Breaking the analytics addiction: My confession

My first gig in digital marketing was as an "Interactive Campaign Analyst." Sounds fancy, right? Basically, it was what most places now just call their Analytics department. My job was to figure out how much bang our clients were getting for their buck from our campaigns and report back to them.

I absolutely loved the job. Imagine me, a cheeky 21-year-old, schooling seasoned media planners about why their beloved campaigns weren’t hitting the mark. I won’t lie—I got a real kick out of that… Of course, I wasn’t exactly the most popular kid on the block with them. Back then, I thought I had all the answers just because I could slap a conversion number on everything with my fancy digital analytics platform. Little did I know, those numbers were way off the mark.

It turns out, I wasn’t the only one getting it wrong. Businesses everywhere were jumping on the digital analytics bandwagon, convinced they could pinpoint down to the last penny how much revenue their campaigns were pulling in. You’d hear teams confidently declare, “Our search campaign raked in exactly two million, three hundred seventy-two thousand, four hundred twelve pounds and 10 pence last month!” The big bosses ate it up, and before you knew it, budgets were being tossed around like confetti in all the wrong places.

As time went on, we all started to see the cracks. There was this huge gap between what the analytics platforms reported and the actual value created for businesses. For years, everyone was obsessed with the "last click" model—it was gospel. If that same logic applied in the “real” world, the sign hanging over a shop’s door would be crowned the king of marketing!

Eventually, people began to question the whole "last click wins" idea. Was the last click really the MVP? Maybe the first click mattered more, or perhaps we needed a fair way to spread the value across multiple clicks? Exploring these questions started to shake up our old ways of thinking, but let’s be real, those new models only made a tiny dent in fixing our biases.

In the midst of this measurement makeover, we lost sight of some marketing fundamentals. Good marketing isn’t about what your analytics platform spits out. It’s about really knowing your customers and crafting strategies that not only reach them where they are but also resonate on a personal level. Byron Sharp and Les Binet talk about this much more eloquently in their books, “How Brands Grow” and “The Long and Short of It.”

Lately, there’s been a bit of a comeback for good old-fashioned media planning. More companies are leaning into solid testing and research to really understand what makes their marketing tick. But, let’s face it, many are still caught between wanting the new shiny strategies and clinging to the old-school need-to-know-every-little-detail approach.

Data-driven attribution (DDA) has looked like a shiny new toy to solve these issues—it promises a sophisticated look at performance with all the detailed numbers you could want. While it sounds great on paper, I’ve found it doesn’t quite deliver. It’s got the same old problems—overvaluing what we can track and undervaluing what we can’t. And it doesn’t help us see the big picture.

This isn’t just a critique of DDA, though. It’s more about encouraging a rethink on how we use analytics. These tools are great for understanding how people interact with your site and where they come from, but they're not so hot at telling you the real impact of your marketing investment.

What’s the answer then? Well, first, get a grip on what role each marketing effort plays in the customer journey and measure it accordingly. When it does come to linking activities to sales, go for tests and models that give you the real scoop, not just the easy numbers.

And hey, maybe we need to chill a bit with how much detail of performance we actually need to give the businesses we work for. Does the board really need to know how much your Google Shopping campaign drove each month, or are they actually more interested in the overall contribution of your marketing strategy to business performance?

Looking ahead, I’m hopeful that machine learning will soon help us make smarter predictions using data from robust methods like media mix models and incrementality tests. We’re not quite there yet, but we’re getting closer.

It’s great to see businesses digging deeper into understanding marketing’s value beyond just the numbers. And to any old colleagues reading this—sorry for being that overconfident kid back in the day. Turns out a bit of experience really does go a long way!