Easy access to data and analytics has transformed marketing. Marketers can now use data to make decisions on a daily basis. However, when data doesn’t line up with our expectations, we can get into trouble.
Sometimes we have cognitive biases that make it difficult to evaluate data objectively. The better we understand these biases, the better we can avoid them. Here are a couple common cognitive biases to watch out for.
- The availability heuristic helps us quickly evaluate situations based on mental shortcuts. However, when looking at data, it’s important to take the long way around and not jump to conclusions.
- Confirmation bias makes us look for information that fits with our current beliefs and ignore data that challenges those beliefs. Remember to look at all your data, especially if it’s counterintuitive.
Examine Unexpected Results
Let’s say you manage a marketing program for a retailer that sells furniture as well as clothing and beauty products. Your gut tells you that customers don’t buy couches over and over again, whereas they are likely to buy new clothes more frequently.
You decide to conduct an analysis to see the average duration between purchases for different product categories. You find that the average time it takes for people to buy their next furniture item is shorter than the average time between clothing or beauty purchases.
Average Time between Purchases
- Clothing — 5 months 6 days
- Furniture — 3 months 26 days
- Kitchen — 5 months 9 days
- Beauty — 5 months 27 days
Your findings are counterintuitive. What should be your next step? How does this change your marketing strategy for this product category?
Sometimes data is there to help us realize some of our campaigns aren’t operating as expected. You may have a standardized new customer program which applies to all new acquisitions.
Believe Your Data
Let’s think about how long it takes before a customer churns from your brand — in other words, stops buying your product. If they bought a lot before, they probably will stick around longer, right?
Survival analysis is a common practice to analyze customer churn in marketing which lets you see how the probability of survivorship varies by different type of customers. When you compare different groups of people based on the number of products they bought from your brand in the last year, you would expect people who purchased more products to have higher survival probability.
What if your data shows that these groups have the same survival probability? Marketers should consider that every customer has their own purchase cycle.
Sometimes data is there to help us realize some of our campaigns aren’t operating as expected. You may have a standardized new customer program which applies to all new acquisitions. However, when you look into your data, you realize that 5% of your new customers do not receive their first communication until 10 days after they signed up to your program. What would you do after that point?
You can do the wrong thing. Assume there’s an issue with the data, ignore the results, and let your cognitive biases control your decisions. Or you can choose to investigate the program and see what might be causing the issue.
As human beings, is it possible for us to be 100% rational? Probably not, but there is always room for improvement. It is your responsibility to be aware of yourself. So what should you do to overcome your biases and be a data-driven marketer?
Always know your underlying data — most of the time it’s not data quality, but how your data is structured that will give you some tips to uncover the real insight
Don’t let your prior assumptions shape your conclusions — think and identify how those assumptions were formed in the first place
Take a step back to see the big picture — what does this piece of information show you?
Do a sanity check for your overall marketing program — identify points where you used your beliefs instead of your data
Ceren Karacasu is an Analytics Consultant at Cheetah Digital. Within the Analytics Services team, she focuses on building scalable marketing analytics tools and works on business analyses for clients using statistical & machine learning techniques. She also focuses on proving the economic impact of data-informed marketing on her clients. She has expertise in analytics, strategic data management, problem solving, and client relationship management. She holds a master's degree in Applied Behavioral Economics from Cornell University.