Most end-of-year reviews look back at targets and forward at a slide deck. People sit through them, then return to work unchanged.
A review people value does something different. It helps the team see how it worked together, not only what it delivered.
This guide gives you a 90-minute format for December. It draws on 2026 survey evidence about what shapes working life, and on research into how leaders make it safe to speak.
Why most year-end reviews fall flat
Year-end sessions tend to fail for three reasons. They repeat numbers everyone already knows. They reward the loudest voices. And they end with actions nobody owns.
There is also a mismatch of purpose. Leaders want a record of results. Team members want to know whether their effort was seen, and whether next year will be easier.
A useful review serves both purposes. It simply starts with the second one.
What the 2026 evidence says matters to people
The CIPD Good Work Index 2026 was published in July 2026. It is based on a YouGov survey of 5,074 UK workers.
It found that line management, leadership, skills development and meaningful work had the greatest effects on the five business outcomes it tracks. Contract type, by contrast, showed no statistically significant link to most of them.
The same report found that 34% of employees use generative AI in their role at least once a month. Two-thirds of those users have had no AI training at all, and 45% of employees do not trust its output to be accurate.
Gallup's 2026 data point the same way. In AI-adopting organisations, 78% of employees who strongly agree their manager supports AI use are frequent users. Among those who do not strongly agree, the figure is 44%.
The pattern is consistent. How people are led shapes their year far more than the tools they were given.
A 90-minute review in four parts
Keep the session short and the structure visible. Send the four questions a week before, so quieter people can prepare.
- What went well, and who made it possible? Spend 20 minutes here. Ask for specific moments and name contributions; recognition that cites real work lands, general thanks does not.
- What got in the way? Spend 25 minutes. Focus on how work flowed: hand-offs, meetings, approvals and tools. Ask what made the work harder than it needed to be, not who dropped the ball.
- What did we learn about working with AI? Spend 20 minutes. Ask where AI helped, where it created rework and where people were unsure whether they were allowed to use it.
- What will we change in January? Spend the last 25 minutes. Agree no more than three changes, each with an owner and a date to check progress.
Build this capability in your organisation
Growth Performance designs bespoke programmes that turn insight into measurable behaviour change.
Use colour energies to hear every voice
Insights Discovery describes four colour energies: Cool Blue, Earth Green, Sunshine Yellow and Fiery Red. They describe preferences, not ability. They work best as a shared language for how people like to contribute.
In a review, that language helps you design for range. People with a Cool Blue preference often want data and time to think, so share material in advance. People leading with Earth Green may hold back concerns unless asked directly and kindly.
Sunshine Yellow preferences tend to bring energy and ideas, so give them room to build on others. Fiery Red preferences usually want decisions, so close each part with a clear choice.
If you hold the budget and want to try the model first, the free Insights Discovery profile is a low-effort start.
Make it safe to say what did not work
Edmondson's 1999 study of work teams described psychological safety as a shared belief that the team is safe for interpersonal risk-taking. A year-end review tests that belief, because the useful content is often the uncomfortable part.
A 2026 study by Resendiz and colleagues interviewed eight senior research leaders about how they build that safety. Their behaviours fell into five themes, including self-awareness and growth, communication and transparency, and accountability and integrity.
It is a small study within one university. Treat its themes as practical prompts rather than proof.
In practice, go first. Name one thing you would do differently and one decision you got wrong. Then thank the first person who raises a problem, before you respond to its content.
Bring AI into the review honestly
Many teams now have quiet, uneven AI use. Some people rely on it daily; others avoid it or keep it to themselves. A review is a good moment to make that visible without blame.
Before the session, ask everyone to complete the Shadow AI Exposure Check. It takes about four minutes and gives you a team picture to discuss in part three.
Then turn what you learn into first rules for the year with the AI Team Charter. It records the tools you agree to use, the tasks where you do not use AI, and how you check output before it is used.
For why manager support decides whether AI use spreads, see the Adoption condition.
Common mistakes to avoid
- Turning the review into a performance appraisal. Keep individual ratings for one-to-ones.
- Running out of time before the actions. Protect the last 25 minutes.
- Agreeing ten changes. Three owned changes beat ten good intentions.
- Leaving AI to "any other business". It belongs in the main agenda.
If you would rather have the session facilitated as part of a wider team day, see our team events for 2026 and 2027.
Try this now
- Send the four review questions to your team today.
- Ask everyone to complete the Shadow AI Exposure Check before the session.
- Book 90 minutes in the diary and protect the last 25 for actions.
