Strategy.

The framework

1. 01

   Sort the capability before you price it.

   Every capability falls into one of three buckets: differentiating — part of why customers choose you over the alternative; necessary — must be done well, but wins you no deals; commodity — must merely be done. Prahalad and Hamel's test for the first bucket is worth applying literally: does this capability open access to markets, does it contribute to what the customer actually values, and would it be hard for a competitor to imitate? Only the first bucket has a strong claim to being built and kept in-house almost regardless of cost. The sorting must happen before any pricing, because the moment a quote is on the table, the smaller number starts arguing for itself — and the buckets answer a question the numbers cannot.

2. 02

   Price the whole transaction, not the invoice.

   This is Coase's correction, and it cuts both ways. Buying costs more than the invoice: the weeks of vendor search, the negotiation, the briefing and re-briefing, the meetings where you explain your business again, the quality you accept because pushing back is exhausting, the switching cost you will meet later. Making costs more than the salary: recruitment, management attention, tools, and the opportunity cost of what that attention was not spent on. Most make-or-buy comparisons put a complete invoice next to an incomplete salary — or the reverse — and the borderline calls flip when you finish the arithmetic. Write both columns in full before you look at either total.

3. 03

   Never rent the learning curve.

   Capabilities compound, and the compounding accrues to whoever does the work. Every project a vendor runs for you teaches the vendor — about the craft, and about your business — and you buy the output while they keep the asset. So ask of each capability not just what it costs today but where it is heading: if this is where the next three years of your advantage will be built, keep it in-house even at a premium, because you are not buying output, you are buying the curve. The same logic runs in reverse for commodity work done internally out of habit or control-comfort: every hour your team spends doing what a vendor does better is an hour subtracted from the curve that matters.

4. 04

   Decide the dependency before you sign it.

   Every buy creates a dependency, and Williamson's work is a catalogue of how dependencies turn expensive: once a vendor holds specific knowledge of your business and switching is painful, the renewal negotiation is not between equals — they know you are locked in, and the price will learn it too. So before any significant buy, answer three questions: how many real alternatives exist, what would switching actually cost, and what happens at renewal when the vendor knows the answer to the first two. Reversible buys can be made quickly and corrected cheaply; irreversible ones deserve the full analysis, a second source, or a written exit. Never single-source anything existential without knowing exactly how you would leave.