Correlation vs. Causation: A Serious Upgrade in Marketing Insights

Share This Post

You could say that marketing campaigns are only as good as the data we collect. These insights are powerful and it helps us understand how to spend our dollars to put in our ideal lead. 

But what if those insights aren’t giving us the full story, but we’re acting on them anyway? Then all of a sudden we’ve spent a lot of budget on a marketing campaign that’s optimized on half-truths. 

In our premiere episode of Revenue Science, I sat down with two of the foremost experts of causality, Dr. Vishal Sikka and Jake Klien to discuss smart marketing and how we can use scientific techniques to transform the way we, as marketers, look at data. Check out the episode here:

Most marketing data is basic

The episode was created to give you a solid overview of causality in just about five minutes. But if you’re ready to dive into the nitty gritty, then keep reading the off screen conversation between Klien and Dr. Sikka…

The amount of data at our fingertips can be overwhelming – especially when working across multiple platforms, each with its own set of dashboards. Weirdly–even with all of these numbers–the insights from this data is quite often elementary at best. Here’s how it plays out: 

Boss: We need more leads.

You: Great, I’ll run ads for a month to see if that brings in more conversions.

30 days later…

Boss: How was your ad experiment, do we have more leads?

You: There was an uptick of ad clicks and conversions within the month, so I would say it was a successful campaign. We can get even more leads if we put even more money behind it.

Boss: DONE!

You run the campaign a second time but this time around it’s a total flop. Now, you’ve got a loss of time and money on your hands…and a lot of explaining to do. 

We’ve all been there. We see datasets trending in a specific way, and it’s easy to make a correlation between the two and assume either success or failure of a campaign. There’s a reason why this happens: data insights based on correlation is simply too basic of a method to understand why people are interacting with your marketing efforts. 

“Correlation is easy to find when we have data.” Dr. Sikka explained. “Modern AI techniques make it easy to find correlations, but causality is the driving force of the universe.” 

We’re getting close to mind reading…

Correlated data uses the past to help guide the future and unfortunately that doesn’t tell us the current mindset of a customer when they are converting. Until we can actually read minds, we have to rely on the next best thing: cause and effect. 

Going back to our scenario on ads, maybe your initial campaign was launched in the summer and your second version was launched in winter. You might consider a few questions. What does seasonality mean to your customer when they’re ready to buy? Or maybe your ads were only showing up in the morning and not evening. How does timing impact your user’s journey?

These types of insights are more meaningful because it helps us understand the root cause of buying and the effects our campaigns have on customers in the moment. Once we have this, we start to get a better picture of the buying behavior… and of course, that’s when we start to see a real impact on revenue. 

In this episode, Dr. Sikka shared the importance of cause and effect in marketing, “When we look at the behavior of our customers, or how marketers look at how customers react to things, the force underneath that is cause and effect,” he said. “Now modern technology, recent AI advances make it possible for us to really understand those causal links between things, not just spurious correlations, things that happen to happen at the same time, but what really causes certain behaviors, what really causes certain outcomes.”

Sounds great right? So, how do we get these insights? 

Later correlation, hello causation!

To do this, we need to reevaluate how we look at our data. “To improve outcomes going forward, what you want to do is basically take a different lens on your observational data and you want to debias it, randomize it, so you can actually narrow in on and focus on the actual effects of how your actions as a marketer affect customer behavior,” said Jake Klien. 

Now, one last time – go back to our ad scenario. You just delivered that second report with the sad numbers. But this time around, instead of using correlated data to provide feedback – you use causation to analyze the data and provide actionable insights. In your report you provide next steps like: 

  • Stop paid ads to see how organic search performs for a month
  • Test new channels with same messaging to see if other channels are more effective
  • Retargert churned customers from ads at different times of the day with new messaging

Actions like these turn sad numbers into happy ones –  they’ll only provide more insight on future campaigns and bring you closer to the full story. 

Automating cause and effect 

At the end of the day, looking at data and seeing how it correlates with each other is an easy task for marketers, but I think we can all agree that it’s too basic of a solution for your complex customer. 

Causality equals better marketing outcomes. And while it seems like a big task, it’s made easy with the right artificial intelligence. Revenue Science is a breakthrough discovery in AI that ushered in a new era of data analytics and in return more marketing ROI opportunities. 

I hope you enjoyed this first episode of Revenue Science. Make sure to tune in for next week’s episode! Until then, if you’re ready to get the causal insights for your own prospects, get a free trial of Dealtale by clicking here.

More to Explore

The Revenue Science Blog

Learn about the latest breakthroughs in
Revenue Science, geek out on quantum theories, and mastermind your next big marketing moves!