The Case for Data-Driven Test and Learn Processes

by Jonathan Hornby

Marketers have used test-and-learn loops for decades. But as the recent recession has put marketers’ budgets under scrutiny, some wonder whether these exercises are cost-effective efforts that can drive business growth or sinkholes for time, money and resources.

The tactical – or perhaps safe – response is to avoid uncertainty, double down on known results and focus on efficiency. That’s fine as a defense. A marketer might even see some incremental growth. But if competitors aren’t satisfied with incremental growth – if they are experimenting – there will be problems.

The strategic approach offers far greater opportunity: Learn a little to profit a lot.

In “Beating the Odds When You Launch A New Venture,” Harvard Business Review reports that the most successful entrepreneurs have redirected their strategies at least five times before hitting a solid growth trajectory. All devised low-cost experiments to test concepts before committing to (potentially damaging) courses of action.

Think about website design. What appeals to one person could distract or confuse another. might perform hundreds of small experiments in a single day. It changes wording, layout, images and order of messages to evaluate which combinations generate the highest interest. Amazon’s traffic volume lets it see impact quickly, and technology lets it flip the switch instantly.

The overarching principle is, if something works, extend it. If it flops, flip the switch back and take heart in having learned in time to avoid setbacks.

Chris Trimble, co-author of the Other Side of Innovation, offers this perspective: “Organizations need to put learning first, profits second.” He affirms that organizations should strive for good results, but argues that good results can either mean success, or (as the HBR article also points out) quick and inexpensive failures.

Unfortunately, many organizations translate failure as, well, failure. But David Reibstien, professor of marketing at the Wharton School of the University of Pennsylvania, celebrates it. "If [marketers] don’t make it safe to fail, few will ever risk trying anything new," Reibstien says. "If they don’t try, they won’t learn and they risk competitors getting smarter and more profitable."

Staples, the office supply giant, is great at successfully using test-and-learn strategies. It began by automating its ability to run and manage campaigns. It used the time saved to explore and analyze ways to improve segmentation and

response rates, conducting many experiments over short periods of time.

One experiment tested different ratios of investment between acquisition and retention. At the onset of the recent economic downturn, Staples wanted to identify its best customers and figure out how to keep them happy and engaged.

The experiment provided a valuable lesson: Even with a significant shift in investment toward customer retention, a portion of customers will still churn. It could be for one of a hundred reasons, many of which are out of a marketer's control and are difficult to predict and change. A marketer that is not adequately investing in acquiring new customers will see its customer base dwindle.

A harsh lesson, perhaps, but executed in a test-and-learn environment, Staples quickly flipped the switch, reverted back to the proven ratio, and was able to gain market share from a number of competitors during the recent recession.

A second experiment that examined buying behavior provided more positive results. Most segmentation strategies use basic demographic information as a proxy to determine appropriate offers and channels. In this experiment, Staples analysts looked for patterns in buying behavior that suggested a customer might be in the wrong segment and then changed how they marketed to them. Response rates soared.

Trimble and Reibstien would agree that the key is not to judge based on the immediate outcome alone, but on the ability to run a disciplined experiment and adapt to the insights gained.

The bottom line is it’s not about doing the same things more efficiently. It’s about doing more in the same amount of time – or even better, in less time. Marketers gain more opportunities to learn what will generate the best result today, tomorrow and into the future. The process starts by putting learning first, profit second.

Jonathan Hornby is director of worldwide marketing at SAS.