Thanks For Calling, How Can I Help Us?

Oct 15, 2016


Making decisions about how to treat customers can be hard. Knowing how to make the right decisions—for both your customers and your business—is even harder. Defining what “right” means can be a challenge in itself: is it doing what the customer wants? Putting your employees first? Maximizing returns for shareholders? Perhaps some ethereal combination of the above? The problem of how to balance these often-competing interests is one of the most fundamental in business.

Companies regularly create guidelines and policies to try and simplify this process; business rules tell us that in X situation, we should do Y, with some flexibility in case of Z. And while these decision processes allow for some latitude in the interest of customer service, they’re still based on rules that may or may not actually be “right” for a given situation. But what if there’s a better way? Modern wonders of big data, statistical inference, and new technology offer companies an opportunity to leverage existing assets into customer-centric decision support systems that can answer some of the industry’s most complex business questions. So how might such a system work to create win-win situations for companies and their customers?

A few weeks ago, I received an automated notification about a schedule change on the return leg of my upcoming flight. My departure was being moved up by 30 minutes, while my connection was pushed back by about the same, resulting in an extra hour of travel time (and, more importantly, an extra hour of time away from my wife and daughter who I hadn’t seen for a week). The airline made this change several weeks in advance, ostensibly because it was more convenient for their network or staffing.

Frustrated by this change, I called the airline and asked to change to a difference route, which would have departed and arrived at my original times while taking me through a different connecting airport. The agent politely informed me that flight schedules are subject to change and that since my schedule was impacted by fewer than 4 hours, I would need to pay a $200 change fee to adjust my itinerary. “But I’m only changing to the departure and arrival times I originally purchased,” I pleaded. “If I’d known you were going to change the flight schedules, I would have just booked that other flight to begin with!” I knew what was coming next: “I’m sorry, sir, but there’s nothing we can do.”

As it turns out, maybe there is.

As I concluded my trip and arrived at the airport to begin my journey home, I heard a surprising announcement over the loudspeaker: my return flight was oversold, and the airline was offering free points, hundreds of dollars in vouchers, and a night in a hotel for anyone willing to fly out the next morning instead.

Overselling seats to maximize yield is a common practice in the airline industry; the challenge revenue management departments face is to use the most and best data possible to forecast how full a flight will be and adjust inventory and pricing levels accordingly. Airlines oversell because they figure any given flight is likely to have a few customers who no-show for one reason or another. However, if the airline sells too many seats, it risks having to compensate excess customers who are denied boarding. Sell too few seats and they risk leaving flight revenue on the table. A fraction of a percent difference in show rate forecasting can mean the difference between tens of millions of dollars in annual revenue. Imagine if they had some way of knowing which customers were more or less likely to show up, or even better, which customers would prefer to be on a different flight to begin with!

Revenue management departments aren’t the only ones using data to make decisions affecting customers. When I asked customer service to change my flight, they used the data available to them and the decision support systems in place (i.e. their customer-unfriendly, revenue-maximizing policy) to make a decision about how to react:

Now, imagine if the airline’s systems were better integrated and could allow them to optimize their decision-making based on data from various departments! What new capabilities might be unlocked?

Upon examining my request and checking my flight details, customer service might have received an alert that one of my fights was in danger of being oversold. Recognizing that accommodating me was in their best interest, they could have done what I asked, positioned it to me as a one-time courtesy gesture, and created a win-win situation for us both.

Alternatively, customer service could have flagged that I was interested in a potential flight change, and contacted me proactively upon realizing my flight was too full to offer me the opportunity to make that change at no cost. Airlines could also combine online flight search behavior with customer price sensitivity models to identify customers who might have preferred different flights/times, but selected a cheaper option to save money instead. A free change could be offered proactively based on extrapolated customer preferences if it becomes advantageous to the airline; at least the customer would be delighted some of the time.

Instead, with a most delicious irony, this airline’s refusal to do right by the customer ended up costing them in every way possible: they expended customer service resources responding to my inquiries and escalations, created a vocal detractor by wasting my time, and ended up having to compensate another customer who was denied his seat so they could keep me in mine.

Elicit works with clients like these every day to help them make smarter decisions rooted in real data. This typically involves breaking through organizational silos, integrating disparate technology systems, analyzing vast quantities of data, and implementing new decision support processes. Getting there is never easy, but when it’s done right, everybody wins.