Services · APIs
Flight search API and fare pricing for high-volume shopping
Most of the load on any air product is people looking, not buying. A flight search API has to answer quickly, show prices that hold up at checkout, and do it without breaching supplier look-to-book limits. This page covers how Voraal's Avunzo handles shopping and fare quotes, and what your team should design around.
Built for
- Metasearch-style front ends
- OTAs
- Fare alert products
- Corporate tools
- Mobile travel apps
In brief
What does a flight search API do?
A flight search API takes an origin, destination, dates and passengers and returns priced itineraries from airline sources. Voraal's Avunzo queries Amadeus, Sabre and Travelport, NDC airlines and low-cost carriers in parallel, merges the results into one schema, and uses caching tuned for search-heavy traffic. A separate price call re-checks the chosen offer against live availability before the customer pays.
By the Voraal team · Updated
Shopping and booking are different workloads
A flight search API serves many requests, most of which never become bookings. That ratio is called look-to-book, and suppliers watch it: every search a GDS or airline answers costs them compute, so heavy shoppers can face limits or charges. The design goal is to answer the browsing customer fast, spend live supplier calls where they change a decision, and confirm price only when the customer is close to paying.
Avunzo separates the two. The search call fans out to connected sources in parallel and returns a normalised result set; the price call re-checks one selected offer against live availability. Booking, ticketing and servicing then belong to the flight booking API.
What to cache, and what never to trust from cache
Fares are volatile because seats are sold in booking classes that open and close. Different data ages at different speeds.
| Data | How fast it changes | Sensible handling |
|---|---|---|
| Airports, airlines, aircraft, city codes | Rarely | Cache for long periods; refresh from reference data |
| Schedules | Seasonally, with ad hoc changes | Cache, but never rely on it for a booking |
| Popular route-date results | Within minutes on busy routes | Short-lived cache to absorb repeat searches |
| Seat availability in a booking class | Continuously | Live; confirm with the price call |
| Final price with taxes and fees | With availability and currency | Live at price-check; show any change to the customer |
Avunzo's caching and aggregation are tuned for responsive search; the price call is the point of truth before payment.
How a search request is answered
- 01
Normalise the request
Origin and destination (airport or city), dates, cabin, passenger types and any filters such as non-stop only or preferred carriers are validated once, in one format.
- 02
Fan out in parallel
Voraal's core calls each relevant source at the same time rather than one after another, so the slowest supplier sets the response time, not the sum of all of them.
- 03
Merge and de-duplicate
The same flight can come back from more than one channel. Results are merged into one schema so your front end can sort and compare like with like.
- 04
Apply channel rules
Markups and filters set per channel are applied before results reach your app, so a wallet and an OTA on the same account can see different prices.
- 05
Price the selection
When a customer picks an offer, call price. A changed fare or lost seat is reported before passenger details are collected.
Key features for shopping traffic
One query, every channel
GDS, NDC and low-cost results in one response, sorted and priced the same way.
Parallel calls and caching
Aggregation and caching designed for search-heavy traffic.
Fare families
Branded fares and ancillaries returned where NDC airlines provide them, so customers compare what is included, not only the headline price.
Per-channel markups and filters
Different price and content rules for each of your channels or partners.
Usage reporting
Search and booking volume per key, so you can see your own look-to-book before a supplier does.
Clear error codes
Distinguish 'no availability' from 'supplier timeout' from 'bad request' in code.
Business applications
Fare alert and price-tracking products rely on frequent, cheap searches and occasional live checks. Voraal's own technology examples include a traveller saving a route and receiving an alert when the fare moves. Corporate tools filter results by company policy before employees see them. Mobile apps need small, predictable payloads; a normalised schema means one parser for every airline channel. Offers from airlines selling through their own NDC channel are covered on the NDC integration page. For the difference between calling a GDS directly and using an aggregated API, read flight booking API vs GDS.
Implementation process
- 01
Model your traffic
Estimate searches per day, peak hours and expected conversion. This shapes rate limits per key and caching.
- 02
Build against the sandbox
Use sandbox keys and test data to build search, filters, sorting and the price-check step.
- 03
Design the UI for price change
Decide what the customer sees when a re-price returns a different total. Silent failure here loses more bookings than slow search.
- 04
Load-test and go live
Test at expected peak in the sandbox where permitted, then move to production credentials and monitor response times and error codes.
Common challenges
Look-to-book pressure
Searching on every keystroke or date-slider move multiplies supplier calls. Debounce input and reuse recent results.
Phantom low fares
A cached fare that has already sold out frustrates users. Show cached prices as 'from' and confirm with the price call.
Currency drift
Fares filed in one currency and displayed in another move with exchange rates; display and settlement currency must be explicit.
Slow sources
One slow supplier should not block the page. Set a response budget and stream or return partial results where your UX allows.
How Voraal delivers this
The product behind it.
FAQ
Frequently asked questions
Questions buyers ask about flight search and pricing api.
Ask us anythingWhy does the price change between search and checkout?
Search results reflect availability at the moment of the query, and airlines sell seats in booking classes that close as they fill. The price call re-checks the chosen offer live, and the customer should see the confirmed total before paying.
What is look-to-book and why does it matter for a search API?
It is the ratio of searches to bookings. GDS providers and airlines incur cost for every search they answer, so very high ratios can lead to limits or fees. Caching, debouncing user input and reporting usage per key keep it in check.
Can we cache Avunzo search results on our side?
Short-lived caching of results for display is common, but never book from a cached price; always call price first. Agree your caching approach with Voraal during discovery so it stays within supplier terms.
Does the search API return branded fares?
Where NDC airlines provide fare families and ancillaries in their offers, Avunzo returns them so you can show what each fare includes. Availability of branded fares varies by airline and channel.
How quickly does a multi-source search respond?
Sources are called in parallel, so response time is driven by the slowest source within the response budget. Voraal does not publish fixed latency figures; measure in the sandbox with your own routes.
Let’s build what comes next.
Tell us how your travel business works. We’ll recommend the right products and walk you through them on your own use case.