← Cases
AustraliaPublic sector2024

Australian Government (Digital Transformation Agency)

Whole-of-government Microsoft 365 Copilot trial

Australia's six-month Copilot trial across federal agencies found moderate use, with only a third of staff using it daily

In progressCompany's own claim

Problem

Federal agencies wanted to know whether a generative AI assistant built into everyday office software would save public servants time, and what risks it would bring for records, privacy and skills.

Approach

The Digital Transformation Agency coordinated a trial of Microsoft 365 Copilot from January to June 2024 with 5,765 licences spread across dozens of agencies, which nominated the participating staff. The evaluation, published in October 2024, relied on surveys and interviews in which staff assessed their own use and time savings. It made no specific recommendation to embed Copilot in daily operations.

Results

1/3share of participants who used Copilot dailyCompany's own claim[1][3]
5,765Copilot licences in the trialCompany's own claim[1]
1.1 h/dayself-reported time saved on summarising informationCompany's own claim[2]
-54%drop in staff belief that Copilot would reduce time in meetingsCompany's own claim[1]

Takeaway

Staff reported saving time on summaries and first drafts, and training was linked to greater use. Savings were self-reported by agency-selected staff, with no direct measurement.

Categories

AI technology
Generative AI
Evidence
Company's own claim

Sources

Source summaries

[1]iTnews23 Oct 2024

iTnews reports the DTA's evaluation of 5,765 Copilot licences used in the first half of 2024. About two thirds of participants used the tool a few times a week or less, and belief it would cut email time fell 32% and meeting time 54%. The evaluation relied entirely on self-assessment, and executives were over-represented.

[2]The Decoder28 Oct 2024

The Decoder reports the evaluation across 56 agencies. Participants estimated time savings of 1.1 hours a day for summarising and one hour for document drafting, and 77% were positive about the tool. They raised concerns about data security, bias, loss of skills and effects on entry-level administrative roles.

[3]GovTech Review28 Oct 2024

GovTech Review reports that the trial ran in the first six months of 2024 and that usage was rated moderate, with a third of participants using Copilot daily. It says 46% found value a few times a week and that the largest efficiency gains were among junior and middle APS grades and ICT roles. It reports over 7,600 participants in more than 60 agencies.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026
Open case page ↗