This is the first of three essays on how airline commercial intelligence is changing. The series follows one evolution in three waves: the competitive benchmark that pricing teams still run every day; the shift from filed fare to full offer, and from cleared price to search intent; and a platform change arriving from outside travel. The question underneath all three is the same. The market airlines price against is moving faster than the instruments built to watch it.
The daily discipline has not disappeared. In a pricing meeting the first question is still comparative. Where do we sit on this city pair, this date band, this competitor set - today, not last week. Revenue management, network, and distribution start from that same picture. What has changed is the object being measured. Continuous pricing, branded products, and channel-specific offers mean the market does not sit still between filing cycles. The snapshot is already ageing when it is taken. Velocity, granularity, and cost are dissolving a benchmark that the organisation still needs.
Continuous pricing breaks the periodic snapshot
Filed-fare intelligence assumed a market that updated on a schedule. Carriers published a ladder, intermediaries cached it, and a morning pull could stand in for the day. That assumption is no longer safe. OAG has reported that roughly 260 carriers now apply some form of dynamic pricing, an increase of about 20 per cent in two years. ATPCO set a public target that 80 per cent of airline offers would be dynamically created by 2026. The figure is better read as a statement of industry direction than as a census. Even so, the direction is not ambiguous. Prices are being constructed at shop time rather than retrieved from a file that the whole market can see.
The economic case for that shift is documented. Work associated with MIT and IATA estimated revenue gains of up to 1 per cent for class-based continuous pricing and up to 2 per cent for classless designs. Lufthansa Group and Air Canada have since reported uplifts in the 1–3 per cent range from production deployments. Those are not rounding errors on a network carrier’s passenger revenue. They are also the reason the old ladder is being dismantled on purpose. A discrete booking-class structure leaves willingness-to-pay on the table. Once a material set of competitors prices continuously, a periodic snapshot of filed fares describes a market that the competitors have already left.
Cache was the first engineering response, and it worked while the underlying price sat still. Reuse a filed fare, serve the look cheaply, refresh on a cycle. Continuous pricing breaks that bargain in two directions at once. The cached number goes stale faster. Personalisation and channel-specific construction make the cached number less reusable even when it is fresh. The instrument still answers “where do we sit.” It answers a narrower version of the question each year.
The lowest headline fare is now one point on a product
The second pressure on the benchmark is product, not speed. The lowest available fare was a usable comparison when the product behind it was similar across carriers: a booking class, a change rule, a bag policy that varied inside a known band. That product has fragmented. Brand families, bundles, and ancillaries now sit around the fare. Comparing “the price” across carriers without first normalising those attributes is a comparison of unlike things.
This is already familiar inside any airline that sells a branded ladder. Basic, standard, and flex are not three prices for one seat. They are three products. Competitors name the same idea differently, attach different ancillaries, and vary the mix by channel. A rank-order of headline fares therefore compresses several commercial decisions into a single number and then treats that number as the market. The number is still useful. It is no longer sufficient.
Normalisation is the unglamorous prerequisite. Before a pricing analyst can say whether a carrier is high or low on a city pair, someone has to map brand names into comparable tiers and decide which bundle elements travel with the fare. That work used to be optional commentary around the file. It is becoming the file. Wave 2 of this series takes up the full-offer problem directly. The point for today’s benchmark is narrower. Every week the industry delays that mapping, the daily ranking describes less of what the customer is actually being sold.
Monitoring the moving market is becoming a cost problem
The third pressure is computational, and it is no longer only a pricing-operations concern. Look-to-book - searches per ticket sold - has climbed from on the order of 20:1 in the early airline.com period to on the order of 20,000:1 in the modern OTA period, a thousand-fold increase in live compute attached to each booking. The next step, already visible in agentic shopping prototypes, is another order of magnitude: on the order of 200,000 searches per ticket if exploratory queries are allowed to hit a live, transactable pipeline.
The cost math is uncomfortable even before that step arrives. A further ten-fold increase in look volume between 2025 and 2030 implies growth of roughly 58 per cent a year. Silicon unit cost has declined on the order of 27 per cent a year by one long-run measure of AI-chip performance. Volume therefore outruns unit-cost relief by about two to one. Cache cannot close that gap once the price being cached is itself a shop-time construction. The airline pays for a picture that is both more expensive to take and less complete when it arrives.
IATA’s look-to-book work with airlines, IT providers, and sellers is explicit that no single vendor or carrier can resolve the volume problem in isolation. That finding belongs mostly to the third essay in this series, where the industry’s options for offloading exploratory demand are the subject. It matters here because it explains why fare intelligence is migrating from a departmental tool to a board-visible infrastructure cost. The commercial team still needs the answer. The finance and technology teams now have a reason to care how often the answer is rebuilt.
The question is unchanged. The instrument has to be.
None of this retires competitive benchmarking. It explains why the same question now requires a different instrument. Pricing still needs a current reading against a defined competitor set. Revenue management still needs to know whether a weakness is a fare problem, a product problem, or a channel problem. Network still needs a market picture that is not a week old. Distribution still needs to know whether an indirect offer is the same product the airline would have built on its own site.
Those teams have historically been able to share one file. Continuous pricing, fragmenting products, and rising look-to-book costs mean that file has to be rebuilt more often, at finer grain, and with an explicit product map rather than a headline rank. The market will not slow down to wait for the older cadence.
About 260 carriers are already pricing in ways the filing cycle does not describe. ATPCO’s 80 per cent target, whatever its eventual hit rate, is the industry telling itself the same thing.
The next essay takes up the imminent wave: NDC and Offer and Order moving first-party grain into indirect channels, competition shifting from a single fare to the assembled offer, and search itself becoming a measurable demand signal rather than exhaust from the booking. The daily question - where do we sit - survives. What sits next to the fare is about to matter as much as the fare.
