I’ve sat through at least a dozen “data culture transformation” kickoff meetings. They all start the same way. A senior sponsor gives a speech about the importance of being data-driven. Someone presents a roadmap with phases and workstreams. There’s usually a new tool involved. Everyone leaves feeling energised.
Twelve months later, nothing has changed. The dashboards nobody asked for sit unused. The data literacy training had 30% attendance. Executives still make decisions the same way they always did, and the data team is quietly updating their resumes.
The problem is right there in the framing. You can’t “transform” culture with a project plan. Culture is what people do when nobody’s watching. It’s the accumulated habits of how decisions get made across thousands of small moments every day. You can’t install it like software.
Why Data Transformation Programs Fail
Most data culture programs fail because they focus on tools and training instead of decisions.
They’ll spend millions on a new BI platform. They’ll mandate data literacy workshops. They’ll hire a Chief Data Officer and give them a charter to “embed data into the DNA of the organisation.” All of these things can be useful. None of them create culture on their own.
Here’s what I’ve observed over 15 years: culture changes when the consequences of decisions change. If an executive can walk into a meeting, make a gut-feel decision, and nobody questions it, that’s the culture. No amount of training changes that.
At Westpac, I watched the data culture shift happen not because of any program we ran, but because a new leader started asking one question in every meeting: “What does the data say?” Simple. Relentless. Uncomfortable. Over about six months, people started showing up with data because they knew they’d be asked. That one behaviour change did more than any transformation program I’ve seen.
The Three Things That Actually Move the Needle
After trying both the big programmatic approach and the small behavioural approach, I’m convinced that culture shifts come from three things.
Leaders who use data visibly. Not leaders who talk about data. Leaders who actually pull up a dashboard in a meeting and change their mind based on what they see. When a VP says “I was going to go with option A, but the retention data is telling me option B is smarter,” that moment teaches the entire room more than any training session.
At Cochlear, I’ve been working with our leadership team to make data part of how we run operating reviews. Not as a separate agenda item, but woven into every discussion. When someone proposes a new initiative, the first question is: “What’s the baseline, and how will we know if this worked?” It’s becoming habit.
Decisions that are tied to metrics. Not vanity metrics. Not dashboards with 47 KPIs where everything is green. I mean a small number of metrics that people actually care about and are accountable for.
The worst thing you can do is build a dashboard that tracks everything. That’s not data-driven decision making. That’s data-drowning. I’ve seen teams with 200 metrics where nobody can tell you which five actually matter. When everything’s measured, nothing is.
We aim for three to five key metrics per team. That’s it. If you can’t fit your metrics on a single screen, you have too many. And each metric needs a clear owner, a target, and a visible plan for what changes if the metric moves in the wrong direction.
Making data accessible without making it a burden. This is the part most organisations get wrong. They invest in data platforms, set up self-service analytics, and then wonder why adoption is 12%.
Accessibility isn’t just about having the tools. It’s about whether someone can get an answer to their question in five minutes without filing a ticket, joining a Slack channel, learning a query language, or waiting for a data analyst to become available. Every friction point is a reason for someone to just go with their gut instead.
At The Smith Family, where resources were tight, we solved this by putting pre-built analyses into the tools people were already using. Instead of asking people to come to the data, we brought the data to them. Adoption went from negligible to meaningful in weeks, not because the data was better, but because the friction was gone.
The Data Team’s Role (It’s Not What You Think)
If you run a data team and you think your job is to build dashboards and models, you’re thinking about it wrong. Your job is to make better decisions more likely across the organisation.
Sometimes that means building a dashboard. Sometimes it means sitting in a meeting and asking “have we looked at the data on this?” Sometimes it means saying no to a dashboard request because what the stakeholder actually needs is a one-time analysis, not a standing report.
The best data leaders I know spend more time in business meetings than in front of their laptops. They understand the decisions being made, where data could improve those decisions, and where the gaps are. They’re not order-takers. They’re decision partners.
This requires a different skill set from what most data professionals are trained for. Technical skills are necessary but nowhere near sufficient. The ability to understand business context, ask good questions, and influence without authority matters far more at the leadership level.
Stop Calling It Transformation
I’ve become allergic to the word “transformation” in the context of data culture. It implies a beginning and an end, like there’s a finish line. There isn’t.
Building a data culture is ongoing work. It’s a thousand small decisions about how meetings are run, how proposals are evaluated, how success is measured, and how failures are analysed. It’s messy and slow and nonlinear.
But it compounds. Each decision made with data makes the next one slightly more likely. Each time a leader changes course because the numbers told a different story, it normalises the behaviour for everyone watching.
You don’t install a data culture. You grow one, one decision at a time. And it starts with the people at the top deciding that “I think” is no longer good enough when “the data shows” is an option.
For a deeper look at why data strategies fail structurally — beyond culture — see my research on The Strategy-Execution Gap.
