Poor models are killing the travel loyalty star, says Booz Allen Hamilton
Airline, hotel, and rental car brands used to be the leaders among consumer industries when it came to using technology to understand their best customers.
Think of the success of loyalty programs, for instance. American Airlines’ pioneered the concept with its frequent flier program. Loyalty is now a major profit center for all types of travel companies.
Yet in recent years, travel companies have fallen behind retailers when it comes to leveraging customer relationship management data. The culprit is out-dated analytical models.
Booz Allen Hamilton has done two case studies that analyze the “breadcrumbs” that travelers drop in the course of their digital shopping.
The studies are intended to drum up business, of course. Travel brands hire the consultancy to create custom algorithms for their live systems.
Yet even if you’re not a potential client, his viewpoint may intrigue you. Cosmas’s sales pitch casts light on some of the strategic mistakes most travel brands are making around Big Data.
He challenges travel executives to ponder whether their brand is asking the right questions about its most valuable customers.
Loyalty is an underachiever
Loyalty programs once had such potential, such promise. But they have become stagnant, failing to engage the highest value customers.
Right now loyalty is just a revenue model. The more you spend with a company, the more rewards you receive. This model drives additional revenue, creates profit centers, and limits market share erosion.
But loyalty programs could deliver much more if companies better understood which factors drive repeat business and market share gains. Says Cosmas:
“Loyalty, at the end of the day, is supposed to drive an incremental willingness to make a decision that may not be considered rational in a strict cost-benefit analysis.
The key question is this: When is someone being loyal rather than booking out of convenience or for the lowest price? As a travel company, I want to reward loyalty but not the other reasons people book.
The current analytics models being used by airlines, hotel chains, and rental car companies generally can’t determine when, say, a traveler books in a particular instance.
But newer models are out there. These models can help determine what that traveler was driven by: convenience, price, stickiness, or some true affinity for the brand.
The most provocative thing we found in our case studies is that the highest travel spenders tend to be more promiscuous than the lowest spenders.
To talk in broad terms, high-spenders tend to be high-spenders de facto. They spend the same or sometimes even more on your competitors. Is that a behavior you want your program to reward equivalently?
Better to try to entice someone who may spend 50% per transaction as your highest spending customer but for whom you have 100% of wallet share. That’s the one you want to target.”
Imagine there’s a traveler named Alice, who rents a minivan from Avis. The company checks its records and sees that Alice doesn’t rent with it often, and usually only rents minivans.
Avis could then compare Alice’s booking history against its database of competitor availability. It could discover Alice is only booking Avis when Hertz is sold out of minivans. What should Avis do?
Loyalty programs present the intersection of these massive beautiful gorgeous data sets. The opportunity is to get more granular.
Following a trail of breadcrumbs
Cosmas argues that travel companies need to adopt next-generation analytics models that are capable of analyzing and learning from why an individual booking, not just a group of bookings, took place.
“Retailers have embraced information theory. But travel companies haven’t. This is not the Statistics 101 that many travel executives still have in their heads from college. It’s more the computer scientists’ domain.
In traditional statistics, you’re designing models by weighting things to generate a statistically significant sample. But information theory turns it on its head by saying every incremental real-life data can improve personalization.
Every single time a customer books a stay, each of those instances, should fit or reinforce my model of that customer. But travel systems aren’t collecting and analyzing that data yet.
None of the major airlines in the US is building customer-level networks or schedules. Everyone is still very much analyzing aggregated historical data. They’re looking at distributions of things rather than individual decisions.”
Be less ambitious
Cosmas says one of the biggest pitfalls he continuously sees when data science tries to make inroads into hospitality and airlines is that the companies are too ambitious.
Some travel companies have spent millions trying to get up to speed. But their efforts flop.
A more effective alternative would be for companies to launch a handful of quick and cheap — think: three-month — projects that tackle bite-sized problems first.
The peril is that executives may set lofty goals, taking on major overhauls of their data sets, technology stacks, and internal decision-making processes. This can require tens of millions of dollars of investment upfront, and more than a year may pass before seeing a return to the bottom line.
But companies may succeed more efficiently if they adapt the lean startup ethos of Silicon Valley. Says Cosmas:
“Look for quick wins. Try small pilot tests that can result in incrementally more sustainable and operationalizable gains. Use the Agile development model: try for a 12-week proof of concept, at the end of which you decide if you’ll invest another 12 weeks and additional resources.
A comedy of vendors
Simplicity sounds obvious. So why doesn’t it happen? There are some incentives in place that tend to distort decision-making.
“Everyone is full of good ideas about how Big Data can change the world. And some of the most provocative ideas come from strategists who have never done it, they’ve never scripted or coded or had to deal with a messy, sparse data set from a travel company.
I’d hesitate to trust or necessarily make an investment of millions based on their ideas. The devil is in the details with big data for airlines, hotel chains, rental car companies or cruise lines.
It’s a little cat-and-mouse. It takes a C-suite initiative to revolutionize a company’s approach to data analytics. But the c-suite tends to talk to the idea generators and evangelists and not the data science community, writ large.”
If Cosmas is right, the travel industry is moving into era where it can squeeze out of the data how much decision was made by true passion and not necessarily out of convenience. That insight could have a mammoth effect on how business is done.
Sean O’Neill had roles as a reporter and editor-in-chief at Tnooz between July 2012 and January 2017.