Product strategy and positioning

Most product teams drown in roadmaps and feature lists while calling the pile a strategy. A strategy is a focused bet on where you will play and how you will win — and an explicit list of what you will ignore. Without it, every request looks urgent, every competitor feature looks mandatory and prioritization becomes politics.

With strategy, engineers can push back on work that does not move the diagnosis, and PMs can say no without sounding arbitrary. The pages below split the layers (vision vs strategy vs roadmap), Rumelt's test for a real strategy, Dunford's positioning method and the one-pager you can actually execute.

Concepts in this section

Read them in order. Each one builds on the one before.

Step	Concept	What it answers
1	Vision vs strategy vs roadmap	Which layer is this sentence on, and what question does it answer?
2	Rumelt's kernel	Do we have a diagnosis, a guiding policy and coherent actions?
3	Product positioning	What alternative do buyers already use, and how do we frame our value?
4	The strategy one-pager	Can we say where we play, how we win and what we refuse on one page?

Couple positioning to discovery: talk to customers about alternatives and value, not about your roadmap. Couple strategy to roadmaps and OKRs: outcomes should flow from the guiding policy. Re-read the one-pager when prioritizing — if RICE ranks something that violates non-goals, either the score is wrong or you are about to abandon strategy for a quarter without admitting it.

You rarely own company strategy as an SDE2, but you feel its absence every sprint. Translate technical constraints into options ("SSO in two weeks for Okta-only, or six for generic SAML — which segment?"). Flag work that fights the guiding policy. Propose cheaper experiments that test the diagnosis before a six-month build.

In interviews, "product sense" questions often hide a strategy question: who is the user, what is the alternative, why would they switch, what would you not build. Answer with diagnosis and trade-offs, not a feature laundry list.

Common mistakes

Strategy as consensus. If every stakeholder got their initiative on the page, you have a portfolio, not a strategy.
Strategy as slogans. "AI-first" without a diagnosis of which workflow AI changes is fluff.
Feature-matrix paralysis. Comparing 40 features across five vendors produces parity races you will lose to the incumbent.
Platforms without a job. Shipping "an API" or chasing network effects before single-player value works is not a moat; see competition and moats.

Key takeaways

Vision is destination; strategy is focused choice under constraint; roadmap and tactics execute.
Use Rumelt's kernel: diagnosis, guiding policy, coherent actions.
Position from alternatives → unique attributes → value → best-fit customer → market category.
One page beats twenty: say where you play, how you win and what you refuse.