2016-02-25 11:03:41
Michael LucaOliver Hauser
February 04, 2016
Consider a delinquent taxpayer who receives one of the following two letters in the mail:
Letter 1: We are writing to inform you that we have still not received your tax payment of $5,000. It is imperative that you contact us.
Letter 2: We are writing to inform you that we have still not received your tax payment of $5,000. By now, 9 out of 10 people in your town have paid their taxes. It is imperative that you contact us.
The UK tax authority, HMRC (or, properly, Her Majesty s Revenue and Customs), had been sending out a letter that looked a lot like Letter 1 for many years. They didn t think that it was necessarily the most effective letter. In fact, they hadn t thought much about the letter at all. After all, it was just a letter, an administrative necessity.
But in 2010, a newly formed team of behavioral scientists inside the UK government, known as The Behavioural Insights Team (BIT), set out to test its effectiveness. Where HMRC saw administrative hassle, BIT saw opportunity, the opportunity to collect overdue taxes by incorporating insights from behavioral economics and social psychology into the tax letter.
BIT ran an experiment in which they sent out two letters something like Letter 1 to a randomly selected group of delinquent taxpayers, and something like Letter 2 (which incorporates a concept that psychologists refer to as social norms) to another group. As it turns out, Letter 1 was costing the UK tens of millions of pounds per year because it was simply less effective than Letter 2. In response to Letter 2, delinquent taxpayers stepped up their game and started paying their dues.
We take two lessons from this tax letter. First, we all have intuition about what s most effective which letter is best, and how much of a difference it makes. But the best managers have humility about their intuition, which is sometimes off. Because of this, experimental methods are an important part of the managerial process and can tell us what works and what doesn t.
Second, situations like the UK tax letter where we just do the same thing over and over again without asking if it s really effective are very common. And managers need to recognize these situations for what they are: unnecessary risks. Why speculate when you can run an experiment to see what works? Just look around your organization, and you will find opportunities all over the place.
To see communication experiments in action, consider the following:
One of us (Luca, with coauthors Duncan Gilchrist and Deepak Malhotra) wrote a paper showing that the impact of higher wages depends on the reference point of the employee and reference points are often set through communication. More generally, compensation letters are a great but often wasted opportunity to engage with employees. For example, should you say, This raise is our token of appreciation for the hard work that you do for this company or This raise is our token of appreciation for the hard work that you do for your colleagues ? Which of these is more effective at encouraging future motivation depends on the motivations of an employee at a specific job, and can easily be tested through a randomized controlled trial.
Companies routinely use text messages to reach customers, but not many take advantage of an experiment to find out what works. Consider an example drawn from government elections: in 2011, Neil Malhotra found (with coauthors Melissa Michelson, Todd Rogers, and Ali Adam Valenzuela) that even a well-chosen cold-text that is, a message sent to people who had not requested to be texted increased voter turnout in the California elections. In this situation, one experiment may not be enough. While the text messages saw large effects, one might expect the effect to be smaller now that texts are such a common tool.
Experiments can also tell when your intuition would have gotten it wrong. We ve worked with organizations that were confident their advertising messages were effective, only to learn through a trial that they had no effect at all.
Every year, there are new ways to communicate with customers and employees letters, emails, texts, instant messages, chat apps. All of these are opportunities for organizations to communicate effectively in order to achieve their corporate objectives. The experimental method enables you to systematically improve your communication with your customers and to roll out changes that have proven successful. When done correctly, experiments can drive success and create a culture of learning and innovation.
Michael Luca is an assistant professor of business administration at Harvard Business School.
Oliver Hauser is a doctoral student at Harvard University.