The Department for Business and Trade handed out 1,000 licences between October and December 2024, with most going to volunteers and about 30 per cent to randomly selected staff. Of those, 300 allowed their data to be analysed.

Overall, 72 per cent of users said they were satisfied with Copilot, especially when it came to routine drudge like summarising meetings or knocking out emails. Many complained when the trial ended, though the real-world productivity picture was rather less flattering.

The trial showed that two-thirds of employees used Copilot at least once a week and 30 per cent at least once a day, which is hardly great value considering the price tag. On average, users performed just over one Copilot action per working day. Word, Teams and Outlook were the most popular, while Loop and OneNote barely registered. PowerPoint and Excel saw occasional bursts of activity but no serious adoption.

Users saved tiny slivers of time on emails and report summaries, with slightly better quality compared to colleagues without Copilot. The shine wore off quickly in Excel, where Copilot made analysis slower and less accurate. PowerPoint users were about seven minutes faster on average, but the slides were noticeably worse, requiring manual corrections afterwards.

Interviewed staff said Copilot made routine admin easier, which freed up time for training or even taking a walk at lunch. But the assessment concluded that this didn’t make them more productive overall.

Vole has been eager to sell Copilot licences to government and enterprise customers with the promise of efficiency gains. This report suggests the benefits are far harder to measure and in some cases, Copilot simply makes things worse.