53% of managers find original work harder to assess. Show them your decisions

A finished piece of work now lands in a manager's inbox already polished, and the manager cannot tell how much of it came from the person who sent it. TechCrunch reported on 21 October that more than half of senior leaders find it harder to evaluate their teams' work since adopting AI, citing a Nielsen survey of 1,701 directors and senior managers across seven countries, commissioned by monday.com. TokenPost's write-up of the same survey puts the figure at 53% who find it harder to assess employees' original contributions, and 78% who have already changed how they evaluate team members. The finished output has stopped being the evidence of your work. Managers are now trying to grade the decisions behind it, and most of those decisions never leave your screen.

What the new scorecard rewards

The survey names the new criteria. According to TechCrunch, managers who changed their approach now put more weight on how well staff write prompts, decide which tasks to hand to AI, and experiment with new ways of working. TokenPost reports that 47% say they spend more time reviewing and guiding their teams' output than before.

Delegation choices are invisible in a finished document. A clean board paper looks the same whether its author spent the afternoon testing the figures or pasted in a first draft and tidied the headings. The only way a manager can tell the difference is to spend more time reviewing, which is what nearly half now report doing.

That review time is about to get scarcer. TechCrunch and TokenPost both report that 58% of the managers surveyed expect to oversee more AI agents than employees within a year, and 35% are concerned about being held accountable for decisions those agents make. A manager carrying that load has less time to reconstruct how each piece of human work was made.

The people grading you delegate more than you do

Workday's AI@Work survey, published on 22 October, supplies the other half of the picture. In a September poll of nearly 6,000 workers, 88% of senior leaders said they had used agentic AI to complete tasks, against 54% of individual contributors. Among senior leaders, 44% report heavy AI use. Among individual contributors the figure is 9%. Senior leaders are 2.5 times more likely to say they understand AI agents well.

The evaluators are the heavy delegators. They know what a whole task handed to an agent looks like, and they know what checking it costs. When they score someone on deciding what to delegate, they are scoring against their own practice.

The gap runs through trust as well. Workday found that 32% of individual contributors trust their employer's approach to AI, against 66% of senior and executive leaders. Only 44% of employees say they can rely on AI output without checking it closely.

The obvious reading of these two surveys is that managers have a measurement problem and it is theirs to fix. The operator reading is different. Whoever makes their judgement legible first sets the standard the rest of the team is measured against.

AI does the routine work. You do the thinking. Your job now includes making the thinking visible.

Your Next Move

1. Attach a decision note to your next three significant deliverables. Four lines at the top: what you handed to AI, what you checked and how, what you changed, and the call you made that the tool could not. It answers the question 53% of these managers say has become harder to answer, and it gives the reviewer a route through your work that does not depend on guessing.

2. Take one task to full delegation this month, then document the handover. Workday's guidance points to work that is repeatable, lower risk and easy to review. Pick one task of that kind, hand it over end to end, and write down the instructions you gave, the checks you ran and the error you caught. That record is the delegation judgement the new scorecards reward, in a form you can put in front of someone.

3. Ask for the new rubric before your next review. 78% of the managers surveyed say they have changed how they evaluate. Ask yours which AI-related behaviours now count and whether that has been written down anywhere. Then confirm the answer by email. A criterion you hold in writing is one you can be measured against fairly.