Blog

Experiment on customers to get more from your self-service pilot

March 29, 2011

In our house, spring means two things: Chicago Cubs baseball and the school Science Fair. As I was helping my son sort out whether or not a cheese sandwich molds faster than a banana one, I started thinking about how to use the scientific method to get more out of self-service pilots.

When most companies do a pilot they don't experiment with variables that may improve results. Instead, they just make broad assumptions and then deploy those. They make minor tweaks before they do the full roll out. However, they'll never know if they could have gotten better results if they had done something differently.

With self-service or kiosk deployment, there are several variables that can affect success. Some of the things that we've found to have an impact on the customer and usage are:

• Color of the device.
• Design of the attract screens.
• Graphics and image placement.
• Form factor.
• Lighting.
• Store placement.
• The voice or screen font.
• Signage.
• Employee training and communication prior to deployment.
• Customer promotions and training.
• Initial rewards for using the device.
• Sounds used for user feedback.
• And different component parts.

There was an interesting article in the March issue of the Harvard Business Review about experimenting on customers. In A Step-by-Step Guide to Smart Business Experiments, Eric Anderson and Duncan Simester outlined why experimenting on your customers can help improve success. Two of the critical success factors in ensuring that the experiments were successful were maintaining a control group and providing a clear feedback mechanism.

So how might a customer experiment in self–service look? Consider the following example:

The experiment
A larger retailer wants to improve participation in their loyalty program and the customer data they get in the store. Moreover, they want to gauge how effective their coupon programs are. Therefore, they've decided to deploy a self-service device that allows customers to swipe their loyalty cards when they enter the store to see which coupons or promotions apply to them.

In this experiment we are testing two variables (1) in-store placement of the kiosk and (2) use of coupons offered. The pilot might look like this:

Store A
The kiosk is situated right at the entrance and the coupons that are in the weekly circular are offered through the device.

Store B
The kiosk is located on the end of an aisle and special coupons are offered to loyalty customers

Store C
This store would act as the control group without the self-service solution. Loyalty program members would be surveyed with the same set of questions as in stores B and C. In the experiment you'd want to measure whether or not they took advantage of any coupons from the weekly circular.

The results
After a set amount of time, the retailer would compare the results from all three stores. Let's assume that our customer had planned to have the customer swipe his card when entering in his implementation plans.

Our experiment could have shown, for example, that there was more customer coupon usage when the device was located at the end of aisle, which could have led us to conclude that dispensing coupons while the customer was shopping rather than getting his basket etc., there was more utilization of the coupons. Our experiment allowed us to deploy with the confidence that we were offering the best solution to meet our business objective.

As you deploy a self-service solution, it is important to consider the variables that can impact your customer or your success. Also, it is imperative that you slice and dice the feedback to understand why one is preferred over the other.

Finally, don't be afraid to experiment and test your hypothesis in the pilot phase. It is far easier to make tweaks and changes in this phase.

Oh, and if you are wondering, banana molds much more quickly than cheese.