Why AI pilots fail: data governance at Reckitt with Kay Smart

Harnessing AI's true power goes beyond plugging in algorithms. It demands a rock-solid strategic foundation. One that aligns your data, eliminates bias, and makes workforce operations efficient.

Recently, I sat down with Kay Smart, Head of Global Talent Acquisition at Reckitt. In a recent episode, Kay shared a jaw-dropping transformation. As technically fascinating as it is effective. You can watch the recording here. This part of the conversation was particularly interesting to me.

Bridging the AI divide: why governance is non-negotiable

When I asked Kay about how Reckitt approached AI for talent acquisition, something she said fascinated me. "We realized that training an AI for recruitment couldn't be separated from work movement efforts. Without a unified data foundation, our AI would perpetuate inefficiencies and biases."

That insight got me thinking. How many organizations still approach AI and skills in silos?

The answer: far too many.

If I automate chaos, I just scale the chaos.

At Reckitt, Kay and her team discovered something crucial. If AI was going to deliver meaningful, unbiased results, it needed consistency. And rigorous governance. Misaligned job requirements across their 68-country operation were skewing algorithms. This caused what we call "model drift." AI predictions lose accuracy over time.

Here's where it gets interesting. Reckitt didn't just build a one-off approach. They constructed a global governance structure. It ensures AI makes consistent, high-quality recommendations. It's a marriage of technology and human oversight. It's one reason Reckitt leads in responsible AI adoption.

Strategic insight: AI is only as powerful as the data it's trained on. If your organization isn't prioritizing data governance and uniformity, your AI isn't just underperforming. It's potentially causing harm.

Meet your people where they work

We also discussed viewing AI as a strategic partner. Not a magic fix. As Kay pointed out, Reckitt's team initially faced pushback. Recruiters felt AI just added extra work. We know recruiters move fast. They're practically Olympic sprinters when sourcing talent.

How did Reckitt solve this? By integrating AI where their people already worked. Like SuccessFactors.

This idea struck me. It wasn't about building a shiny new UI for the sake of it. It was about embedding AI in a way that streamlined existing workflows. Not complicated them. That's a huge lesson for any company implementing AI. Meet your people where they are. Respect how they already get things done.

Strategic insight: The UI is secondary to data quality. Your AI needs good, structured work and skills data. Focus on building an unbreakable data foundation first.

Savage take: managing model drift

Model drift derails AI performance faster than you can say "bad hire." At Reejig, we regularly run independent audits. This ensures our algorithms stay aligned with evolving skills data. It keeps them from causing harm. The focus stays on skills and potential. Not personal characteristics.

AI capability is compounding. Work visibility is not.

Wrapping up

What's the big takeaway? AI is a supercharger. But only if you've laid the groundwork. It won't fix bad data, misaligned job expectations, or disconnected talent strategies. But if you do it right, like Reckitt has, you reach a level of workforce agility and efficiency you never thought possible.