Strategy & Alignment

These four words are routinely confused. Vision is the long-arc future state you are creating ('every developer has access to programmable money'). Strategy is the choices about how to get there given current constraints ('focus on internet-native businesses; lead with API quality; price for volume; defend with developer experience'). Roadmap is the sequenced expression of that strategy in shippable bets. OKRs are the in-period commitments that show whether the bets are working. If your team can't fit each on a different page, you are conflating them.

Richard Rumelt's three-part kernel: a diagnosis (what is the actual problem?), a guiding policy (what is our approach?), and coherent action (what specifically will we do?). Most 'strategies' are list-of-goals dressed up; they skip the diagnosis (which is the hardest part) and jump to the actions. The diagnosis is what reveals which actions are coherent and which are random.

• Diagnosis: a frame that simplifies complexity and identifies the critical aspect of the situation.
• Guiding policy: an overall approach that addresses the diagnosis and creates an advantage.
• Coherent action: the steps that the policy demands, mutually reinforcing rather than conflicting.

A strong product vision describes a future user state, names the user, and is concrete enough to disagree with. 'Be the leader in X' is not a vision; it's a press release. 'Anyone can deploy a production application from their laptop in under an hour without a DevOps team' is a vision: it names the user, the future state, and an implied current state of pain.

• Make the vision falsifiable: in 5 years, has it happened or not?
• Pair with a 3-5 year strategy and a 12-month roadmap; vision alone doesn't ship.
• Re-validate the vision when the market shifts, but resist re-writing it every quarter.
• Test it on engineers and designers — if it doesn't excite them, it won't survive Q3 fatigue.

Simon Wardley's mapping technique plots a value chain along an evolution axis (genesis → custom-built → product → commodity). It exposes which parts of your product are differentiating today and which are about to become commoditized. Strategy decisions become visible: where to invest in genesis, where to product-ize, where to outsource to commoditized utilities.

• X-axis: stage of evolution (genesis → custom → product → commodity).
• Y-axis: visibility to user (high at top, low at bottom).
• Anchor on the user need; chain dependencies downward.
• Components drift right over time — anticipate the drift, don't be surprised by it.

Porter's framework — supplier power, buyer power, threat of substitutes, threat of new entrants, competitive rivalry — was built for industries but works for product strategy if you treat the product surface as the industry. It helps you see whether the competitive structure is friendly or hostile before you commit to a position.

• Buyer power: how easy is it for users to switch to alternatives, including doing nothing?
• Supplier power: how many of your dependencies (cloud, payment, model APIs) can hike prices on you?
• Substitutes: what completely different solutions could solve the same job?
• New entrants: how high is the technical/ regulatory / data moat against the next 50 startups?
• Rivalry: are competitors converging on the same playbook, or differentiating?

A North Star Metric captures the core value exchange between user and business. Good NSMs are leading indicators of long-term retention or revenue, behavior-based, and unique to the product. Amplitude's framework distinguishes two NSM archetypes: attention products optimize time/engagement, transaction products optimize transactions completed, productivity products optimize jobs done.

• Avoid vanity (registered users, page views) — they rise without value.
• Pair the NSM with input metrics the team can directly influence.
• Re-evaluate the NSM at maturity inflections; the metric that brought you to PMF may not be the one that takes you past it.
• An NSM that no team has any leverage over is a dashboard, not a North Star.

OKRs (Andy Grove → John Doerr) express change: an Objective is a qualitative aspiration, Key Results are 3-5 measurable outcomes that prove the objective happened. KPIs are health monitors — operating metrics you watch but don't necessarily try to move. Most OKR programs fail because teams cargo-cult the format, write outputs as Key Results, set goals that are sandbagged or impossible, and never review them mid-cycle.

• Objectives must be inspirational and time-bound; Key Results must be quantifiable and binary in outcome.
• Cap at 3 Objectives, 3-5 KRs each — more than that is a wishlist.
• Calibrate confidence: at start of cycle, predict 50-70% confidence; if everything is at 100%, you're sandbagging.
• Hold a mid-cycle check; OKRs not reviewed mid-quarter become quarterly performance theater.
• Score at end-of-cycle on the 0-1 scale; 0.7 is the goal, not 1.0.

PMF is famously fuzzy. Two practical tests: Sean Ellis's 'How would you feel if you could no longer use this product?' (40%+ very disappointed indicates PMF), and the Superhuman engine, which segments very-disappointed users to find your high-expectation core and double down. Both work better than NPS for early-stage measurement.

• Pre-PMF, retention curves do not flatten — every cohort decays toward zero.
• At PMF, retention flattens (some users keep using indefinitely) and word-of-mouth becomes a meaningful acquisition channel.
• Different products have different PMF retention thresholds; consumer social ≠ B2B SaaS ≠ vertical SaaS.
• Don't optimize for PMF metrics until you have qualitative evidence of repeat use; you can drive surveys without driving value.

April Dunford's framework: positioning is the act of defining who your product is for, what category it's in, what it does that's different, and for whom that difference matters. Bad positioning is invisible — the team can't articulate it consistently. Good positioning is the lens that turns vague messaging into resonant copy and turns sales conversations into qualified pipelines.

• Identify true competitive alternatives (often spreadsheets and 'doing nothing').
• Isolate unique attributes — what only you have or do.
• Map attributes to value: 'so what?' for each.
• Identify the buyer who values that bundle.
• Choose the market category that frames your value most favorably.

Pricing is one of the highest-leverage product decisions and the one most teams under-invest in. The four levers: pricing strategy (penetration, value-based, premium), pricing model (per seat, per usage, tiered, hybrid), packaging (which features in which plan), and discounting policy. Tom Tunguz, Patrick Campbell, and Madhavan Ramanujam all converge on one core idea: price your product to the customer's willingness to pay, not to your costs or competitors.

• Survey willingness to pay with the Van Westendorp four-question battery before pricing changes.
• Stratify by segment — enterprise, mid-market, SMB will tolerate different prices.
• Tier features around 'good / better / best' anchored on user job depth, not feature counts.
• Annual discounting compounds retention; use it deliberately.
• Re-price annually; pricing decay is the silent killer of growth.

Every PM faces this trade-off. Build when the capability is core differentiation; buy when it is undifferentiated infrastructure (auth, payments, observability); partner when the capability is in someone else's wheelhouse and the integration is the value. The classic mistake is building auth or payments from scratch because it 'looks easy' — and discovering the long tail of compliance, edge cases, and maintenance.

• Core capability + differentiated execution → build.
• Undifferentiated capability + commoditized vendors → buy.
• Adjacent capability + strong partner → partner with clear data and revenue boundaries.
• Re-evaluate annually; markets evolve and yesterday's build can become tomorrow's buy.

Janna Bastow's Now / Next / Later format is the roadmap antidote to date-based feature commitments. 'Now' is in-progress; 'Next' is committed for the upcoming cycle; 'Later' is directional and uncommitted. It buys teams the room to discover and iterate without the political cost of moving Gantt bars.

• Express each item as a problem theme + target outcome, not a feature name.
• Limit Now to active capacity; Next to one quarter ahead; Later to broad themes.
• Pair with OKRs so themes ladder to outcomes.
• Communicate confidence explicitly: 'committed', 'planned', 'exploratory'.