Build, buy or integrate: deciding without a supplier in the room

Build or buy is rarely a single decision. A travel business is a collection of capabilities, and the right answer differs for each one. Asked at the level of "our platform", the question has no good answer. Asked at the level of "quote versioning", it usually has an obvious one.

Start with what is genuinely differentiating

A useful first filter: would a customer or a partner ever choose you because of this capability?

For most travel businesses, the itinerary construction workflow, the supplier relationships and the commercial model are differentiating. Accounting, messaging infrastructure, document generation and identity are not. Building the second group is a cost with no commercial return, however satisfying it is to control.

The trap is that the non-differentiating capabilities are frequently the easiest and most enjoyable to build, which is why so many businesses have a bespoke one.

Then ask how unusual your process really is

Every business believes its process is unusual. Most are unusual in one or two specific places and entirely conventional everywhere else.

Being precise about which parts are genuinely unusual is what makes buying viable. A product that covers eighty per cent with a configured or integrated remainder is nearly always cheaper than building all of it — and the remaining twenty per cent is where your differentiation actually lives.

Precision about what is unusual is what makes buying possible. Vagueness is what makes everything look bespoke.

Cost the whole life, not the first version

• Build: initial development, plus maintenance, plus the opportunity cost of the team not doing something else, plus the risk concentrated in whoever understands it.
• Buy: licence, plus implementation, plus configuration, plus the cost of adapting your process, plus the risk of the vendor changing direction or pricing.
• Integrate: integration build, plus ongoing maintenance as both sides change, plus the coordination cost of a dependency you do not control.

The comparison people usually make is build cost against licence cost. That comparison flatters building, because it omits the largest line on the build side and the smallest on the buy side.

Weigh the cost of being wrong

A decision that is expensive to reverse deserves more analysis than one that is not, and this is frequently the deciding factor rather than the economics.

Buying is usually easier to reverse than building, provided your data can leave. That proviso is worth establishing in writing before purchase rather than discovering during an exit — an exportable data model is a genuine commercial term, not a technical detail.

The integrate option is systematically underrated

The third option gets less attention than it deserves. Many capabilities can be obtained by connecting something you already own to something you already pay for.

It is unglamorous, it is fast, and it produces no impressive announcement. It is also, in a meaningful share of cases, the correct answer — and the one least likely to be recommended by anyone whose revenue depends on a larger project.

A short decision test

1. Would a customer ever choose us because of this capability? If no, do not build it.
2. Is our process here genuinely unusual, specifically how? If not, buy it.
3. Could this be obtained by connecting two things we already have? If yes, do that first.
4. If we are wrong in two years, what does reversing cost? Weight the analysis accordingly.
5. Who maintains this in three years, and is that person in the plan?

A general decision framework. Appnox builds software and also publishes products, so we have an interest on more than one side of this question — which is exactly why the test above is written to be used without us.