The history of in-flight service is rich with – in some cases apocryphal – tales of carriers removing one peanut or piece of fruit per passenger and consequently saving tens of thousands of dollars a year. Such decisions were straightforward, born of a desire to save money.

Today, big data offers a theoretically much more nuanced and powerful approach to optimising passenger services, carrying the potential to reduce wastage and drive new revenues. That might, for example, mean tailoring the products loaded on the aircraft to individual flights or customers, or personalising the ancillaries offered during the booking process.

Such initiatives can be facilitated by the structured harvesting of existing airline data – everything from onboard sales through to a specific customer's buying habits.

There are limitations to consider, however. Presuming the recording and collation process is to a high standard – a significant challenge in its own right – data is great at telling you what has happened. But establishing the "why", and how the business should therefore respond, is fraught with pitfalls.


Confirmation bias is a common problem, for instance. This occurs when data is wittingly interpreted to confirm a pre-existing hypothesis. For airlines fixated on big data as a money-saving opportunity, that is a real danger. There is also the potential confusion between correlation and causation, where it is incorrectly assumed that because two things co-exist, one is the result of the other – when both may have been caused by a third factor.

When hasty or ill-informed interpretation of data happens, all that is likely to be achieved is the entrenchment of suboptimal norms – at best.

Looking at sales of in-flight food as an example, data will not tell you why very few people are buying your sandwiches. But a carrier with a fixation on using data to cut wastage might see that data and cut the number of sandwiches loaded on to aircraft, then celebrate the financial saving. In reality, however, it could be missing out on a significant growth in revenues, simply because it is ignoring the fact that its cheese and tomato baguette is revolting.

Data might also tell an airline that duty-free purchases on evening flights to holiday destinations are very low. Could it be, however, that overworked cabin crew are spending so long on the food and drinks service that duty-free sales fall by the wayside? Based on data alone, cutting the amount of duty-free loaded on such flights might seem logical, but perhaps more revenue could be achieved by instead rethinking how service is delivered.


A similar conundrum applies to personalisation strategies. It might be established, for example, that a certain passenger never books hotels through a particular airline, only flights. One interpretation might be that marketing hotels to that passenger is a waste of time; more revenue could be achieved by marketing other products to them instead.

Making that decision based on data alone, however, risks ignoring other factors – might your hotel offering be poor or too pricey for a particular destination? Might this passenger prefer to book flights first, then hotels much closer to the departure date using another provider? Again, data will capture what has happened in a particular scenario for a particular business, but not why.

Problems with interpreting data still vex those dealing with some of the most critical information handled by humans – such as that associated with trials of life-saving medicines – so it should be no surprise that airlines would be vulnerable to the same.

None of this means big data is a bad thing – far from it. Handled correctly, it carries the potential for airlines to differentiate their services in what is an increasing commoditised market.

But big data alone cannot provide answers to everything. A holistic and open-minded approach to decision making will always be required, as will a recognition that even those most adept at interpreting data face a daily fight to ensure they are reaching optimal conclusions.

FlightGlobal has launched the Aerospace Big Data Conference series in Miami, London and Singapore, find out more:

Source: Cirium Dashboard