In his most recent book, Great By Choice, Jim Collins categorically states that “in the face of instability, uncertainty, and rapid change, relying upon pure analysis will likely not work, and just might get you killed. Analytic skills still matter, but empirical validation matters much more.” And most executives agree.
Fundamentally, these executives want to distill causal relationships between their own actions (e.g. I changed the room rate for weekends) and their consumers’ reactions (e.g. Did the customer tweet about the new rate? Did they go online to find out more? Did they compare rates on an OTA website? And did they actually book because of the new rate?). Economic changes, competitor actions, geographic factors, and other “noise” make isolating cause-and-effect relationships between a business action and consumer behavior exceptionally difficult. The most robust way to isolate the impact of any change is to try a new business idea in a few hotels or with some guests, analyze whether it worked and where or with whom it worked better, and then target the rollout to maximize ROI.
This robust “clinical trials,” or Test & Learn approach to decision-making can be used to evaluate a gamut of questions, for example: how should I vary rates between 3rd party channels vs. my brand’s website? What are the true incremental bookings from promotions? How do I correctly set brand standards and motivate franchisee investment? How would a new breakfast offering impact my limited service brand? Hotels have always analyzed large amounts of data to measure the performance and implications of various programs. However, isolating cause-effect relationships between decisions and their impacts has traditionally been difficult.
Frequently, there is too much “noise” in the data, and different analytical assumptions lead to confusing or even conflicting answers. Informal testing that isn’t conducted in a robust, repeatable manner often creates more questions than it answers. The issue for most hotel chains is not that they are not conducting tests, but that the methodologies they use (e.g. comparing test hotels to the rest of the chain, or another group of hotels in the area) often lead to wrong or inaccurate decisions.
"Quantifying the impact of new programs or initiatives is particularly difficult for the hotel industry, due to the industry's extremely cyclical and seasonal nature, the impact of local conditions, as well as ongoing competitive activities," said Wayne Goldberg, President and Chief Executive Officer of La Quinta. Various stakeholders distrust results, and they have good reasons to: maybe the test hotels were not representative of the chain, or the impact could not be determined in the short time frame, or maybe competitors closed or opened properties in the test markets, or simply warmer weather led to better RevPAR.
Over the past decade, advances in computational power and database technology have allowed industry leaders to reliably test and distill cause-and-effect relationships. With an abundance of loyalty program data, hotels are well-positioned to make profitable decisions backed by rigorous cause-effect findings. Ultimately all scientific tests should answer three key questions about any new initiative or program: (1) Does it work?, (2) Does it work better in some locations or with guests than others?, and (3) What are the most valuable components of the action? This evaluation can then predict the impact of any tested action on any location or guest, maximizing returns while focusing costs. As Wayne Goldberg observes:
"[Test and Learn’s] ability to cut through the noise and accurately predict the impact of various programs has enabled us to make business-critical decisions with confidence. [Test and Learn] has already aided us in identifying the highest ROI remodel strategy, optimizing our sales force and reducing the cannibalization impact of our growing brand. By helping us accurately measure the impact of our initiatives and prevent costly missteps, the [Test and Learn approach] is becoming core to the innovation process at La Quinta.".
There are certain best practices that every hotel chain needs in order to institutionalize scientific testing:
- Choose a Representative Sample: Where possible, design tests with properties that are a representative sample of your hotel chain. Oftentimes, we see that changes are made in very specific markets or in only a few properties, which makes it difficult to generalize the findings from the test.
- Measure Incremental Impact: It is not sufficient to measure what happened just in the test hotels. Sound test analysis involves measuring the “lift” or incremental change in RevPar, profits, etc. as compared to a similar set of hotels where there was no test (i.e. the control group) before and after the initiative was launched.
- Identify Performance Drivers: No program will work well everywhere, and some great ideas may not be useful to implement across the chain. Segment the results of a test to understand which types of properties respond well to the program.
- Learn: Record key lessons, disseminate / provide broader access to them, and use findings from current tests to generate ideas for future testing.
As emerging technologies and competition disrupt business models and upset well-worn strategies, 13 of the top 25 hotel chains are already using scientific testing to de-risk decisions and generate tens of millions of dollars in incremental revenue. Bill Carlson, Senior Vice President of Performance Analytics at Choice Hotels recently commented that, “the hotel business is dramatically changing with the advent of new technology, new industry participants and new channels for booking hotels … this new environment provides great opportunities but potential risks/costs as well. Now more than ever, it is critical for us to precisely allocate marketing, capital, and promotional resources in order to generate maximum profitability for Choice Hotels and its licensees. [Test and Learn’s] ability to accurately predict the impact of various decisions helps us foster confidence before rolling out strategic ideas in the quickly changing hospitality industry.”
Across industries Test & Learn is ending the internal debates on the value of various initiatives, and unlocking profits through powerful and nuanced insights. For example, Family Dollar Stores used a Test & Learn approach to understand whether it was worth installing refrigeration units in its 6,800 outlets that, up to that point, had sold only dry goods. Based on a test of only a few dozen stores, Family Dollar found that the impact was far greater than just the sales gains in milk, eggs, and frozen pizza. The bigger impact on profit came from increased volume in its existing business. Data-driven targeting of the rollout drove substantial value. Taking a similar approach, Wawa tested and discovered that adding 50 hours of store labor per week produced dramatically higher sales. In Europe, Boots applied test versus control analysis of past investments to inform the best deployment of their store refurbishment budget.
In the current volatile economy, the key to success lies in knowing the true incremental benefit of any action. By correctly giving credit where credit’s due, leaders can effectively speed up the adoption of value-adding activities, curtail those that don’t work, and, most importantly, tailor and target the more promising initiatives for enhanced profitability.
Maryam Wehe and John Howard are senior executives at Applied Predictive Technologies, a predictive analytics software company that is used by several dozen of the world’s largest corporations to apply structured experimental methods and determine the causal relationships between business programs and financial outcomes.