As pandemic restrictions are unwound, travel demand returns, and borders reopen, airlines are facing consumer behavior and competitive landscapes that look entirely different from the world that existed two years ago.
How can airlines forecast demand when our historical data is meaningless? Many are turning to intent data to fill this gap.
Such data is derived from online text searches via engines like Google, plus airfare query data from online travel agents (OTAs) and metasearch providers like Kayak and Skyscanner.
Many of these companies are providing airlines new data streams to help decide which routes to restore first, how much capacity to offer – and even how to price tickets.
Transitioning from Traditional Demand Data to Intent Data
“Before COVID, we were primarily using traditional sources like [US Department of Transportation] DB1B and T-100 for traffic and fares,” says Jonathon Nield, who managed network planning at Frontier Airlines throughout the first year of the COVID crisis.
Such files are often released months after departure, reflecting market realities that may no longer exist.
“Once the pandemic hit, we started leveraging new types of data because capacity became meaningless beyond one to two months out,” Nield adds. “Historical demand data was just pointless.”
The team at Frontier Airlines was already using first-generation data products from Google and Expedia before the pandemic and were trialling others.
“We relied heavily on Google because they were very deep into development of their data products, although many of these were still in beta testing. They worked with us to help develop the products and validate their data.”
“Historical demand data was just pointless.”
Google started out providing organic search queries by origin and destination (O&D), which were largely based on keywords like “flights from Philadelphia to Cancun.”
Nield says this data offered basic insights but didn't have the same crispness as searches with specific dates and other parameters.
Intent Data is Just the First Layer of Airline Transaction Data
Intent data comprises the first of three sets of ticket purchase data: ToFu, MoFu, or BoFu (Top of Funnel, Middle of Funnel, Bottom of Funnel).
ToFu describes how many searches are made, by how many shoppers (including an indication of party size), to which destinations, and for which potential travel dates.
After the preliminary search but before the booking transaction, the shopper is shown a series of itineraries with live pricing.
This MoFu data can provide some indication of consumer willingness to pay based on what price points result in consumers clicking through to an airline website, or what meta providers call “redirects.”
“Customers are actually choosing these fares,” says Skyscanner commercial manager Thomas Walsh, “so an airline can know the customer is willing to pay them.”
"MoFu data can provide some indication of consumer willingness to pay based on what price points result in consumers clicking through to an airline website."
On a carrier's website, it is more possible to observe other MoFu data points like abandoned shopping carts or look-to-click ratios for ancillary products. Heat maps and other tools help illustrate the shopper’s journey through the airline "store" and which products make it to the checkout page.
BoFu is ticketing or booking data from an airline’s own data systems or third-party sources like IATA's DDS, GDSs (e.g., MIDT data from Amadeus), or OTAs.
Growing Data Sophistication Highlights Opportunities for New Routes, Capacity Optimization
Later in 2020, Google released its Google Flights product, which was normalized around historical volumes and gave context across the entire selling window.
This allowed airlines to look at specific origin cities, O&D city pairs, and search dates. “We were able to apply this data to compare search volumes in, for instance, December 2019 vs December 2020 to give us a view on the strength of recovery,” says Nield.
“We could see which states were hot spots for demand, and which ones were less recovered. It became clear there was a propensity to travel to places that weren't all that popular before COVID, such as ‘Great Outdoors’ states like Montana or North Dakota.”
This data helped Nield’s team see where new demand was materializing, and which markets were likely underserved. Google augmented this with the percentage of Google queries by O&D that were resulting in traffic to the carrier’s website.
Airline route cases are often network-led, as was the case before COVID. However, the planning cycle at most airlines has gone from 6-12 months before departure down to just 2-4 weeks. Intent data is fed to airlines on a weekly or even daily basis, driving nearly real-time capacity and pricing decisions.
“We could see which states were hot spots for demand, and which ones were less recovered. It became clear there was a propensity to travel to places that weren't all that popular before COVID."
Metasearch provider Skyscanner captures 295 distinct data points, not just about the potential purchase but also the consumer: their nearest airport, distance from that airport, etc.
The company has agreements with several carriers to provide its data to airlines, airports, and destination marketing organizations (DMOs).
“Airlines are using this data to answer the question: ‘If we put an aircraft on this route, what's the likelihood we can fill it?’,” says Walsh. “This also provides airlines with a view to new destinations.
We can even extrapolate from our search rates and [meta] market share by country to estimate O&D market sizes.”
He adds that airports and DMOs are using the data to know which other destinations are competing for traffic from a given origin market.
Itinerary Search Data Drives Market-Specific Insights
“Let's say Las Vegas is down 60% year-over-year in search volume, but capacity was down 80%,” Nield explains.
“That could be used as a marker for us to look at increasing flying, because it was potentially a sign that the decrease in capacity had outpaced the reduction in true demand.”
However, Las Vegas has long been known as a market with lots of fare shopping but relatively low conversion.
So, Frontier Airlines' network planning team developed metrics to evaluate the volume decrease, capacity decrease, and the historical search conversion rate to decide whether to add capacity from a particular origin.
Nield says, “Looking at the O&D data allowed us to avoid piling onto the same routes that most other carriers started flying.
For instance, we knew that every carrier would pile on in Florida, Cancun, and the Caribbean, because beach destinations were obviously popular as soon as leisure demand began to recover.”
“But intent data also helped us in unique ways to expand our footprint into new origins from which we could profitably serve those destinations with less-than-daily frequency.”
“Looking at the O&D data allowed us to avoid piling onto the same routes that most other carriers started flying."
Search trends also provide specific data on how customers react to virus outbreaks. “Whenever Australia has another outbreak,” Walsh says, “we see daily search volume falling by 20,000 to 30,000.
Customers also shop for a shorter average stay at their destination, average party size falls, and the booking window moves closer to departure dates.”
According to Skyscanner data, Europe’s top unserved routes are Manchester to Bangkok, Amsterdam to Cancun, and Amsterdam to Zanzibar. There also appears to be significant latent demand from most German markets to Zanzibar.
How is Raw Search Data Turned into Actionable Insights?
Intent data providers take raw search data and run it through a series of filters to improve accuracy. First, search bot traffic must be separated from actual human users.
Duplicate searches from the same user (or group of users) across different days must be consolidated to avoid oversampling. Metasearch queries also provide the itinerary party size for each “intent event” or booking request.
Data summaries show all the itinerary options a user was presented during their intent event and what they ultimately chose, providing an estimate of potential market share at specific price points.
"Airlines can compare actual bookings to metasearch data to come up with a more precise estimate of how searches convert to buying."
Metasearch data usually lacks insight into the final transaction as the meta site isn’t a party to it, but they know how many shoppers clicked through and might have ticketed after that redirect.
Airlines can compare actual bookings to metasearch data to come up with a more precise estimate of how searches convert to buying.
For an airline with operations in more than one region, working with multiple data providers is advisable as the dominant OTA and/or metasearch provider can differ by country.
As global air travel continues to recover, these new data sources are already revolutionizing how airlines identify and respond to demand trends in their markets.