NDC and ATPCO are usually set up as a contest, which is the wrong shape for the question. They are two answers to the same problem: how does a seller that is not the airline find out what the airline sells, on this itinerary, today?
In short
ATPCO carries content the airline has filed in advance, so any seller reading it derives the same answer. NDC carries an offer the airline generates in response to a specific request, priced at that moment for that itinerary. Filed content gives breadth and repeatability. Generated offers give accuracy and context.
Filed content: the airline writes it down first
Filing is the older idea and still the wider one. An airline records what it sells and under what conditions: fares, the rules attached to them, and structured records describing optional services such as a checked bag or an assigned seat. That content is published on a schedule, and every seller in the market reads the same records.
The consequences all follow from one property. Filed content is deterministic. Two sellers reading the same records for the same itinerary should reach the same result, which makes the answer auditable and usable at the scale of a display that prices thousands of itineraries without asking anyone.
Its limits come from the same property. A filing describes rules, and a rule has to be evaluated against a real trip before it becomes a price. Conditions stack up across route, fare basis, direction, point of sale and traveller status, and two implementations of the same rule set can diverge in the corners. Filed content also describes what the airline sells in general. It cannot describe what the airline would like to offer this particular traveller this afternoon.
Generated offers: the airline answers the question itself
The newer model moves the decision back to the airline. The seller sends the shopping context and receives an offer the airline has constructed for it: items, prices, conditions, and an identity for that offer. Nothing is re-derived by the seller, because nothing needs to be. The airline was the pricing authority all along, and now it is answering directly.
That closes the interpretation gap entirely, and it opens a door that filing never could. An airline that builds an offer at request time can vary it by context: the itinerary, the channel, the traveller, the moment. This is what the industry means by retailing rather than distribution.
The cost is that the answer exists only inside a conversation. An offer is specific to the request that produced it and valid for a period rather than forever, so it cannot be treated as a fact about the world. Coverage is also uneven in a way filing is not. It depends on each airline having a retailing programme, and on how far that programme has actually got, which is why reach here is honestly described airline by airline rather than as one number.
The difference in one table
| What differs | Filed content | Generated offer |
|---|---|---|
| Who sets the price | The airline, in advance, expressed as rules | The airline, at the moment of the request |
| Who interprets it | The seller, against the itinerary | Nobody. It arrives already priced |
| Breadth across airlines | Wide, including carriers with no retailing programme | Airline by airline, and capability by capability |
| Repeatability | The same inputs give the same answer | The answer belongs to one request |
| Traveller context | None. The rules are the same for everyone | The airline may use it if it chooses to |
| Range of items | Whatever has been filed | Whatever the airline chooses to construct |
What this changes about what an intermediary can sell
Two things, and they pull in different directions.
The first is range. Filed content is the widest baseline anyone has. A seller reading only generated offers has a catalogue that stops at the edge of the airlines with mature programmes, and that edge moves slowly. A seller reading only filed content misses items an airline never files and misses prices the airline would rather set itself.
The second is accuracy, where the failure modes are genuinely different. Filed content goes wrong quietly: an edge case in a rule, or a record read after the airline changed it. A generated offer goes wrong loudly: no answer, or an answer that has aged out. Quiet failures are worse for a traveller, because a price shown confidently and wrongly is discovered at the airport rather than on the screen.
Why the industry is in a transition, not a switchover
Underneath both standards the commercial shape is identical, which is exactly why one can substitute for the other and why neither has to win.
The airline decides what it sells
A checked bag, an assigned seat, and the conditions attached to each.
That decision reaches sellers
Either as content filed ahead of time, or as an offer built in reply to a request. Two routes, one intent.
A seller matches it to a real trip
The itinerary, the fare and the market decide which items apply, and at what price.
The traveller decides
One clear price, at a moment the question is live for them.
The reservation reflects it
The outcome has to reach the booking, or none of the work above it mattered.
A switchover would require every airline and every seller to move at the same time, and no distribution change in this industry has ever done that. Airline programmes mature at different rates and in different markets. Filing is not being retired; for a large share of carriers it remains the only way they publish anything. Sellers therefore carry both paths, and will for years.
The practical questions worth asking anyone who says they have solved this are narrow and answerable. Which airlines, and which capabilities for each of them? What happens when the airline does not answer? Is the filed baseline still read underneath? And when the two disagree, which one is shown to the traveller?
Frequently asked questions
Is NDC replacing ATPCO?
Not in the sense of one being switched off. Filed content remains how a large part of the industry publishes what it sells, including airlines with no retailing programme at all. Most sellers will read both for a long time yet.
Which gives more accurate ancillary prices?
An offer generated by the airline is authoritative by construction, because the airline priced it for that request. Filed content has to be interpreted against the itinerary, which is where two sellers can reach two answers from the same records.
Why does ancillary coverage differ so much between airlines?
Because coverage is per airline and per capability rather than a single industry level. One airline may publish rich seat and bag content, another only the basics, and the same airline may differ by market and by route.
Does an intermediary need both?
In practice yes. Filed content gives the widest baseline across carriers, generated offers give depth and accuracy where an airline has invested. The interesting design question is which one wins when the two disagree.