McKinsey: 80% vs. 37%—Why AX Needs a Nudge

In McKinsey's 2026 global survey, 80% of respondents said AI improved their personal productivity, but only 37% said AI positively contributed to their organization's operating profit. In an October 31, 2026 column for Hankyung Business, Professor Kim Kwang-jin of the IGM World Management Institute proposed the 'AX nudge' as a device to close this gap—a small mechanism that naturally guides people toward behaviors that help performance.

Why doesn't faster work driven by AI translate into organizational results?

The column pointed to scenes where a report that once took half a day is now produced in an hour, yet review and decision-making still lag, and a completed proposal still goes through repeated checks and verification across departments. The diagnosis: individual work has sped up, but organizational work and outcomes haven't changed much.

Microsoft's '2026 Work Trend Index,' cited in the column, found that the success of AX depends less on how well employees use AI and more on how much the organization changes its work and operating systems to work alongside AI. Even in terms of correlation with AI impact, organizational factors—culture, manager support, talent management—mattered more than individual factors.

Professor Kim noted that AX pursued individually, each on their own, tends to produce three kinds of loss.

• Loss of knowledge— Without knowing a colleague's good approach, people repeat the same trial and error.
• Loss of collaboration— Loss occurs when the next person has to re-verify the work.
• Loss of capacity— Saved time isn't redirected to important work.

What three AX nudges should leaders design?

1. Make it easy to find and use colleagues' good methods

When useful prompts and know-how stay confined to an individual's chat window, others repeat the same trial and error. The column noted that simply collecting examples on a shared board isn't enough—people need to know what tasks it applies to and what to check before they can actually use it. One suggested method: use 15 minutes of the existing weekly meeting for an 'AX Experience Review.' Pick one recurring task, share what problem was solved with AI and how, and briefly summarize the conditions for use, required materials, and verification method.

What leaders should check is not the number of entries registered, but whether others actually picked up and used the method, and whether trial and error decreased. According to a September 2026 study by Boston Consulting Group (BCG) cited in the column, 'writers' introduce colleagues' adopted work methods and results at company-wide meetings and give credit where due.

2. Find where work gets stuck and reconnect the flow

Even if a proposal that used to take a day is now produced in an hour, if approval and review still take two days, the customer waits just as long. The column recommended picking one recently handled request, writing out the sequence—intake, preparation, review, decision, response—and marking the actual working time and waiting time side by side at each stage. How to reconnect the flow depends on why it got stuck.

• If information is lacking, align what needs to be conveyed.
• If approvals are delayed, define the scope of what the person in charge can decide.
• If completion criteria differ, agree on standards so the next person can start immediately.

When sales hands off a customer request, it should include 'the outcome the customer wants,' 'facts and evidence confirmed,' 'remaining items to verify,' and 'decisions needed from the next person in charge'—and the receiver should mark it as 'ready to start' or 'needs further confirmation.' Success is measured by whether the time from customer request to final response, the number of tasks sent back, and the number of long-waiting cases have decreased.

3. Reinvest time saved by AI back into performance

In BCG's '2026 AI at Work' survey, 42% of frontline employees who regularly use AI said they save the equivalent of one day per week. However, 66% said they received little to no guidance on how to use that time, and more than half said they weren't actively reinvesting it into more strategic work.

The column's suggested starting point is weekly planning. Leaders and team members should jointly decide on 'one task to reduce or eliminate' and 'one important task to advance with that time,' then put the chosen task on the schedule. Afterward, check whether customer pain points decreased, stalled proposals moved forward, or recurring mistakes were prevented. The column closed by urging teams to keep this question in mind during AX meetings:

"What better outcome did it create for colleagues and customers when someone on our team got better at something?"

Implications for AX at Korean Companies

When reviewing AX performance, it's more useful to first count how long work waits as it moves between people and departments, rather than counting how many people use AI. Even if an individual's working time decreases, if wait times at the approval and review stages stay the same, the time customers wait won't shrink.

If you run a prompt-sharing board, tracking whether other team members actually adopted and used a given method—rather than counting entries—directly targets the loss of knowledge. Deciding where to reinvest saved time shouldn't be deferred as a separate task; it's better placed within the same weekly planning process.

At this week's team meeting, pick one recently handled request and write out the working time and waiting time side by side, from intake to response. If the waiting time is longer, fix the handoff format and decision scope before choosing the next AI tool.

Source: AX Needs a Nudge, Too [IGM's Management Strategy]