The green-screen era of flight booking was hardly glamorous, but in terms of search efficiency, it was close to a golden age. According to OAG, the look-to-book ratio, the industry metric that measures the number of shopping requests per booking, was as low as 5:1.
The Internet changed that equation, sending L2B ratios into the hundreds. Bots, metasearch, and NDC later drove them into the thousands, while advanced travel platforms pushed ratios to tens of thousands in parts of the leisure distribution.
How much does all this add up to? OAG estimates that total spending on airline shopping infrastructure rose from around $100 million in the 1990s to $15 billion in the 2010s, a 150-fold increase. Industry experts say a single airline can waste millions of dollars a year on PSS charges from excessive traffic alone, before accounting for the additional expense of pricing and constructing NDC offers.
Agentic AI could take the problem to a new level, generating hundreds of thousands, or even millions, of shopping requests before a single purchase is made. The potential result? Total spending could reach $90 billion. Even more strikingly, around 80 percent of that spending ($72 billion) could support low-converting traffic responsible for less than 25 percent of bookings.
But is this an inevitable future or an exaggerated fear? And how can industry players address today’s look-to-book challenges while preventing uncontrolled and unproductive search spikes?
These are questions our engineering team is already working through as more travel companies raise concerns about search volumes and infrastructure costs. At AltexSoft, we believe there is little reason to postpone what can already be improved today to prepare for tomorrow’s reality. We can help with better agent design, smarter caching, and more efficient shopping flows to contain the L2B problem before agentic traffic makes it worse.
To get a broader industry perspective, we spoke with aviation and distribution experts and combined their insights with AltexSoft’s own experience designing search-optimization solutions for travel platforms.
Travel search mathematics
As travelers jump from one website to another in search of the best itinerary or lowest fare, they rarely think about what each search costs behind the scenes. Every query consumes computing resources and can trigger anywhere from a handful to hundreds of billable backend transactions.
For example, a month-long calendar search can generate roughly 450 separate API calls, as the system checks different date combinations, with no guarantee that this activity will result in a reservation and generate revenue to offset those expenses.
The story of L2B growth
The problem is not new, though. “Look-to-book ratios have been rising steadily for decades,” says Ann Cederhall, travel technology strategist and airline retail expert. “There has never been a reversal: Only more searches, more traffic, and more pressure on airlines. I remember KLM having to shut down its host system after look-to-book traffic pushed it into a continuous loop. That was 21 years ago, and the industry has been dealing with the problem ever since.”
What has changed is the diversity of the supply chain. As Mark Lenahan, director at a consultancy specializing in travel technology and airline distribution, puts it, “Twenty years ago, a mid-sized OTA might have had one GDS connection. Today, it is common to have multiple GDSs, aggregators, and direct connects in play to broaden content coverage or find a better fare”
Imagine you enter a flight search on a metasearch platform. Depending on the market and how the platform is designed, that single query may be sent to five, ten, or fifteen OTAs, each drawing on its own mix of suppliers. “Different intermediaries search in different ways,” Mark explains. “Some call APIs for a specific flight, while others send broader origin-and-destination requests that retrieve a wider range of options. This has created a fan-out effect, where a single consumer search multiplies into dozens or even hundreds of downstream requests.”
By the time this traffic reaches an airline, it no longer looks like a single traveler’s search but a stream of separate requests. Multiply that effect across similar and repeated queries, and the volume quickly adds up. In some cases, L2B ratios have already climbed “from 1,000 to one to 100,000 to one,” says Paul Ryumugabe, Vice President of Business Intelligence at TPConnects, a global travel aggregation and distribution technology platform. “This 100-fold increase means that look-to-book is no longer a theoretical metric. It’s a challenge that drives costs up and must be addressed.”
The emerging agentic reality raises the stakes even further. Human travelers eventually “get tired and stop,” says Martijn van der Voort, Director at AstraNomad, a strategic consultancy operating at the intersection of agentic AI, travel tech, and payments. “AI agents do not. Amadeus and Sabre have both floated figures as high as 200,000 searches for every ticket sold.”
Extensive searches drive up the cost of every successful booking. So who bears the largest share?
Who pays the price for a rising L2B
Sign up once for all future AltexSoft articles and research
If you're tired of AI slop everywhere… same here. We added a free registration to keep bots out and actual humans in.
We won't spam you, promise.