Skip to content

Resources

Build vs. buy, argued fairly

We sell the buy side, so treat this page with appropriate suspicion — and then read the section on when building is genuinely right, which is the part most vendors leave out.

Should I build travel technology or buy a platform?

Build when distribution itself is your product, when you have exactly one supplier and always will, when your volume makes per-booking fees dominate, or when your product shape is one no platform models. Buy when the technology is a cost of doing business rather than the thing customers choose you for. Both answers are legitimate.

AreaBuilding in-houseOn infrastructure
First supplier integrationSix to twelve weeks for the first, including learning the supplier's rate model, content shape and cancellation grammar.Configuration plus a mapping run. The supplier's peculiarities are already absorbed.
Each additional supplierRarely faster than the first by as much as teams expect. Every supplier is different in a new way.Connector configuration and a mapping pass.
Hotel mappingA product in its own right. Companies exist that do only this. It is never finished, because supply changes.Included, with a human review queue for the uncertain band.
Booking correctnessIdempotency, timeout reconciliation and refund arithmetic are usually learned by getting them wrong in production.Already encoded, including the unknown-outcome path.
Pricing changesOften a code change and a release, because margin logic starts life in the storefront.A rule change in the admin, versioned and effective-dated.
Team requiredDistribution engineers, a data engineer for mapping, and someone who owns supplier relationships technically.Product and front-end engineering on the layer that differentiates you.
What you own at the endEverything, including the maintenance burden and the parts that will never differentiate you.Your brand, your customers, your supplier contracts and your data. Not the plumbing.
Switching cost laterNone — it is yours. This is a genuine advantage and should not be dismissed.Real. Mitigated by documented data export and by keeping supplier contracts in your own name, but real.
Stated plainly:The timings above come from how travel distribution projects generally go, not from a study we ran. Treat them as a shape to sanity-check your own estimate against, and be suspicious of any vendor — including us — who gives you a precise number for your specific case before understanding your suppliers.

The other side

When you should build it yourself

There are real cases where buying infrastructure is the wrong call, and a vendor who will not name them is not being straight with you.

Distribution is your product

If your differentiation is the connectivity itself — you are building a bedbank or an aggregator — then buying the layer you intend to sell makes no sense.

You have one supplier and will always have one

A single direct contract with no ambition to add more does not need an aggregation layer. Integrate it directly and keep your money.

Your volume makes the per-booking cost dominant

Above a certain scale, infrastructure fees exceed the cost of a team. If you already know your volume is there, model both.

You need something no platform models

An unusual product shape that no infrastructure supports will be a fight against the platform forever. Better to build it than to bend someone else's model.