Gurleen Saraon decodes the ‘corelation vs causation’ debate in programmatic

The complicated world of marketing analytics revolves mainly around three pillars- understanding the consumer, resonating with the consumer and proving the outcome.

At exchange4media’s second edition of Real Time Summit, Gurleen Saraon, director- digital media, India and Middle East, Publicis Epsilon, decoded the debate between correlation and causation for digital-first brands

“Correlation and causality are fairly misunderstood terms as far as measuring in programmatic is concerned,” she said.

Saraon then presented a slide that showed the divorce rates in the state of Maine and the per capita consumption of margarine on the same graph. Both of them seemed to have a directly proportional relationship. Both of them decreased and increased together.

Saraon explained, “So would this mean that increased higher consumption of margarine or butter leads to higher divorce rates? No. There exists a positive, coefficient of correlation definitely but there is no causal relationship between the two. This is exactly the problem we face in marketing analytics.”

When it comes to programmatic, there is attribution where the sale drives the media, which simply translates into trying to touch as much of the audience as possible closer to a sale. On the other hand, the media driving the sale means optimising the incrementality away from the 100 percent conversions that are going to happen anyway.

The Publicis executive further deep dived to explain the two types of attribution, which are last-touch and multi-touch, via a soccer field graphic. In the last-touch, a brand focuses on getting the last click, but in multi-touch, they focus more on all the touchpoints of a consumer journey. The former helps a marketer understand how much value and ROI is brought in by every touchpoint. 

“If you see soccer, the last goal is hit by their strongest player like Ronaldo. That person gets 100 percent credit and recognition for the win. But this becomes a bit of a problem when we look at various omni channels and that is where multi-touch helps,” she added.

Top consumer brands measure marketing by multi-touch attribution, a marketing mix modelling, and conversion lift studies.

Towards the end, Saraon quoted the famous statistician George Box who said, “All models are wrong but some are more useful”.

“This basically means depending upon where a business is, certain models work more efficiently in those scenarios, but it really matters what goal the business is taking on. If the business is aligned to report click-throughs or easy to navigate touchboards, then that's what it's useful to go with,” she concluded.