Loyal Versus Flexible Customers: How to Avoid Competing on Price Alone
PROS Lead Scientist Ravi Kumar and Senior Director of Science Darius Walczak had this unique theory that could seriously impact airline ecommerce strategy. Ravi believed that competitive pricing and scheduling were not the only two factors predicting airline bookings. It turned out, Ravi’s theory was worth pursuing.
Ravi connected with Etihad Airways to model the findings, test data and further explore a solution.
So, what did he discover from the data testing? Loyal customers have a dramatic impact on bookings.
Two Airline Customer Personae
Airline customers fall into two camps: fully flexible and loyal. Fully flexible customers will take the cheapest flight despite the weird departure time from Gary, Indiana and the inconvenience of more than one layover and having to pay for luggage.
The loyal customers are at the other end of the spectrum. Think of the friends and family members you’ve traveled with in your life who have rapidly and firmly refused that cheaper flight option. It could be because their favorite airline’s main hub is in their city and it’s more convenient. It could be their American Express charges convert to free air miles. It could be that they love that airline with the comedic flight attendants. Everyone has differing motivations, but there are airline loyalists out there who will not budge on their commitment to book with specific carriers. These folks also are subscribers to an airline's loyalty (frequent flier) program and they want to fly that carrier since they get benefits like upgrades, fast access, and miles.
When airlines price lower than their competitors, we get both loyal and flexible customers. But once an airline starts pricing higher than their competitors, the models show the flexible customers leaving. The remaining segment is the loyal customer population.
So how do you use this information to maximize revenue and profit?
Putting Theory into Action
Based on this knowledge, and in partnership with Etihad Airways, we created a willingness-to-pay model that factored in more realistic competitive dynamics plus both types of customer behavior—fully flexible and loyal. Etihad was willing to collaborate with PROS by granting us access to their data enriched with competitor pricing data to train the willingness-to-pay model. This was key because synthetic data simulation methodologies are limited in their ability to model realistic interactions between competing airlines.
Through the feedback we received from Etihad, we were able to refine our models using a Bayesian, machine learning-based demand forecasting methodology. This is a unique approach that airline revenue management systems have not used before in their modeling.
The delicate balance for the PROS team was ensuring that our modeling was not so complex that it would become challenging to estimate the model parameters while still providing the flexibility to capture realistic consumer behavior. Adding new dimensions to the equation was not a simple task.
PROS Unique Approach
Our model took a distinctive approach--helping airlines take a holistic view of their overall brand vs. the competitor’s overall brand instead of just price matching. PROS enhanced the willingness-to-pay forecasting by integrating two methodologies: competitive market analytics and the customer persona model. The competitive market analytics module identifies the reference competitive price signal for each market such that this price signal is most correlated with airline's demand.
To develop a market model, we separated the overall demand volume into two customer populations: the loyal customers (who prefer to buy from their airline of choice despite higher pricing) and the flexible customers (who always buy on the lowest price point). Both customer types’ willingness-to-pay distributions were exponentially distributed. Next, PROS added the competitive price ratio model. In this model, as the competitors increased their prices, demand increased.
Apart from our qualitative testing, we performed extensive quantitative testing, looking at how these competitive models improved forecast accuracy. We took the real data and divided it into training sets and test sets. We trained our models on the training sets, and then we evaluated the forecast accuracy, constrained demand forecast, and accuracy on the test sets.
Increased Intelligence with the PROS Module
Our price modeling will give Etihad Airways enhanced visibility into the following:
- Distribution of various carriers competing in the market
- Trends and patterns related to competitors' prices in the market
- Competitors’ historical pricing strategy (how often they matched, ignored or overpriced Etihad)
- Which competitive statistics impact carrier demand the most
- What is the optimal real-time pricing strategy (undercut, match or ignore)
The (Up to) 10% Solution
Etihad’s forecast accuracy uniformly improved with PROS competitive willingness-to-pay forecaster. Etihad saw statistically significant improvements ranging from five to 10 percent, depending on the markets. Our results were outlined in the recent Journal of Revenue and Pricing Management paper we published in 2021 in collaboration with Etihad.
Airline loyalty expert and StatusMatch CEO Mark Ross-Smith recently said “Number crunchers can take a back seat” to customer loyalty marketing. Contrarily, PROS believes airlines can achieve a dominant market position because of our predictive analytics, built on a sophisticated foundation of “number crunching.” With this new forecast modeling our airline customers can achieve better forecasting accuracy while also getting one step closer toward their collective goals of contextualized offers.
To listen to a full discussion of this joint project, check out our webinar PROS Science Showcase Using Competitors' Prices to Improve Revenue Management & Pricing | PROS
And look for this innovation in one of the upcoming iterations of our product.