ActivTrak has released the results of its Productivity Lab, which tracked 120,620 employees in 1,009 organisations over three quarters from the fourth quarter of 2025 to the second quarter of 2026. The data reveal that the highest productivity and work-health scores occur when AI tools are used at a moderate level, rather than being fully embedded in every workflow.
What the study found
The lab categorised AI use into three stages. In Stage 1, 27% of workers used AI like a search engine for research and summarisation. Stage 2, involving 14% of employees, saw AI assisting with drafting content, generating ideas and completing routine tasks that were later validated. Only 2% reached Stage 3, where AI became an integral part of daily workflows. Overall, 43% of the workforce used AI in any capacity.
Productivity and health metrics rose as employees moved from little or no AI use to regular, task-level assistance, peaking when about three-quarters of the workforce adopted AI. Once AI was embedded in workflows, utilisation dropped by roughly five percentage points, returning to levels similar to those who barely used AI.
Why the findings matter
Traditional AI maturity models focus on licence counts or login frequency, measuring deployment rather than impact. The ActivTrak study instead looks at behavioural data, showing that deeper AI integration does not automatically translate into higher output. Companies that push every employee toward the most powerful models risk runaway costs, more tokens, larger models and greater infrastructure spend, without proportional gains.
Moreover, when AI is adopted without redesigning underlying processes, organisations may create sophisticated but isolated workflows that do not improve overall efficiency. The result can be higher expenses and a disconnect between individual task optimisation and broader business objectives.
What organisations should do next
Leaders are advised to map existing workflows before introducing AI tools, matching the right model to the right task. For example, a sales representative can generate a quote from multiple systems with a single prompt, saving time without needing the most advanced model.
ActivTrak's own operations team discovered that employees were routinely using the newest, most expensive AI model to rewrite routine customer emails, a task that does not require such sophistication. The company responded by creating internal guidance to help staff select appropriate models, curbing unnecessary spend.
Because AI adoption is a durable change, the study shows that 82% of employees who start using AI continue to do so, and those who move beyond casual use tend to stay at that level. Pushing everyone to the deepest tier may lock firms into a plateau where productivity gains stall while costs rise.
Going forward, firms should aim for the sweet spot identified in Stage 2, where AI eliminates repetitive work without over-automating. By monitoring usage patterns and aligning AI deployment with clear business goals, organisations can reap the competitive advantage of AI while avoiding hidden expenses and workflow fragmentation.
Looking ahead
As AI tools become more accessible, the pressure to adopt them at scale will increase. The ActivTrak findings suggest that a measured approach, focusing on task-level assistance and ensuring that AI complements, rather than replaces, existing processes, will be the most sustainable path for European companies seeking to boost productivity and protect employee well-being.