What the surveys say
| Finding | Figure |
|---|---|
| Organisations piloting Microsoft 365 Copilot (2025) | 40%, with a further 19% planning to |
| Of those that finished a pilot, moving to a larger deployment in 2025 | 5% |
| Organisations that delayed rollout by three months or more over oversharing concerns (2024) | 40% |
| Respondents saying deployments needed more change management than expected (2024) | 73% |
| Respondents saying Copilot already gives “significant” value (2024) | 3% |
Sources: Gartner, 2025 Microsoft 365 and Copilot Survey — 187 IT leaders (reported by techpartner.news, 26 June 2025); Gartner survey of 132 IT leaders (reported by Computerworld, 27 September 2024). Checked 5 October 2026.
The same 2024 survey found 98% of respondents saying employees were excited to use it. Enthusiasm is not what is missing.
Why it stalls
| Where it stalls | What actually happened |
|---|---|
| Oversharing | Copilot answers from what each user can already open. Years of “anyone with the link” sharing become visible overnight, and security pauses the rollout |
| Value | Licences were bought per seat; nobody agreed what a better week looks like, so nobody can show one |
| Adoption | Usage spikes in the first weeks, then falls back to the people who were already enthusiastic |
| Data | The business case assumed clean, current documents. The real estate is duplicated and stale in places |
Microsoft’s own deployment blueprint starts in the same place: its first pillar is remediate oversharing, before guardrails and regulation (Microsoft Learn, updated 18 August 2026).
A four-step recovery plan
- Pick one workflow. Not “productivity”. One task — drafting first replies to customer queries, summarising contract changes, preparing a weekly report — where a better answer is worth real money or time.
- Fix access for that workflow first. Find out which sites and files it draws on, and who can see them. Fixing one workflow’s sources is a project; fixing the whole tenant is a programme.
- Agree what “working” means. A test set from your own documents, a measure (time saved, errors caught, answers accepted without edits), and the level that justifies the licences.
- Measure, then decide. If it clears the bar, widen from that workflow. If it does not, you have a number — not an opinion — to take to the licence renewal.
When the workflow needs more than Copilot
Some workflows need answers that are the same every time, traceable to a rule, and checked against your systems of record — an approval, a calculation, a compliance decision. A general assistant is the wrong tool for those, however good it is at drafting.
For those, Pricient quotes IOanyT’s Proof Sprint: in three to four weeks, one workflow is compiled — LLMs write and test the logic at build time, and deterministic code runs it, so the same input gives the same answer (what Compiled AI means). It ends with a consistency and accuracy report on your own data and a written go / no-go, you keep the code either way, and the price is fixed and agreed at scoping.
Who signs off
The team that builds or configures the assistant should not be the one that certifies it. Independent review exists for exactly that reason.
How Pricient handles a stalled Copilot rollout
Describe what stalled and which workflow mattered most. A person reads it; this lane’s own intake is not live yet. You receive the specification and a gap list — data, access, integration, success measure — before anyone quotes a sprint. We do not make you compliant, and we do not sell licences; we help you find out whether one workflow is worth them.