Ask ten product teams for their strategy and most will hand you a roadmap, a list of features with quarters next to them. That's a plan, not a strategy, and the gap between the two is where a lot of good products quietly stall. A roadmap tells you what you're building next. It doesn't tell you why those things and not a hundred others, or what winning even looks like. That "why" is the work of product strategy and vision, and they are two different jobs.

The quick version

  • Vision is the future you're trying to create, inspiring, customer-centred, and stable for years. Strategy is how you'll get there, a small set of hard choices about where to play and how to win, revisited often.
  • Marty Cagan frames vision as a 3–5 year north star and strategy as the sequence of product-market-fit "beachheads" you'll take on the way to it.
  • A real strategy says no. Roger Martin: the heart of strategy is choosing where to play and how to win, which means deliberately not playing everywhere.
  • Richard Rumelt's test: good strategy starts with an honest diagnosis of the core problem, then a guiding policy and coherent actions. A list of goals isn't strategy.

The idea in depth

The cleanest way to separate the two terms comes from Marty Cagan of the Silicon Valley Product Group, whose book Inspired is close to a standard text for the discipline. In Cagan's framing, the product vision describes the future you want to create, typically a 3–5 year horizon, and its job is to inspire: to make the team, and the people who fund them, want to help build it. Product strategy is the sequence of product-market-fit targets you plan to hit, like beachheads, on the path to realising that vision (SVPG, "Vision vs. Strategy"). Vision changes rarely. Strategy you should expect to revisit, Cagan argues at least quarterly, because the market keeps moving under you.

So write them as two separate artefacts and never let one masquerade as the other. A quick gut-check: a vision should still read true in three years; a strategy should make at least one person in the room slightly uncomfortable, because it rules something out.

That discomfort is the point, and it's where Roger Martin sharpens things. In Playing to Win (co-authored with former P&G chief A.G. Lafley, 2013), Martin reduces strategy to an integrated cascade of five choices: a winning aspiration, where to play, how to win, the capabilities you'll need, and the management systems to support them. The middle pair carry the weight. Martin calls where-to-play and how-to-win "the heart of strategy" and says you can't have a great strategy without a great heart (Martin, "Decoding the Strategy Choice Cascade"). The reason this matters for products: choosing the wrong field means you can't win no matter how good your team is. Strategy is as much about the segments, channels and use-cases you decline as the ones you chase.

flowchart TD
    V(["Product vision<br/>(3–5 yr north star)"]) --> S(["Product strategy<br/>where to play / how to win"])
    S --> B1(["Beachhead 1<br/>first product-market fit"])
    S --> B2(["Beachhead 2<br/>adjacent segment"])
    S --> B3(["Beachhead 3<br/>…toward the vision"])
    B1 --> R(["Roadmap<br/>features & sequencing"])
    B2 --> R
    B3 --> R
Vision sets the destination; strategy picks the beachheads; only then does a roadmap make sense. Leaders Loop

The practical step is to force the choice explicitly. Before the roadmap, answer two questions on one page: for this horizon, which customers and use-cases are we fighting for, and how will we win them that competitors can't easily copy? If you can't name what you're not doing, you haven't finished. A roadmap that follows from a clear where/how is a strategy in motion; a roadmap that precedes it is just a to-do list with ambition.

What separates a real strategy from a wish list

Plenty of documents called "strategy" are really a stack of goals, grow revenue 30%, delight customers, be the market leader. Richard Rumelt, the UCLA strategy scholar, named this directly in Good Strategy / Bad Strategy (2011): the second most common flavour of bad strategy is mistaking goals for strategy, and the most common is a weak or absent diagnosis of the actual problem. Rumelt's "kernel" of a good strategy has three parts, a diagnosis that names the crux, a guiding policy for dealing with it, and a set of coherent actions that carry the policy out (Rumelt, summarised on Lenny's Newsletter).

"The most common cause of bad strategy is a weak diagnosis.", Richard Rumelt

Start your strategy with the problem, then, not the ambition. One honest paragraph: what is the one thing standing between us and the vision right now? Then a guiding policy that addresses precisely that, then two or three actions that reinforce each other. If your actions don't obviously serve a single guiding idea, you have a portfolio of activity, not a strategy, and the team will feel it as whiplash.

This connects back to the daily grind of roadmapping and prioritisation: when prioritisation arguments turn into a tug-of-war between loud stakeholders, the real fault is usually upstream. There's no agreed diagnosis to prioritise against. Fix the strategy and a lot of prioritisation fights dissolve, because now there's a shared answer to "does this move us toward the next beachhead?"

An honest limitation: strategy on paper isn't strategy in practice

It's worth being candid about what these frameworks don't do. They are organising lenses, not engines of insight. A choice cascade or a Rumelt kernel will structure your thinking; it won't tell you which segment is actually winnable or which capability is actually rare, that knowledge comes from discovery and validation, from talking to customers, and from evidence about product-market fit. A beautifully filled-in template built on guesses is just a confident guess.

There's a second trap. A written strategy means little without the conditions to execute it, and the discipline to hold the line on focus when the next shiny request lands. Plenty of organisations write a sound strategy and then undermine it weekly by saying yes to everything. The framework isn't the bottleneck; honouring a "no" usually is. So treat any strategy document as a hypothesis you'll defend under pressure, not a settled fact you can file away.

A worked example

Consider a fictional company, call it RotaPay, a small startup building payroll software. (The figures below are illustrative, used to show the mechanics, not real data.) The founders' instinct is a sweeping vision and a roadmap to match: "payroll for everyone," with features queued up for enterprises, mid-market firms, accountants and sole traders all at once. Six months in, they're shipping steadily and winning nobody decisively.

Re-run it through the three lenses. Vision (Cagan): a stable 3–5 year north star, "every small business runs payroll in ten minutes a month, without a bookkeeper." Inspiring, customer-centred, unlikely to change. Diagnosis (Rumelt): the crux isn't features, it's focus, they're spread across four buyer types with four different needs and beating none. Strategy (Martin's heart): where to play, micro-businesses with one to five staff, who can't afford an accountant and are underserved by enterprise suites. How to win, radical simplicity and a setup that takes minutes, not the feature-completeness larger rivals compete on.

flowchart LR
    D(["Diagnosis:<br/>spread thin across<br/>4 buyer types"]) --> G(["Guiding policy:<br/>win micro-business<br/>via radical simplicity"])
    G --> A1(["Action: 10-min<br/>guided setup"])
    G --> A2(["Action: drop<br/>enterprise backlog"])
    G --> A3(["Action: pricing<br/>for <5 staff"])
RotaPay's strategy kernel: one diagnosis, one guiding policy, coherent actions that reinforce it. Illustrative. Leaders Loop

Now the roadmap writes itself, and so does the list of things to stop. The enterprise payroll backlog gets shelved, not because it's bad work, but because it serves a field RotaPay has chosen not to play on yet. The team's first beachhead is plumbers, sparkies and cafés with a handful of staff; later beachheads (small accountancy practices, then mid-market) become the path to the broader vision. The strategy hasn't shrunk the ambition. It's made the ambition reachable by sequencing it, exactly as Cagan's beachhead model suggests. The uncomfortable choice, saying no to enterprise for now, is the strategy.

Frequently asked questions

What's the difference between product vision and product strategy in one line?

Vision is the destination, the future you want to create, usually 3–5 years out, and it rarely changes. Strategy is the route, the specific, revisable choices about which customers and use-cases you'll win first, and how. Vision inspires; strategy decides.

How is product strategy different from a roadmap?

A roadmap is the output of a strategy: the sequenced features and bets that follow once you've chosen where to play and how to win. If you have a roadmap but can't explain the choices behind it, you have a plan without a strategy, which is why feature lists so often feel arbitrary.

How often should we change our product strategy?

More often than the vision, less often than the backlog. Cagan suggests revisiting strategy roughly quarterly, because evidence from discovery and the market keeps shifting. The vision should hold for years; if it's changing every quarter, it was probably a plan all along.

Do small teams and startups really need a formal strategy?

They need the choices, not the ceremony. A startup can't out-resource incumbents, so where-to-play and how-to-win matter more, not less, focus is the only edge a small team reliably has. One page (diagnosis, where, how, what we're not doing) beats a fifty-slide deck nobody revisits.

What's the most common product-strategy mistake?

Confusing goals with strategy, Rumelt's point. "Grow 30%, win the market, delight customers" are ambitions, not a plan for achieving them. The fix is to start with an honest diagnosis of the one obstacle in your way, then choose actions that all serve a single guiding policy.

Related in the Toolkit

Where to go next