Have you ever wondered what happens behind the scenes when you search for a flight?

Jerome Daniel

Head of Search, Shopping & Pricing, Core Shared Services, Amadeus

This content is only available in this language.

“This is your captain speaking. Please fasten your seatbelts for take-off.”

To many, this phrase brings joy. It can mean the beginning of a new adventure or that you will soon reunite with your loved ones. To others, it may bring anxiety: another tiring business trip or a call back to reality. 

What is certain is that if you’re traveling, you will encounter many steps before you hear this phrase. These include research on the chosen destination, asking your boss for some days off, going to the airport and passing through security… But one of the first steps, whether traveling for business or leisure, is searching for your flight. 

Initiating a search for a flight online is easy – be it through an airline.com, a travel agency website, or a metasearch site like Kayak.com, you just need to enter the departure city, destination, and dates. Or sometimes you don’t even need to enter all of these as some search engines give you the flexibility of seeing a range of dates or destinations by price.

But what actually happens behind the scenes? In contrast, the technology at work that brings up the possible best combinations in a matter of milliseconds is more complex than you think. 

Did you know that when a user searches for a flight in an online travel agency, Amadeus considers more than 562,500 possible itineraries and 560 billion possible fare combinations? However, only the top 250 itinerary recommendations are shown, a very small sample of all possibilities. So, have you ever wondered how these itineraries are built?

Algorithms and machine learning: the keys to building the best itineraries

Itineraries are built using an algorithm, which is a set of rules to find specific information among a collection of data. Machine learning improves the quality of the results by calculating trends and other characteristics based on historical data. 

This data comes from the airline’s schedules and includes elements such as the flight number, departure city and arrival and time, the type of aircraft and whether there is a code-shared flight with another airline. Then the types of itineraries are also taken into consideration - for instance, whether it’s a direct flight, or a connecting flight in the airline’s hub or with a partner airline.

After combining these very large sets of data with a series of complex algorithms, travelers get access to the best possible itineraries to book their flights. 

However, more often than not, travelers overlook the most convenient itinerary in favor of getting the best possible deal. 

Up & down: why do flight prices change constantly?

Many travelers will have faced a situation where they wait to buy a ticket only to find that the price has changed. Prices constantly change simply because the data used to set prices are constantly changing. Schedules, fares and availability are continuously updated by the airlines to optimize their sales in an increasingly competitive and complex environment. 

Amadeus receives new schedules several times a day. We get new fares and rules in as short a span as every 15 minutes. However, the most volatile element is availability, which at peak times, can change a few times each minute! 

Ultimately, the price of an itinerary depends on multiple factors:

  • Fares published by the airline: these will vary depending on many factors such as type of passenger (adult, child, infant), day of the week, time of the day, seasonality, carrier and flight number, duration of the trip, number of connecting flights, and others such as fuel and airport charges.
  • Airline revenue management systems which model customer purchasing behaviors. They rely on big data andMachine Learningmodels which triggers opening and closing booking classes with the aim to “sell the right product to the right customer at the right time for the right price”. 
  • Negotiated fares: travel agents can negotiate with airlines specific fares that will only be distributed by them 
  • Commissions: travel agencies can sometimes apply a commission on top of airline prices.

There is a best time to buy flights. Myth or reality?

Unfortunately, there is not an ideal time to buy flights. There are opportunities to find cheaper flights, but it’s a job in itself. This has been the original idea behind businesses like lastminute.com, an online travel company that initially based its model on the re-opening of lower price class availabilities when flights were not full enough before departure. 

Looking into the future with artificial intelligence

As travel agencies and airlines focus on delivering a more personalized offer to travelers, machine learning andartificial intelligencewill play a fundamental role in the future of fare search. Right now at Amadeus we are at the forefront of what these technologies can bring, looking into areas such as customer choice modeling – in other words – how the best combinations with the best prices are selected from all possible options. This helps us find the best solutions for a certain traveler profile, allowing our customers to maximize sales.   

Although shopping online for plane tickets may seem straight forward, flight search is a highly complex task from a mathematical point of view. There can be hundreds of thousands of possible solutions to a given flight search so considering all of them is hardly realistic nor feasible with a fast response time. As travelers won’t settle for anything but the lowest deal, with the backdrop of a rapidly shifting industry, our teams are harnessing the power of technology to continuously improve our capabilities. 


Flight search - the facts*

  • There can be up to 1 million possible fare combinations for a single itinerary
  • 57% of users choose the cheapest option when searching for a flight
  • 24% of users choose the top 20 cheapest flights
  • 11% of users value convenience/service over price

*Based on Amadeus research


Do you want us to email you the latest insights?

Subscribe to updates


Airline IT, Travel sellers, Machine Learning, Travel Search, Artificial Intelligence, Big Data