Putting the Six to Work: Where to Start
The six capabilities sound like a year-long programme. They don''t have to be. This capstone shows McKinsey''s own starting move: build a thin baseline of all six for one value domain, learn, then repeat — rather than perfecting any single capability across the whole company.
Six capabilities sound like a five-year programme. McKinsey's own first move is the opposite of that — and it's the reason their transformations actually start.
Here's the trap the capabilities set if you read them as a to-do list: you decide to “get the talent right” first, spend a year on it, then start on the operating model, then the data — and three years in you have strong capabilities, no value captured, and a board that's lost patience. Sequencing the capabilities one after another is the slow road to nothing. So is the other extreme: rolling out all of them across the entire company at once, which collapses under its own weight. McKinsey's answer is a different cut entirely. Don't slice by capability and don't slice by the whole org — slice by domain.
Perfect one, company-wide
One capability polished across the whole org — a strong row, and nothing shipped. You learn about that capability; you capture no value.
Thin slice of all six, one domain
A complete slice — all the capabilities, thin, aimed at one domain. It captures value now, and hands you a template to thicken and repeat.
Pick one value domain from your roadmap, and stand up a thin version of all the capabilities aimed at just that domain: a small roadmap for it, a pod with enough in-house talent to staff it, just enough shared technology and data to ship, a real adoption push — and, now, agentic workflows where they earn their place and the governance to run them safely. Thin, but complete. You're not perfecting anything; you're getting one full slice of the system working end to end, because a transformation is a system and you only learn how the parts mesh by running them together.
Which domain first: a value-by-readiness call
Choosing the first slice isn't guesswork. Weigh each candidate domain on two axes — the value at stake, and how ready you are to capture it — and start where both are high:
Where the nine lessons land
Step back and the whole course is one argument. The value-led roadmap tells you which domain to start with. Talent and the pod model give you a team that can own it. The technology and data foundation lets that team move without rebuilding the basics. Gen AI and agentic workflows are how the work itself gets redesigned, not just assisted. Adoption turns the slice into captured value, and governance is what lets you keep scaling without a stall. And this capstone is the rhythm that strings them together: not a row of sequential projects, but one thin slice repeated and thickened, domain after domain, until the capabilities are real across the company because you built them by using them. The 2026 reset, in three moves: from scattered initiatives to strategic programmes, from use cases to whole processes, from experimentation to industrialised delivery.
Try this
Pick the single highest-value domain you could realistically ship in a quarter. Now sketch the thinnest possible version of all the capabilities pointed at just it — one line each, governance and agentic workflow included. If even the thin version feels too big, your domain is too big; cut it down until a thin slice of all of them fits in the time you have. That slice, shipped and adopted, teaches you more than any plan — and it's how you actually start.
Grounded in Lamarre, Smaje & Zemmel, Rewired (McKinsey, 2023; 2nd ed. 2025), on starting with a thin baseline across one domain and scaling by repetition.