Accenture rolls Microsoft Copilot out to all 743,000 employees

TL;DR:

• Microsoft is deploying Copilot 365 to all 743,000 Accenture staff, scaling up from a 2024 plan to reach 300,000 — the largest single enterprise commitment to the assistant to date.
• Accenture says 97% of staff in an earlier deployment reported some routine tasks running up to 15 times faster, while a NBER survey of 6,000 senior executives in February found nearly 90% saw no impact of AI on employment or productivity over three years.
• The deal is a marquee Microsoft reference win — and a useful real-world test, given Accenture UK’s significant footprint, of whether large-scale Copilot rollouts deliver the productivity gains big consultancies are now selling to UK SMEs and corporates.

Microsoft is rolling out its Copilot 365 assistant to every one of Accenture’s roughly 743,000 employees, in what is now the largest enterprise commitment to the chatbot. The companies disclosed the scope but not the financial terms in a joint statement on Monday. The agreement materially expands Accenture’s earlier 2024 plan to deploy Copilot to up to 300,000 staff, and gives Microsoft a high-visibility reference customer at a moment when its own shares are down 12% year-to-date — a quarterly drop, January-March, that was Microsoft’s worst since the 2008 financial crisis.

The deal matters commercially because just over 3% of Microsoft’s 450 million-plus 365 enterprise users currently pay for Copilot at $30 per user per month. Accenture has emerged as one of the most aggressive corporate adopters of generative AI and, according to media reports, has tied senior promotions to AI usage. Accenture said its initial Copilot deployment “paid off”: a self-reported survey of 200,000 users found 97% saying Copilot helped them complete some routine tasks up to 15 times faster, with 53% reporting major productivity gains. CEO Julie Sweet said her teams “are already doing higher-value work because of it”.

The productivity claim under pressure

Those numbers sit alongside more cautious external evidence. A National Bureau of Economic Research survey of nearly 6,000 senior executives across the US, UK, Germany and Australia, published in February, found nearly 90% saying AI had no measurable impact on either employment or productivity over the past three years. The Whitehall Copilot trial that prompted HMRC’s 28,000-staff rollout reported a more modest average of 26 minutes saved per user per day — useful, but several orders of magnitude shy of “15 times faster”. The tension between vendor and customer self-reports and external academic data is now the live debate inside many UK enterprise procurement teams.

Microsoft’s product strategy is also broader than OpenAI alone. Charles Lamanna, who runs Microsoft’s M365 apps and Copilot platform, told Reuters that supporting multiple AI models — including Anthropic’s — and adding tools like “Critique” (one model checking another’s output) is helping demand. The Accenture announcement landed on the same day as the company’s formal end of exclusivity with OpenAI, which clears the way for Microsoft to push Anthropic harder inside Copilot.

Why UK readers should care

Accenture UK is one of the largest consultancies in the country, with a heavy presence across UK financial services, public sector and retail clients. The firm is effectively turning itself into Microsoft’s productivity laboratory — and into the case study Accenture itself, and rivals like EY (which this week launched UK Forward Deployed Engineer roles for AI delivery), will use to sell large-scale Copilot programmes to UK boards. UK SMEs and corporates considering similar rollouts should treat Accenture’s “97% / 15x” survey numbers as marketing input rather than independent evidence — and ask suppliers for unit-economics, time-tracking data and quality outcomes alongside any productivity claim.

Looking forward

The Accenture rollout is essentially a high-stakes natural experiment in scale-Copilot productivity. If the firm can publish hard, externally verifiable outcomes — billable hours per project, win rates, client satisfaction — over the next 12 months, it will become a watershed moment for enterprise AI adoption. If it cannot, the gap between vendor self-reports and the NBER-style external surveys will only widen, and UK procurement teams should plan accordingly.