I often think of revenue management as the science of supply and demand. Like science, revenue management should be void of feelings and emotions. Decisions should be made with measurable data, and results should be carefully measured.
Too often revenue managers tell me they're testing a new strategy, then I find they haven't considered the basic components of a real test: hypothesis, timing, metrics and analysis.
A test is not a test unless there is a clearly defined plan to measure the results. That plan needs to include a way to analyze data free of bias. If you don’t have a plan to measure the results, you’re just shooting in the dark.
Step 1: Make a hypothesis
To test something, you have to have a hypothesis.
Simply put, a hypothesis is an educated guess: If I do X, I expect to get Y. It's perfectly fine in revenue management to guess or assume something will happen as long as you have a plan to check your assumption.
An example of a revenue-management hypothesis: If I lower the price of my suites, I’ll increase average daily rate in that room type by selling more above the standard room rate (instead of using unsold suites for complimentary upgrades).
Another example of a hypothesis: If I increase the AAA discount from 10% to 20% off the best available rate, I’ll increase revenue by selling more AAA roomnights.
Step 2: Determine your timing
The next step is to determine the timing of the test. Timing has to include three sets of dates:
The booking period is the set of dates during which you will allow a reservation to be transacted via the new strategy as driven by your hypothesis. You need to allow enough booking lead time so the results aren’t skewed by committed roomnights already on the books before the test. You don’t want to start the test once you’re already in the booking window for that customer.
The stay period is the set of dates for which the customer will be permitted to check in and out of the hotel with a reservation under the new strategy. It is the period of time you will allow the customer to consume the reservation.
The comparison period is the set of dates from which booking and consumption data under the old strategy will be compared to the same booking and consumption data under the new strategy. This is an essential component of a test.
When determining your test window, make sure it is relatively short so the revenue manager can stop and measure results and change the strategy if needed. Also, it is important to identify periods of time not affected by something unusual, such as a holiday or special event. If you’re testing a transient strategy, avoid time periods that are either heavy or light on group business that would significantly change the availability of transient inventory.
Also, the booking and stay dates for your test should be similar in nature to your comparison dates.
Step 3: Define your metrics
The third step in a revenue-management test is determining the metrics you’ll use. This is the most complicated part of the test—typically changing one aspect of your strategy affects other channels, segments and customers, and you have to consider the effects one strategy change will have on others.
The key is to identify which customers most likely will be affected by the primary group of customers in the test.
For example, if you believe that by changing your advance purchase booking window from 14 days to 10 days you’ll increase revenue by booking more advance-purchase roomnights, you have to consider how much demand shifts out of other segments into advance purchase simply because you now have inventory in that program open for an extra four days. It wouldn’t be fair to measure the roomnights and revenue in the advance-purchase segment alone. It’s likely that advance-purchase revenue will increase in this example, but you have to measure how much revenue decreases in the programs that are most likely booked when advance purchase is not available, such as retail and AAA.
Determine exactly how you will measure success. Consider metrics such as mix of sales, booking pace, occupancy, ADR and revenue per available room, and determine which combination of metrics you might need to evaluate to get the entire picture.
Determine the source of the metrics to be evaluated, such as brand reports and tools, your company’s own proprietary information and third-party business intelligence vendors.
Determine if the results should be measured using a comparison to a recent period and/or the same time last year. If you decide to measure against the same time last year, decide what level of change would be considered to be normal. For example, if demand is up 10% compared with prior year, you should expect a 10% improvement even before the test; success of the test would start at 11%.
Avoid using metrics that are too broad or that could be tainted by other factors. Before implementing the strategy, document what is already on the books so you’ll be able to identify how much business is new as a result of the test.
Step 4: Measure your results
A test is usually complete once 1) you’ve reached the last day of the predefined stay date period, and 2) you have the consumption data available.
While you need to wait for the test to be complete to measure the final results, you shouldn’t wait until then to start measuring preliminary results. Almost every brand and third-party business intelligence vendor provides daily and weekly booking and consumption data that can be analyzed throughout the test. If, by chance, your test is showing signs of critical failure, you don’t want to continue on with the test and lose revenue that cannot be recouped. On the other hand, you don’t want to completely abandon the test if you’re not happy with the results early on. Give it enough time to have reliable data to measure.
Avoid changing the test midstream. If you change your test, you’re changing your hypothesis, which means you’re starting a new test. Before you start a new test, you have to close out your original test, completing all four steps including the measurement.
Try to avoid testing too many things at once or you won’t know which strategies are working and which are not.
The best test can still have emotional creep if it’s not controlled. A good revenue manager should communicate the steps of the test to the revenue committee and clearly explain how and when he or she plans to measure the results, giving updates along the way.
At the end of the test, report the conclusion (which might be to consider the test a success and make it part of your ongoing strategy), determine whether the results were inconclusive and test another period or alter the hypothesis and test again with more or less extreme changes.