Airline acquiring is collateralised against the wrong number
Acquirers hold one to two months of card sales. The unflown tickets they may have to refund run two to six months. That gap is the real exposure.
Writing
Notes from the field
Long-form thinking on travel, payments, and building companies, written from inside the work.
Seven weeks in: the pattern behind all of it
The expensive mistakes I get called into were mostly made by good teams with the wrong specialist in the room. Three examples from the first seven pieces.
NDC does not break at booking. It breaks after
Four places where NDC gets hard once the ticket is sold, and why I think servicing rather than standards is behind the flat share number.
Travel Industry Explained: the first two volumes
Airline Distribution Explained and Travel Payments Explained are out today. Why they exist, what is in each, and the two scored diagnostics, free.
The two diagnostics from the books, free
The scored diagnostics from Airline Distribution Explained and Travel Payments Explained. Thirty items on distribution, forty on payments, as fillable PDFs.
Payment failure is worth about three times payment cost
17% abandon an ancillary when the payment fails. One added method lifts conversion 7.4%. The whole payment cost line is 2.27% of volume.
Offers and Orders is now a contract question, not a standards question
Sabre calls Amadeus dominant. IATA says adoption is voluntary with no deadline. The leverage sits in the core system contract, not in the schema.
What actually shipped in agentic travel booking in 2026
Five agentic travel announcements in 2026. Four of them sit next to the booking, and the one that books real flights is a closed alpha behind a waiting list.
Most airline payment business cases are built on 2019 data
The $20bn figure everyone quotes is 2019 data published in 2022. IATA's own 2024 numbers work out at 2.27% of payment volume.
NDC has been flat at one fifth of the US agency market for a year
Twelve months, twenty basis points: NDC's share of US agency transactions stopped growing even as the market grew 12%.