Edge cloud strategies for airlines

Introduction

Airlines struggle with high “look-to-book” ratios (low conversion), and big challenges related to complicated infrastructure and reliance on legacy, 3rd-party backend systems like Sabre, Amadeus, Travelport, or SITA. Bot and DDoS attacks are also a constant threat. 

All of these challenges make it hard to improve site and application performance, security, and operational efficiency. However, many solutions are available today by implementing a smarter edge cloud strategy.

Performance plays a role

Performance is important. It has a direct impact on bounce rates, conversions, and whether organizations achieve their goals. One of the most important performance indicators for airlines is the look-to-book ratio. This is the ratio of visitors to its website vs. the number of tickets actually sold. As reported by Amadeus, airlines used to experience lows of around 10:1, but today it often exceeds 1000:1 due to overall growth in the number of travelers, and an increasing volume of price searches made by consumers. It’s also impacted by bot traffic deployed by competitors, price-checking services, and malicious actors. Even with traffic from real human visitors, it’s now common for visitors to make multiple searches, compare routes and dates, use several price-comparison websites and do much more due diligence before a booking is made. 

The point is that web and application performance for airlines is an important way to improve look-to-book and conversion metrics, just as faster sites improve conversions in more traditional online ecommerce businesses. If you want to improve look-to-book, application performance is key.

• A 0.1-second improvement of mobile site speed increases conversion rates by 10.1% for travel sites (and 8.4% for retail sites) Source

• Improved site speed has a positive impact at every stage of a mobile funnel from a product listing page to completed order Source

• The probability of bounce increases 32% as page load time goes from 1 second to 3 seconds. Source

Security plays a role

None of the performance points above matter at all if the inventory for a flight is blocked from visitors by a malicious attack (DDoS and bots, for example). Airlines face the same security threats as most highly visible enterprise organizations, but with unique bot issues that can even come from competitors and price comparison sites deploying bots at frequent intervals to make ticket price inquiries. Lack of security can hurt performance, but some security strategies can also introduce performance bottlenecks that wipe out any performance gains.

Results

Largest Contentful Paint (LCP) and Time to First Byte (TTFB) are two metrics used by Google when evaluating the performance of web applications. These metrics report on initial load times of a site when a visitor lands on it, and they are not representative of every element of site performance that happens after that initial load, but they’re a good starting point to compare how performative an experience will be across different sites because there is publicly available data. When looking at LCP, Google defines a “good” user experience as 2.5 seconds or less, anything over 4 seconds is considered “poor,” and anything in between is considered “needs improvement.” But it’s not just about hitting these marks – conversions and user experience always benefit from performance improvements, whether your starting point is in the red, yellow, or green. 

For the fifteen airline websites in this report, as ranked against Google’s Core Web Vital thresholds:

• Zero airline sites in this set showed a “good,” with LCPs of 2.5 seconds or less

• Four airline sites showed a “needs improvement” score (Emirates, Aer Lingus*, Lufthansa, and Iberia*)

• Eleven airline sites showed a “poor” performance (American Airlines, Southwest Airlines, KLM, Air France, Air France, Turkish Airlines, China Southern Airlines, United Airlines, Delta Air Lines, British Airways, and Air Canada) 

More information about how airlines were selected is available at the bottom of the report.

Optimization challenges for airlines

Look-to-book ratio is one of the most important KPIs for airlines trying to improve their sites and applications, and it’s impacted in a major way by two factors: 

1. Site performance 

Poor site performance contributes to lower conversions and more abandonments similar to how it does with more common ecommerce examples, and in other parts of the travel and hospitality industry. In addition, airlines make more profit by selling tickets directly, rather than through travel aggregator sites that resell their tickets and receive a cut of the revenue in exchange. If an airline’s site is performing badly it can drive visitors to another site to purchase the same ticket and cut a chunk of profit out of the airline’s revenue for a sale they should have been able to make themselves. 

2. Site security

Ineffective website, app, and API security can contribute to bad performance and create big drags on look-to-book ratios. Bots are a particularly big problem for airlines. They do a ton of price scraping and various activities that mess with inventory and seat availability. Sometimes bots will falsely hold tickets to make them unavailable, even for entire flights at a time, and then release them all. The reasons for this behavior aren’t always clear: they may have an interest in pushing customers to purchase elsewhere, or are trying to reserve availability on flights until a later date for some reason. This may be actions from tour operators or pricing engines, or it may just be malicious operations that cause mayhem and costly issues for the airlines. But no matter the intent, airlines will pay site hosting and infrastructure fees to serve bot activity that can hurt their customers and their bottom line. 

Airlines face several challenges addressing these issues, including: 

Complicated infrastructure

Price, inventory, and trip data comes from many sources, including one or more 3rd party back-ends whose performance is out of the airlines’ control. On top of that, if someone is booking a multi-leg flight where one or more legs are served by a different airline, then they need to aggregate the data about the flights they serve with the flights they don’t serve. The ways these systems are strung together are often archaic and fragile, and its technical debt that the airlines themselves cannot address directly – they have to wait for partners and providers to make improvements as well. 

Comparison shopping 

Shoppers today do far more comparison shopping for flights than ever before, and this means many requests per completed booking. Airline booking infrastructure must be built to scale to huge proportions, so faster solutions may be sacrificed in favor of greater stability or scalability. Engineering teams build with scale as their primary concern, and performance as a secondary issue.

Risk aversion

Airlines are risk-averse to failures more than other industries because they need to have a public perception as stable and reliable. They’re not willing to move fast and break things the way some smaller retailers or publishers may feel comfortable doing. Aside from perception, downtime means significant lost sales, not only through lost transaction volume, but also by potentially driving more sales to travel sites and aggregators during that time that cut into the airline’s profit margin for tickets that would have been bought directly. 

Edge cloud strategy for airlines

Optimizing a complicated infrastructure is hard. It’s even harder when you don’t own or control big pieces of your data supply chain or rely on a 3rd party vendor who constitutes a lot of your backend. Both are true for the airline industry, and in a situation like that you have to look to what you CAN control, and start making gains there. Strategically moving more to the edge is a smart place to start, but it’s important not to recreate the complexities on the edge as well. Point solutions that exist on different edge networks and get cobbled together don’t bring the same benefit as a more unified approach. 

Edge cloud platform benefits

You shouldn’t think about addressing CDN services, edge security improvements, and edge computing without thinking about looking at unifying them on a single network. Not all edge cloud is created equal. For example, a company might offer edge security, edge computing, and a CDN network under the umbrella of a platform, but what’s really under the hood are three completely separate networks that are basically duct taped together under a brand name. Platform benefits are also unavailable when engaging with point solutions like a WAF vendor who doesn’t offer other edge capabilities. In this case your WAF is operating separately from the other potential activities at the edge that are important to your organization. Optimizations across security and edge computing, or security and CDN services are much more limited. Some security services like DDoS protection actually benefit from protection at different layers of the network, and there are disadvantages if your DDoS protection is split between a CDN operating at a network layer, and an unrelated WAF operating at an application layer. 

It’s important to select solutions that actually run together on one network in order to get the performance and savings benefits necessary to make an edge strategy successful. As a bonus, you also get other benefits like higher developer productivity. There’s less context switching for DevOps and SecOps because more of their toolkit and reporting is under a single dashboard, and vendor consolidation can result in savings as well as more predictable budgeting. Unified customer support experiences also means that when you run into problems across different parts of your edge services, the team you engage with for help has insight into your whole story, rather than a limited view. 

But the platform efficiencies are just the tip of the iceberg. Here the some of the specific benefits that will come with selecting a unified edge cloud platform:

Enhanced CDN, delivery and network services capabilities

In short, because so much of an airline’s infrastructure is hard for them to control, they should aim for the best possible CDN performance on top of CDN capabilities that allow them to cache, optimize, and configure as much as possible, and with as much flexibility as possible. This could include caching dynamic content and APIs (plus their responses), performing image optimization and transformation at the edge, and being able to reconfigure things as needed, and without delay. Faster purge times (instant if possible) should be applied to any content on the site after changes and updates to remove it for improved performance and better user experiences. These are the first places to start. 

Enhanced security capabilities

Better security and bot protection can help airlines ensure their sites stay up and available, but the best solutions can do so without causing additional latency. Preserving site availability is important, but you don’t want to offset your availability gains with latency that harms your look-to-book ratio. 

Bot and DDoS protection is important for performance. Bots can cause problems for airlines with incessant scraping of the site for pricing information. More malicious bot activity can aim for account takeovers, personal data breaches of customer accounts, fraudulent bookings, use of stolen credit cards, and sometimes bots will meddle with inventory by holding large numbers of tickets so that entire flights become unavailable to consumers. Worst of all, a DDoS attack could make the whole site unavailable for long stretches of time – and not just the customer-facing applications, but internal applications as well like those used for flight management, scheduling, and more. If internal applications aren’t protected, they can be vulnerable to DDoS and ransomware attacks that cause cascading problems. 

Some security solutions, especially on-prem WAFs, can create their own bottlenecks as requests for security checks pile up in the queue. Edge security solutions can help, but not all edge security solutions are built equally well for latency. If your security checks happen on a separate edge network from the rest of your operations then there’s a latency tax every time a request has to hop from one network (like from your CDN network or edge computing) over to the security network and back.

Enhanced edge computing capabilities

Doing more at the edge on a modern edge cloud platform is a great way for airlines to build modern, performant solutions on top of a legacy backend (such as Sabre, Amadeus, Travelport, or SITA). Identifying storage and workload opportunities that make sense to move to the edge can bring big performance gains and other efficiencies as well. Three examples of edge computing opportunities for airlines are: 

1. Advanced Caching and Personalization at the edge - Personalization is expensive in terms of performance because it adds latency, and it can also be expensive in terms of money because of its impact on egress charges as less is served from cache, and more unique calls are made to origin for each individual user who shows up at your site. Edge computing makes it possible to move more storage and workloads to the edge so that operations can be handled there rather than at origin – saving both time and money. Airlines could explore using the edge for: 

• Caching personalization data for customer profiles in edge storage with a Time To Live (TTL) expiration like traditional caching so that user data is ready for subsequent activities after their first action

• Bringing up flights and routes that are likely to be searched based on a user’s history or location

• Storing interactive seat maps for the planes on the edge

• Proposing seat assignments based on past preferences to reduce friction in the checkout process

• Look for other ways to autocomplete more of the booking process by storing a customer profile at the edge

2. Workloads on the edge - When edge computing is in the mix, it’s like being able to cache more complicated data and workloads because they’re stored or running at the edge. Airlines could explore more advanced caching opportunities like storing the seat maps for their planes in edge storage so they come up instantly and make users feel like response times are great. It can be taken further with a headless commerce approach where more of the front-end presentation layer is decoupled to the edge for a speedier interaction that is only dependent on the backend for populating the data not available in cache or from edge storage.

3. Savings on egress charges - Doing more on the edge also means you don’t have to go back and make requests from your origin as much, and save on the egress charges that get racked up through those requests. Commonly used data and workloads should be identified so that they can be stored or run at the edge rather than within legacy backends. This will result in better performance while saving meaningful amounts of money. 

Additional information on this report

Airline inclusion decisions

The top 11 airlines by revenue were selected from this Forbes Global 2000 list for 2023. Two of these 11 are holding companies rather than individual airline brands, so the largest airline brands held by each were included, while smaller and regional brands like Transavia, Vueling, and Level were excluded. Emirates was also included even though they are not on the Forbes list. As a state-run company, they were not included by Forbes, but they are a major player in the airline industry.

*Air France-KLM is represented by Air France and KLM. Furthermore, Air France runs separate websites with unique performance data for each country they operate in, so the two instances of their website with by far the most traffic were included – the sites for the United States and France. 

**International Airlines Group is represented by British Airways, Iberia, and Aer Lingus. 

Methods

We focused on Time to First Byte and Largest Contentful Paint to measure performance because they’re good measures of initial loading activity for a site, and because they can be gathered and measured from publicly available metrics. These are not the only performance measurements to look at, but they are important and have real impacts. 

The table in this report shows the loading metrics of airline websites from Dec 10, 2023 - Jan 6, 2024, based on CrUX data.

• Main website only

• All pages

• All form factors

• 75th percentile

The CDN which delivered the website was detected 10 times from 3:27 PM - 3:30 PM PST, Jan 8, 2024 using Method 1.

You can read more about these metrics and methodologies here.

Let's look at your performance.

Answer a few questions and our team will tailor a list of recommendations to meet your needs.

Talk to an expert