Every business repeatedly faces the choice between constructing a capability internally and purchasing it. The decision is usually framed as a cost comparison and usually decided by something else.
Differentiation is the first filter
If a capability is what makes customers choose the firm, building it keeps control of the thing competitors would need to copy. If it is table stakes, ownership adds cost without advantage.
Most functions fall into the second category. Payroll processing, accounting systems and email infrastructure are necessary and identical across competitors.
The error is treating something as differentiating because it is important. Importance and distinctiveness are different tests, and only the second argues for building.
Internal cost is routinely understated
Build estimates typically cover development. They rarely include maintenance, support, documentation, security patching, staff turnover and the opportunity cost of the engineers involved.
Ongoing cost dominates over a multi-year horizon. Software built once continues to consume attention for as long as the business depends on it.
Vendor pricing appears higher precisely because it is complete. Comparing a full subscription price against a partial internal estimate reliably favors building.
Buying transfers risk and creates dependency
A vendor carries responsibility for uptime, compliance updates and feature development, backed by contractual commitments the firm can enforce.
In exchange the firm accepts exposure to price increases, changes in product direction, acquisition of the vendor and eventual discontinuation of the product.
Switching cost is the measure of that exposure. Data portability, integration depth and staff familiarity all determine how expensive an eventual exit becomes.
Time to capability often decides it
A purchased system can be operating in weeks. An equivalent internal build takes long enough that the market condition prompting it may have changed.
Where the capability supports a time-limited opportunity, buying is frequently chosen despite a worse long-run cost, and that trade should be stated rather than hidden.
Hybrid approaches follow from this. Firms commonly buy to establish the capability and build later once requirements are proven and the volume justifies it.
Integration is the cost that surprises
Purchased systems must exchange data with everything already in place. That connective work is built internally regardless of the decision and is frequently omitted from the comparison.
Configuration is similar. Substantial vendor products require specialist configuration, and the internal or consulting effort involved can approach the license cost.
Decisions revisited on a schedule age better than decisions made once. Requirements, vendor markets and internal capability all move, and a build justified three years ago may no longer be.