Most people asking should buy. Here is how to tell.

The three conditions

Building wins when at least one of these is clearly true. The process is the differentiator; the integration cost between bought products exceeds the build cost; or the data shape does not exist in the market.

The condition that is usually a mirage is "nothing does exactly what we want." It is almost always true and almost never sufficient — the relevant question is whether the gap costs more than owning the whole system would. The write-up on buy software covers the four questions that predict whether you will still be happy with a purchase in five years, none of which appear on a scorecard.

On the other side, the write-up on software build sets out the five obligations building creates — patching, platform drift, bus factor, support, and decision debt — none of which appear in an estimate because none of them is a feature.

The answer is usually neither

Framed as a binary, this decision almost never resolves as one. What companies do is buy the commodity and build the differentiator, and the quality of the outcome depends on where that line is drawn. The write-up on building custom covers the test I use: if a competitor had this capability, identical, would it hurt us.

The reason people arrive at this question in the first place is usually a product that fights how they work — the write-up on buy shelf covers the three kinds of mismatch and which of them configuration can fix.

Related engagements: fractional cto services for ongoing decisions of this kind, and the outsourced cto services engagement where the whole technical function needs an owner.

What you can inspect

The egress policy agent demo is a system of mine that declines to produce output the evidence does not support — on a real capture it enforces 0 of 19 destinations and explains why. It is the same disposition this analysis is built on: a recommendation you can check, with the conditions for reversing it written down.

What this cannot settle

The three conditions are a filter, not a decision. They reliably tell you when to buy. When they indicate building is plausible, someone still needs to look at your actual data, your actual integrations and your actual team before money is committed.

I also cannot tell you whether a specific platform's API is expressive enough for what you have in mind without reading its documentation against your requirements. That check is cheap and it is the one most often skipped before a platform is chosen.