From vision to advantages – why one vision needs several strategies, not one

The previous part covered Rumelt's kernel: diagnosis, guiding policy, coherent action — the logical structure that separates a real strategy from a wish list with a fancy name. But most organizations don't start with a diagnosis. They start with a vision: a picture of where they want to end up.

The problem is that a vision, however inspiring, doesn't by itself say how you get there. "We will become the obvious choice in our industry" is a vision. It is not a strategy — it lacks everything Rumelt requires: no diagnosis, no guiding policy, no coherent action. This piece takes on how to build the bridge between the two.

One vision, several advantages

Once a vision is reasonably concrete, it can be broken down into a small number of competitive advantages — the specific things that must be true of the organization on the day the vision is achieved. Not activities. Not departments. Advantages: durable differences from competitors that make customers, employees, or the market choose you specifically.

Take the vision above — "the obvious choice in our industry." Broken down, it might result in three advantages:

Fastest delivery in the industry — customers choose us because they can't wait for anyone else.
Lowest total cost for the customer — not necessarily lowest price, but lowest cost across the whole relationship.
Strongest trust — customers feel confident we won't disappear or let them down.

This still isn't strategy. It's a refinement of the vision into something concrete enough to actually be diagnosed. And here's where it gets interesting: each advantage requires its own strategy kernel. Not one shared strategy for all three — three separate ones, because the obstacles, conditions, and actions required for fastest delivery have very little in common with what's required for strongest trust.

One kernel per advantage

For each advantage, the same three parts from the previous article apply, aimed specifically at that advantage:

Diagnosis — what must we actually overcome?
Not a general description of the market, but the specific obstacle between where the organization stands today and the named advantage. For "fastest delivery," the diagnosis might be that the warehouse structure is centralized in a way that routes 70 percent of deliveries through a detour. For "strongest trust," the diagnosis might instead be that the organization has historically communicated poorly during delays, building a culture of distrust rather than a single technical problem.

Guiding conditions — within what boundaries do we work?
This is Rumelt's "guiding policy" applied concretely: not a list of everything that's allowed, but a clear boundary that rules out alternatives. For the delivery advantage, the guiding condition might be "we build regional hubs instead of a single central warehouse, even though it increases inventory cost" — a deliberate choice that closes the door on other, cheaper but slower alternatives. Precisely because it excludes something, it becomes useful; a principle that permits everything guides nothing.

Coherent goals — what will we do, and how does it fit together?
The concrete initiatives, resources, and priorities that together realize the guiding condition. For the delivery advantage: three regional hubs built in a specific order, the route-optimization system procured before hub number two opens (not after), and warehouse staff bonus structures recalculated to reward delivery time instead of inventory cost. Each of these actions reinforces the others. Remove one, and the other two lose much of their effect.

Why separate kernels, not one shared one

It's tempting to merge all three advantages into a single strategy — one document, one plan, one presentation. Resisting that is the whole point: a diagnosis meant to explain three different types of obstacles at once becomes so general it stops pointing at anything specific — exactly what Rumelt calls fluff. A guiding condition meant to accommodate three different logics stops excluding anything at all, which makes it useless as guidance.

Three separate kernels doesn't mean three unrelated projects. They still connect through the vision they came from, and the organization still needs to prioritize between them when resources are limited — which advantage matters most to reach first, and which can wait. But the analytical work itself, diagnosing what's actually in the way, has to happen separately for each advantage. Otherwise the exact mistake article one warned about repeats: a vision assumed to be its own strategy, just in a marginally more detailed disguise.

A simple test

Before the next strategy document gets written, a few questions worth asking out loud in the room:

Can we name the specific competitive advantages our vision actually requires — no more than a handful?
Does each advantage have its own diagnosis, or do they share one vague description of "the market situation"?
Do our guiding conditions actually rule out any alternatives, or do they in practice permit everything?
Do the actions within each advantage reinforce each other, or are they just three parallel to-do lists?

If the answer drifts toward "we really just have one vision and a long list of activities," the work isn't done yet. The advantages may already exist, implicitly, in the leadership team's heads — but they haven't become strategy until each one has its own diagnosis, its own guiding condition, and its own coherent actions.

Want to go deeper: Richard Rumelt's Good Strategy / Bad Strategy remains the foundational text behind the whole kernel argument. A.G. Lafley and Roger Martin's Playing to Win: How Strategy Really Works offers a complementary framework specifically for how a vision breaks down into "where will we play" and "how will we win there" — a process closely related to the multi-advantage breakdown described above.