Make-or-Buy.

A usable decision emerges along a handful of criteria. First, differentiation: does the business earn its money with this workflow, or is it merely necessary support? Second, market supply: are there products that cover eighty per cent of the need? Third, the fit of the remaining twenty per cent: are those habits or genuine requirements? Fourth, data sovereignty and regulatory obligations. Fifth, time pressure. And sixth, the internal ability to run a system of your own for years — which is, in our experience, the point most often overestimated.

The comparison has to run across the lifetime of the solution, not the purchase price. On the buy side sit licences, rollout, training, adjustments and an eventual migration. On the build side sit development, operations, maintenance, security updates, further development and the risk that knowledge sits with individual people. Internal hours belong on both sides of the calculation, because they are real and frequently tip the result. Comparing licence costs against a development quote means comparing two different things and deciding, in effect, on instinct.

A neutrally calculated case: a business with fifteen office staff considers building its own CRM because the licence costs grate. The five-year calculation puts licences, rollout and training for the standard product against the cost of a custom build — including maintenance, operations and further development, because a first version alone is never enough. At that scale the standard product almost always wins clearly, because the ongoing responsibility weighs more than the licence total. It looks different when one genuinely distinctive workflow sits at the centre.

There is a third route that often gets overlooked: buy the standard product and build only the differentiating piece yourself, connected cleanly through interfaces. That keeps the in-house share small and leaves the rest to the vendor to maintain. Either way, document the decision together with its assumptions and revisit it after twelve to twenty-four months. Market supply, user numbers and requirements all shift — a decision that was right in 2026 is not automatically still right in 2029.