ActivTrak: heavy AI users work up to 90 minutes longer a day, CIO reports

Employees who use AI most heavily are working up to 90 minutes longer each day, according to new research from workplace-analytics vendor ActivTrak, reported by CIO on 3 November 2026. The same company's previous report said the opposite: that AI users were working slightly less. For anyone building a productivity case for an AI rollout, the pair of reports is a warning about how the number gets made.

What ActivTrak's latest data shows

CIO says the new findings come from behavioural work data on 9,140 knowledge workers across 467 enterprises, collected from 1 April to 25 September 2026. According to that report, heavy AI users work between 8.42 and 9.08 hours a day, against 7.58 hours for those who do not use AI.

The report also describes more fragmented work. Heavy AI users had focus sessions 22% shorter than light users (37.8 minutes versus 48.2) and switched between apps 15% more often. People using AI embedded in the tools they already work in switched apps 0.6 times an hour; those using web-based AI tools switched 1.7 to 4.2 times an hour, which ActivTrak calls a "swivel-chair tax". Light users spent more active time in core job-specific apps (81.4% against 78%).

It contradicts the same company's earlier report

ActivTrak's earlier report, published earlier in 2026 and based on 443 million hours of work by 163,638 employees across 1,111 organisations from January 2023 to December 2025, found average workdays had fallen by roughly 9 minutes (2%) with AI. It also found daily productive time up 19 minutes (5%) to 6 hours 36 minutes, and productive sessions up 13%. In the same report, focus efficiency fell to a three-year low of 60%, and multitasking rose 12%, or more than 10 minutes.

CIO says the gap may owe something to the different time spans and sample sizes: five months and 9,140 people against three years and 163,638. ActivTrak's Sarah Altemus is quoted as saying the larger issue is how organisations use the capacity AI creates: when staff finish tasks faster, managers may assume they have room for more, and without workload boundaries the saved time can turn into extra work.

What the measurement leaves out

CIO quotes Sanchit Vir Gogia, chief analyst at Greyhound Research, who notes that ActivTrak sells the capability that produces these measurements. He says recorded AI use depends on what the instrumentation can detect: an employee can use AI and still appear among the non-users, work on an unmanaged personal device sits outside the data, and an API or background agent can finish work without producing a desktop session. App-switching is also easier to observe than attention-switching inside one application.

His test for the board, as quoted: it should be able to say whose day improved and whose work made that improvement possible, because an improvement per paid hour can conceal effort left off the timesheet.

What to do with it

• Treat any single time-saving figure, including a vendor's own, as one measurement with a stated method, not a result.
• Define the metric before the rollout: paid hours, focused hours, throughput or quality each give a different answer.
• Prefer AI embedded in the apps people already use if switching cost matters; this report associates it with far fewer app switches than web tools, though it does not show that embedding caused the difference.
• Decide what happens to freed-up time. If it silently becomes new work, expect longer days rather than the saving your business case promised.