Why Do You Urgently Need to Escape the Data Fallacy
A new way of thinking has taken the management and marketing establishment by storm in the past decades. The idea that decisions should be based on facts – not just on manager’s opinions or experts judgment. Business intelligence and evidence-based management became not just established buzzwords but service industries. The advent of unlimited data feeds by the internet and social media started a new wave of data enthusiasm – Big Data.
How a badly voted product design can turn out to be a sales booster
Recently I got a call from a major brewery. The management wanted to launch a new label and case design and was wondering whether the new design would foster sales. So far, a standard request. What market researchers do in those cases is testing the design concepts by asking target customers: “Will you buy this case of beer?” If the new case attracts few “yes” answers and subsequent buyers, it’s a failed design. Correct?
Wrong! The result of our analysis was that while the new design scored slightly lower with respect of buying intentions, it was in fact more appealing to customers, and it has been associated with properties that drive sales. Still the new design felt somehow “unknown” – it lacked familiarity. Exactly this turned out to be a second reason for buying the product. – in this case a reason for not buying the new design.
Familiarity usually grows automatically over time. That’s why it can be expected that over time the new case will substantially increase sales, although market research didn’t show it. Should we really decide on facts?
Why is a product bought by senior customers more attractive to young people?
Recently I found myself sitting in a kickoff meeting. A meeting in which a company was trying to stop declining sales. As it turned out, a major reason for this unfortunate situation was the management’s own reliance of “facts”.
This lottery enterprise instructed me that their target customers are older than 40 years and winning the “jackpot” is their main reason to buy lottery tickets. The reason for this conclusion was the consistent observation that a majority of customers were seniors and that sales exploded every time a jackpot increased in value. Based on these insights the management adjusted their marketing material and strategy. It seemed to be a reasonable strategy that ads were mainly promoting jackpots.
We studied the true reasons for customers to buy lottery tickets and were astonished about the simplicity of an alternative explanation for the “facts”: What drives sales is a ritualized playing behavior and the experience of winning in general (not just jackpots). Both factors grow over time, which is why seniors are more common among customers. However, we also found that younger people are actually more likely to buy lottery tickets – given that they had no lottery experience so far.
By focusing on jackpots in ads, the company discouraged ritualized playing behavior, and by targeting seniors, the company missed out to win new customers with a long-term attachment to the company’s product. So, I am asking: “should we really decide on facts?”. Sure we should, but facts can be more complicated as it seems.
Why do customers who pay higher prices may have a low willingness-to-pay?
It was a rainy November day several years ago that ended with a great “Aha” moment. I was a sales and marketing director at that time and sat together with my sales reps to review the team’s monthly performance. I checked our Business Intelligence (BI) tool and asked some questions:
“Why did our sales in that product category decline?” is a typical question. The answer from one of my team was immediate. “Oh, because a particular customer’s demand has declined”. But after this answer I realized that I just had selected the wrong timeframe and sales actually increased: “Sorry my fault. So, why did our sales increase so much?”. Likewise, the answer was immediate: “I was successful in cross-selling with customer Y” my team member responded. Ok. Obviously any fact will find explanations.
After this amusing experience we sat down and used our BI tool more systematically in order to identify potential target market segments for the whole team. We found that companies producing pharmaceuticals were paying much higher than average prices. “Oh great. let’s target those companies”. Our sales reps immediately became active and contacted the promising companies. Unfortunately, with little to no success. Why?
My own methods from my days as a PhD student brought something more meaningful to light. Those pharmaceutical companies did not pay a penny more compared to average customers. They simply ordered smaller than average volumes and thus received higher priced proposals. Should we really decide on “facts”?
Many managers have had such experiences but many draw the wrong conclusion. They simply try to dig deeper into data. May the “facts” that result from such actions be misleading because they are too aggregate? The answer is “typically no”. The reason is that outcomes are simultaneously influenced by many factors. Disaggregation cannot separate their effects. It remains a dead end.
Instead, what we need to know is what causes our success measure to change. And correlation simply does not tell us anything about causation. In markets there are always dozens of factors that simultaneously influence success. In order to find out the contribution of a single potential success driver, you need to consider and model all other success factors. Looking at just two rows of data is a dead end – no matter if it is “age”, “design” or “customer industry” vs. “likelihood to buy” or vs. “accepted price”.
If you find this post useful, we appreciate if you “like” it and share it.