This article was contributed by Leon Young, Principal Consultant - Data & AI at Allura Partners

Few people arrive in a career in Data & AI via a PhD in physics and publications in some of the world's most prestigious scientific journals.

But for Mick Pullen, Head of Product – Data, Analytics & AI at Endeavour Group, that grounding in translating complex, specialised work for a broader audience turned out to be the perfect foundation for a career at the frontier of artificial intelligence, one that has since taken him across banking, software development and FMCG.

I recently sat down with Mick to explore why so many organisations struggle to generate commercial returns from their AI investment and what it actually takes to change that.

Q. Despite enormous investment in data and AI, many organisations struggle to see measurable outcomes. Why does that gap exist?

For me, there are a couple of reasons. The first is that there’s a need for data leaders to understand the business strategy – and to do this, they need a connection with the executive team and the board.

Without that connection – without that understanding – you can’t enable the business strategy with data and AI. You find yourself just doing cool things, building cool things that are interesting, but maybe not driving value, not driving the business in the direction it needs to go.

So, you might get an AI tool in and build something you think is going to help the business, then find there’s no engagement from the stakeholders. It fails – there’s no return on investment.

The result of experiences like this is a degree of caution – I’m now seeing execs want to be more confident that the investment they’re making is going to return value before they commit.

Q. How do you translate that principle into practice – especially when you’re dealing with stakeholders who are sceptical or resistant?

From my perspective, the questions I always ask are: what’s the business strategy? And how do I enable that with data and AI? That’s key.

But it depends on the stakeholder. The great execs to work with are the ones who are really inquisitive and open – who say, “I’m not an expert here, please help me understand”. You can take that relationship from start to advocacy pretty quickly.

For the stakeholders who are harder to convince, or who have a bias against this type of thing, you have to be much more deliberate. It’s about deeply understanding that stakeholder’s business priorities – what they’re looking to achieve – and positioning the work from that perspective.

You’ll often find resistance from people who fear that the introduction of an AI tool will take them out of a job. Aggressively introducing a new process doesn’t work in this environment. It’s about education and it can take multiple months to help them understand that what you’re introducing will help them become more effective – and more highly regarded within the organisation.

Q. How do you move AI investment from proof of concept into genuine business value?

In a product development environment, you don’t have to go for the full-blown solution straightaway. It can be an iterative process where you get a quick win, present it, get feedback, and build from there, curatively developing a solution to the problem.

That’s quite different from the traditional transformation-style program, with key milestones defined upfront. In product development, you encounter things you didn’t anticipate, you get new information, and you have to adjust. Checking in regularly, showing progress and showing how you’re solving problems over time is really valuable.

It also allows you to course-correct. If you get 80% of the way there and an exec says, “We’re happy with where we’re at, maybe we should pivot to something else,” you can do that because you’re iterating. It’s much harder to course-correct in a six- or twelve-month program of work where you don’t check in until later.

Going back to the question of why we’re not seeing returns on AI investment – a big part of it is companies going out and getting the biggest model from one of the major providers and applying it to a problem without doing the groundwork. A more iterative approach allows you to solve problems over time. Once the board or exec have endorsed a program, you can start playing with bigger models and getting real, tangible outcomes.

Q. What about when leadership changes and the alignment you’ve built is suddenly at risk?

It is a journey. You have to have conversations, understand what they care about, and over time help them understand the value you can provide. Changing just the framing of the way you speak about things can play a very important role, because if you’ve got five minutes with a new exec and you’re talking about the wrong things, that’s enough for them to disengage.

You need to use your network as well. A new executive is coming into a very tough situation – getting up to speed with an entire organisation in a small amount of time. Working through your network to understand their positioning; what they’re hearing and what’s important, helps you collaborate more efficiently to support them.

Q. How should organisations think about building data and AI teams?

If you’re building a data and AI team, you might be tempted to go after the most technical person you can find – the best credentials; a PhD in AI. Nine times out of ten, you’re probably making the wrong call because the value proposition of an analyst or data scientist is now squarely in their ability to understand and connect to the business. They need to be able to communicate, build relationships, and understand non-technical things – how does this business run? What are the revenue streams?

I’ve often said: take someone who’s got 70% of the very best technical skills, but who you can speak with, who you can coach, who you can put in front of stakeholders – take them over the most technical person in the world. It’s easier to teach the technical components than it is the interpersonal and commercial ones. Soft skills and business understanding are even important now, because AI can increasingly handle the technical execution.

Q. Finally, how do you stay on top of the pace of change, and how do you build that into your team?

You have to personally, constantly invest in this space because technology is developing at such a rapid pace – and that can be hard.

A key part of how I stay current is my network. I have connections in the industry, we meet up every few weeks, and I attend regular roundtables. I trust people who are closer to the tooling to tell me what the most important advancements are. LinkedIn alone will overwhelm you – you won’t know how to absorb all of it.

The other thing I do is play around with AI tools at home, so I understand what’s possible and can translate that back to my team. The world of product management that existed five years ago is dead. A product manager today should be on the tools, prototyping, getting iterations out quickly, getting feedback fast. Setting that expectation with your team – that they should be thinking that way – has been critical. It’s not easy. Even doing all of that, I still see posts and think, “I didn’t even know about that”. It’s a rapidly changing world, and you have to give yourself permission not to know everything.