Redefining Revenue Management and Marketing Data and Analytics through the COVID-19 shock
Part 1 of 2: Introduction + Data Signals
by Kelly McGuire and Naman Gandhi
Hospitality companies had started to embrace data-driven decision making and invested in analytical systems to support revenue management, marketing and operations. After the longest ever period of economic expansion, and continued innovation, some executives were waiting for the other shoe to drop. And it finally did, in a way that no one in hospitality would have predicted.
When COVID-19 was declared a pandemic in March 2020, the hospitality and travel industries quickly became among the most affected sectors, and even as companies learn to reopen safely, demand has been spotty and volatile. The US FDA issued emergency authorization of the first vaccine on Dec 11, 2020, and yet, the trajectory of the travel and hospitality industry’s recovery remains uncertain as infection rates and virus mutations continue to cause travel restrictions and erode consumer travel confidence.
Complicating the recovery are the macro-economic trends and micro-consumer behavior changes created by this pandemic, which are much more extreme than previous global events. The markets continue to be highly volatile, so it is difficult to detect any persistent travel/stay patterns. It can be said that while the initial crisis is past, the “new normal” has not yet appeared. Instead, industry is in a period of “business unusual”, which will be marked by volatility and continued uncertainty.
This drastic disruption means historical data, even recent history, is not a reliable predictor of future patterns. The hospitality industry is now confronted with a unique and unprecedented problem. Traditional leading and lagging indicators are not behaving normally. Macro and micro economic trends are unstable at best. The demand patterns and consumer behavior trends that commercial leaders relied on in the past evaporated overnight. There are no precedents to follow, and the analytical solutions that supported key functions are struggling to respond properly to current market conditions.
In this two-part blog series, we discuss two ways to overcome the impact of short-term disruption to analytical models: (1) new data will need to be leveraged and existing data signals suppressed until traditional sources stabilize, (2) algorithmic adjustments need to be made to adapt to new economic conditions including increased uncertainty and volatility. Both methods will need to be deployed to ensure that insights can be relied on to support business decisions.
Fine-tune the data signals
All good analytics start with good data, so the first step is to assess the data. As you approach this evaluation, it is critical that you do not discard any data, either historical, pre-pandemic data or pandemic data itself. Any data adjustments made now are a short-term fix. Historical data will be a “foggy window” into the new normal as patterns stabilize and could also be useful today to understand things like seasonal patterns or guest preferences that have not been impacted by the pandemic. Current data will obviously be more reflective of recent market conditions but may also be useful in years to come should a similar disruption occur (God Forbid!).
Data should be categorized according to whether it is a leading signal, evolving signal, stable signal or experimental signal. Table 1 is a version of our implementation strategy recommendation.
Based on our data assessment, the best way to understand how demand is materializing is to evaluate economic and health indicators, along with upper-funnel and meta-search data. Reduce reliance on historical booking pace, occupancy and mix, as these will remain unstable through business unusual. Recent booking patterns should be continually evaluated to pulse-check recovery rate and customer travel confidence in the market. Keep in mind that demand patterns and recovery pace will remain hyper-local until global herd immunity is achieved. Hoteliers will need to track patterns at an extremely detailed location and segment level to ensure that all opportunities are identified. Ensure you take a close look at your competitive set to evaluate whether any changes in the composition or the strategy are needed. While benchmarking against a comp set is important, without knowing what the competitors’ survival needs are, over-reliance on competitive set actions will result in deep discounting and price wars. Carefully evaluate who is included and how you plan to react to their pricing and promotion actions.
Fine-tune the analytical models
In Figure 1 below, we have depicted the analytics ecosystem of hospitality, and we have filled each box with shade of the black gradient to denote the average degree of disruption to the current process caused by the recent economic shock. The actual impact may vary depending on each company’s implementation complexity, the sophistication of each model, and the specific reliance on historical/longitudinal time-series data.
Price sensitivity, demand forecasting, and consumer behavior models including segmentation, promotion response and recommendation engines are the most impacted due to their typical reliance on historical data. In next part of the series, we discuss how we can make appropriate adjustments to these analytical models to ensure they provide reliable insights through business unusual and into the recovery.
Check out Part 2, where we describe how to adapt analytical models to manage through “business unusual”.
