4,99931 %
  • Fortune 50064 %

Source: Stackmatix, Microsoft Copilot adoption statistics 2026

Large firms are not better at this because they are clever. They are better at it because somebody's job is to run the rollout: pick the use cases, fix the permissions, train the people, report the numbers. In a 40-person firm nobody owns that, so the licence lands in everyone's toolbar and the toolbar is where it stays.

What the idle two-thirds actually costs

The arithmetic is unpleasant and worth doing before renewal.

Twenty-five seats, one year, at the published Australian price

25 × AU $31.40 × 12 months = AU $9,420 ex GST, split by the 35.8 % regular-use rate

Seats in regular use
about AU $3,370
Seats paid for and idle
about AU $6,050

Illustrative arithmetic on published list pricing, not a measurement of any particular business.

Source: Microsoft list pricing, July 2026, and Recon Analytics, January 2026

Six thousand dollars a year is roughly a fixed-scope engagement to fix the thing. That is the arbitrage most firms miss: the cost of doing adoption properly is usually less than the cost of not doing it, and it is already being spent.

The blocker is rarely training

When a rollout stalls, the reflex is to book another training session. Sometimes that is right. More often, three other things are in the way.

Permissions, first and worst. Copilot surfaces whatever a person can already open. It does not invent access; it operationalises the permission debt already sitting in Microsoft 365. One assessment puts the average enterprise tenant at 150 to 300 overshared SharePoint sites. The four patterns that cause most of it are the "Everyone except external users" group, broken permission inheritance, anonymous or organisation-wide sharing links, and legacy "All Employees" security groups. Sharing at the file and folder level with "Everyone except external users" is the worst of them, because it does not show up in site-level reports.

The consequence is not usually a breach. It is that somebody asks Copilot a reasonable question, gets back the salary review or the redundancy list, and the rollout quietly dies that afternoon. As one write-up put it, Copilot did not overshare your data, your permissions did.

Second, no use case that survives contact with the job. "Use it for email" is not a use case. "Draft the first reply to a quote request using the price list and last month's similar quotes, then check these three things before sending" is. Three of those per role beats an hour of feature demonstration.

Third, no measurement. If nobody reports active users and hours at week 4 and week 12, the licence renews on hope. Microsoft 365 gives you the usage report; the honest version pairs it with a two-question staff pulse, because seat activity is not the same as value.

The order that works

Permissions, then use cases, then training, then measurement. In that order.

Fixing permissions after training means the first thing people learn is that the tool is unsafe. Training before use cases means people learn features they have no reason to use. Measuring nothing means you find out at renewal.

This is also why "let's do a pilot with the keen people" tends to mislead. The keen people would have adopted anything. The number that matters is what the indifferent majority do in week 12.

If you already pay for it

Three things you can do this month without any help.

Pull the Copilot usage report and get the real active-user count, not the licence count. Run a permissions review of your top ten most-shared SharePoint sites and look specifically for "Everyone except external users" at file and folder level. Then pick one role, write three genuine tasks for it, and measure the time those tasks take today so you have a baseline to compare against.

If active use is under about a third and it has been more than a quarter since the licences landed, the licence is not the problem and another training session will not fix it.

Sources

More posts

Not sure where to start? Book a call.

Twenty minutes. We confirm fit, propose a fixed scope and a date, and tell you if you do not need us.

Book a call

+61 490 083 850 · Wollongong, NSW