The future of marketing is built on open data infrastructure

I was eight years old when I launched my first company making custom scrunchies on my grandmother's sewing machine. I didn't have a business plan, but I had something better: I paid attention. I noticed which colours sold out first. I tracked which friends came back for more. I adjusted. That scrappy, data-driven instinct is still how I think about marketing today, just with bigger stakes and better tools.

Here's what I've learned after years in this industry: most marketing teams are flying blind, and they know it. They're making gut calls dressed up as strategy, stitching together reports from a dozen disconnected systems and calling the result "insights." It isn't. It's noise.

Getting data right isn't optional anymore

Nearly three-quarters of enterprises manage more than 500 data sources. And yet most marketing teams can't tell you in real time how a campaign is performing across channels, why a specific cohort converted, or which touchpoints actually moved the needle. That's not down to a lack of creativity – that's a data infrastructure problem.

The fix is centralising and automating your data so that clean, reliable information flows across every system in real time, without manual intervention. When that's in place, you stop reporting on what happened last quarter and start making decisions about what to do tomorrow. 

This is what I mean when I talk about marketing with confidence. Not confidence in vibes or experience, but confidence backed by evidence.

AI only works if your data foundation does

There's a lot of breathless talk about AI transforming marketing. It will – but only for teams who've done the unglamorous work first. Fivetran's own research found that 42% of enterprises have seen AI projects delayed, underperform or fail outright because of data readiness issues. That number should stop every CMO in their tracks.

AI doesn't create insights from chaos. It amplifies whatever signal you give it. Give it fragmented, stale, siloed data, and you get fragmented, stale, unreliable outputs. Give it a clean, unified data foundation, and suddenly you can anticipate customer needs, personalise at scale, and adapt in the moment instead of in hindsight. The AI opportunity in marketing is real. But it's downstream of the data work, not a shortcut around it.

The future is open by design

I believe strongly that the future of AI and analytics has to run on open data infrastructure: vendor-neutral, interoperable systems where organisations actually own and control their data. The alternative – closed, proprietary ecosystems where your data is held hostage to a single vendor's roadmap – is bad for innovation, bad for competition, and ultimately bad for customers. 

The best outcomes happen when data can move freely, teams can collaborate across platforms, and no single vendor controls what you can build or how fast you can move. This is both a technical and a strategic opinion. Companies that lock themselves into closed systems today are making a bet that one vendor will always have the best answer, and that bet rarely ages well.

This requires people, not just platforms

Even the best data infrastructure can fall short if the teams behind it all look and think the same. AI systems reflect the data and assumptions that shape them. When the people building data pipelines, models and architectures don't bring enough perspective to the table, blind spots show up quickly. And in marketing, those blind spots tend to surface where it hurts most: targeting, personalisation and customer experience.

That's part of why representation in technical roles matters more than many organisations realise. The women working across data engineering, AI infrastructure and data architecture today are helping shape how these systems behave at scale, which is critical because AI doesn't stay static. Reinforcement learning systems train on their own outputs. A model that starts with narrow assumptions doesn't just reflect those assumptions; it doubles down on them. Homogenous teams are less likely to catch that drift early, when it's still fixable.

On International Women's Day, it's worth saying plainly: building diverse technical teams isn't just the right thing to do. It's how you build more accurate, more resilient, and more trustworthy AI. Companies that treat this as a side conversation are missing what's actually at stake.

What's actually at stake

The scrunchie business taught me that success comes from understanding your audience and using that understanding to act. The tools are different now, but the principle isn't. Marketers who invest in real data infrastructure – centralised, automated, open – will be the ones who actually know their customers, move faster than their competitors, and earn the right to have a point of view in the market, while everyone else stays guessing. The data is there – the question is whether you'll build the foundation to use it.