You probably have considered a host of ways to change your business or draw new customers, but are unsure whether they will work.
Why not take a "test and learn" approach, trying out some of those notions in experimental form, like a research scientist checking the results against a control group, to determine if it's worth implementing the change?
In Harvard Business Review, Eric Anderson, a professor of marketing at Northwestern University, and Duncan Simester, a professor of management at MIT's Sloan School of Management, offer the following tips, with a reminder that the goal is not to conduct perfect experiments but to make better decisions:
Focus on individuals. Think short term
The most accurate experiments involve actions affecting individual customers, rather than segments or geographies, and observing their responses. You want to measure their purchasing behaviour with the change, not their perceptions, and determine whether higher profits will occur.
Focus on settings in which customers respond immediately - U.S. Bancorp, for example, looked at experiments to improve customer acquisition in its wealth management business rather than lifetime customer value.
Keep it simple
Look for experiments you can pull off with existing resources and employees. Don't start with ideas that require extensive retraining of staff or manipulation of store layouts. Keep it small and simple.
Establish 'proof of concept'
In academic experiments, researchers change one variable at a time so they know what caused the outcome. In a business setting, you first have to establish proof of concept, so change as many variables in whatever combination you believe is most likely to get the desired result. The academics note that when a chain of grocery stores wanted to test the best way to shift demand from national brands to its private-label brands, it increased the prices of the national brands and decreased the private-label prices. When that showed shifting demand was possible, the company then drilled down to look at individual price changes.
Slice the data
While the goal is simplicity, the reality is that when customers are randomly assigned to treatment and control groups you may have multiple experiments to decipher.
For example, if the sample includes men and women, you will want to check if there is a difference in impact by gender. Most actions affect some customers more than others, so look within your control and treatment groups for sub-groups to study, such as age and historical purchase patterns.
Think beyond your box
Don't run experiments that only incrementally change your policies - tinkering with prices to resellers, for example, when you might want to rethink your sales approaches.
This is a time for out-of-the-box thinking, the professors note. They point to Britain's Tesco PLC, which reportedly discovered it was profitable to send coupons for organic food to customers who bought wild birdseed.
Measure what matters
Check the results carefully to understand the full context in which any changes in purchasing behaviour occurred.
Look for natural experiments
Your business may naturally churn out experiments that you can take advantage of. One example is an online apparel retailer that, when it opened its first store in a particular state, found it now had to pay state retail taxes there for online and catalogue sales, offering an experiment into the impact of sales tax on online and catalogue demand.