Let’s Move Past The Correlation vs. Causation Debate

creelawson"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Cree Lawson, founder and CEO at Arrivalist.

After 18 years in online marketing, I’m noticing that attribution has bubbled up into ongoing debates about digital marketing strategy. All too often, the debate devolves into a discussion of correlation vs. causation.

I think we often make the mistake of distilling the whole dialogue about marketing attribution into these two rhyming words, when the truth – the deeper intricacies of how media influences human behavior – deserves a much deeper discussion.

There are a number of reasons that I believe we owe it to ourselves as an industry to go beyond the simple concepts of correlation and causation. These ideas have an important place in dialogues about science or psychology, but they’re not the best fit for online marketing. Here’s why.

First, this debate is too polarized. Causation assumes that marketing directly influences human behavior, and that it’s the only influencing factor – just like when mice in an experiment receive direction through electric shocks. This gives too much credit to the marketer and not enough credit to the consumer. Aren’t there other aspects of the product or service – or even the consumer’s free will – that drive buying decisions?

On the other hand, correlation gives too little credit to the marketing process. Who would buy ads if their display had merely correlated to human behavior? Correlation – which is little more than a strong coincidence – gives all the credit to the consumer and no credit to the marketer. It’s probably slightly closer to the truth, but still misses the mark pretty widely.

The fact of the matter is that there are many ways to evaluate the impact of marketing. There’s a lot of gray area between the Orwellian notion of causation and the statistical apathy of correlation.

While it’s easy to confirm that media is the cause of certain transactions, such as music downloads or ecommerce purchases, I personally caution against accepting and propagating the notion that our marketing plans alone cause people to buy or do a given thing. It just gives too little credit to product, price point, availability, and issues beyond our control.

On the spectrum between correlation and causation is influence. Banner ads from travel companies, for instance, don’t cause people to just hop around the globe. But I’ve seen a lot of evidence that the marketing delivered by advertisers makes users more likely to visit a destination than users who are not exposed to those same campaigns, which is the real impact of a campaign.

We can’t peer into people’s brains and see what creates the desire to do something. We can only measure the external stimuli that make people more likely to do a particular thing.

Over the last few years, I’ve observed a concept I call “latent intent.” For example, there is a working theory among travel professionals that some people will never go to Mexico (perhaps someone who lives next to a rough border town) while other people will always go to Mexico (maybe someone who owns a timeshare in Cancun). The broad spectrum of people in between are the target market. After dividing this group into exposed and unexposed users, the number of people who later traveled to Mexico per 1,000 impressions of ads shows the latent intent of this in-between market.

Ultimately, using metrics like influence and latent intent gives more weight to consumer intelligence and free will. We are in a privileged position as online marketers. We can measure, illustrate, and optimize the value we create at a higher level than ever before. But setting the bar at causation and asking marketers to assume responsibility for all of the audience’s behaviors will likely produce some disappointed advertisers. Let’s stop letting the terms we use to talk about results devalue the complexity of what we’re doing, and let’s start shedding light on the dark, unexplored area between correlation and causation.

Follow Arrivalist (@Arrivalist) and AdExchanger (@adexchanger) on Twitter.

1 Comment

  1. bob sacco

    Every once in a while, I come across an article that is simple brilliant. This is one of them. Advertisers have a built-in reflex to default to the lowest-common-denominator when it comes to extracting value from their advertising buy. This notion was propagated by early digital advertising agencies that stood on a mountain and announced that the "click" was king and the key to the online universe. Then soon after, they changed their tune thus announcing it was attribution for the sale and they were only going to pay for those Users that actually converted to sale. Research studies such as the comScore/NBC study show that affluent Users don't click and that lower income Users are more likely to click. We are evolving with the new set of technology driven be the ever-evolving computer code we write. We are still in the early stages of the Internet. Aim higher.


Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>