A new survey is making headlines across the business press. Only 44% of employees globally feel they are thriving at work. In Japan, the number is even lower — just 31%. And only 8% of employees in Japan report being engaged at work. Eight percent. The global average is 20%, which is already nothing to celebrate.

The reason being cited? AI anxiety. Workers are losing confidence. They fear their roles are about to be replaced by machines.

Now, I want you to predict what happens next. What will HR departments around the world do with this data?

They will launch wellbeing initiatives. They will commission more surveys. They will hold town halls about psychological safety and resilience. They will do everything except the one thing that would actually make a difference.

Here is the fundamental flaw in engagement surveys — and it applies equally to this thriving statistic. They treat every employee as equal.

The excellent and the mediocre get the same weight. The star performer who is quietly updating her resume and the perennial underperformer who has been doing the bare minimum for six years — they each contribute one data point. And so what you get is a number that tells you about the average. And the average is not what builds your business.

So what does it mean if 56% of your workforce isn't thriving?

The honest answer is: you don't know. And that is precisely the problem.

That 56% could be your mediocre performers, grinding along as they always have. But it could just as easily include your best people — frustrated by the tolerance for mediocrity around them, bored by a lack of challenge, held back by managers who mistake compliance for performance.

Those are two completely different problems. They require completely different responses. And the aggregate number tells you nothing about which one you have.

In fact, the more troubling scenario is not a workforce of disengaged mediocre performers. It's a workforce where your excellent people are the most disengaged — because they see clearly what the mediocre cannot, that the organisation is not serious about excellence. Those people will not wait for your next engagement initiative. They will leave, quietly, without warning. And your survey scores will barely move — because the mediocre who remain will fill in their questionnaires just as they always have.

Now let me talk about what AI anxiety actually reveals — and what the right response looks like.

Your best people are not afraid of AI. The excellent are adapting to it, experimenting with it, figuring out how to use it to outrun their peers. AI is a tool, and excellent people pick up tools.

It is the mediocre who are most threatened. Those whose value to the organisation was always marginal, and who now face technology that can do marginal work at scale. That is a real fear — and it is not entirely unfounded. But it is not your most urgent problem.

If your engagement scores are dropping because the mediocre are anxious about AI, the instinctive response is a wellbeing program. Resist it.

The most successful leaders I know invest their time, their energy, and their resources in their best people. They play favourites with the excellent — without apology. HR departments encourage the opposite. Pour effort into the underperformers. Run remediation programs. Set improvement plans. And in doing so, they systematically neglect the top performers, who need no remediation but who nonetheless need your attention, your stretch assignments, and your advocacy.

The math is straightforward. Invest in your best people, and you get a multifold return. Invest the same effort in your mediocre performers, and you move the needle a percentage point or two — and that is being optimistic.

Worse — and this is the part that leaders too often miss — tolerance of mediocrity is never invisible to the excellent. They see it clearly. They resent it. And eventually they leave for an organisation that shares their standards.

Fire the mediocre, on the other hand, and you send an unambiguous signal about what you stand for. The excellent notice that too. And they stay.

So when AI anxiety shows up in your engagement scores, read it as a signal, not a crisis. Ask who, specifically, is anxious — and why. The answer to that question tells you more than any arbitrary response statistic on an employee engagement survey. It tells you whether you have a performance problem worth solving — or a performance problem worth accelerating.

So ignore the 44% headline. Ask one question instead: are your best people thriving?

Walk your floor. Go out among your people without an escort. Watch. Listen. Who is energised? Who is going through the motions?

Talk to your top performers directly — not through their managers, directly. Ask them what is getting in their way. Ask them whether they feel they are growing.

And ask the managers who matter whether they are losing their stars. Not their headcount. Their stars.

If the answer is that your best people are thriving, growing, and staying — you have nothing to worry about. The 44% can stay where it is.

If the answer is no — that your excellent people are restless, frustrated, or already gone — no amount of wellbeing initiatives will fix it.

Only standards will.