Strategy · the sequence of bets
The densest module, narrowed to what carries the judgment. It funnels from many raw inputs to a few pillars, tests each aspiration for who you win with and against, and gives every bet a defensibility spine. Deeper worked treatment lives in PRO: strategy.
Strategy Blocks (Janakiraman) — the funnel from noise to a few winning-aspiration pillars.
Above the fold
Strategy is what you say NO to.
A focused sequence of bets bridging vision to reality — insights → bets → objectives, never a dated feature backlog (Cagan).
Name who you win WITH and AGAINST.
A winning aspiration names customers and competition. An internal metric or a statement that survives the substitutability swap is playing-to-play (Lafley & Martin).
Don't drop Hard-to-copy.
Defensibility is the leg teams skip in DHM. Product/tech innovation is the easiest to copy — a Product-Performance-only roadmap is the default trap.
Foundations
The base layer: empowered teams given problems, strategy as a sequence of bets, the funnel that produces pillars, a vision skeleton, and the litmus test that separates winning from participating.
Cagan / SVPG
The shift from feature teams (told what to build) to empowered teams (given problems). Core dimensions: staffing/competency, product vision, team topology, product strategy, team objectives (problems framed as OKRs, not feature lists), discovery + delivery practices.
Empowerment needs stronger managers, not fewer — "empowered in name only" (autonomy without coaching or context) underperforms a well-run feature team.
Cagan
A focused sequence of bets bridging vision → reality; pick a small number of significant problems. Don't flatten to "have a strategy" — strategy is what you say NO to. Insights → bets → objectives, never a dated feature backlog.
Janakiraman · Lenny's 2025 — a named framework, not settled canon
Two tracks: small-s (this-quarter focus) + big-S (multi-year). Funnel: 50-150 raw inputs → cluster to 10-15 → 3 pillars + explicit non-goals → ~3 HMW questions. Score pillars on a 4-dim rubric that MUST include uniqueness/defensibility — the dim most teams omit.
Write each pillar as a winning-aspiration headline — the assertion, not the topic (headline style traces to Lafley & Martin, Playing to Win). A 5-phase rollout timeline; the two tracks feed one roadmap — don't run them as two disconnected plans.
Strategy-doc skeleton → §TemplatesCagan / Amazon
"For [customer] who [need], [product] is a [category] that [benefit]. Unlike [alternative], we [differentiation]."
A 3-10yr aspiration, not a roadmap. Pressure-test by writing the Amazon PR/FAQ (press release + FAQ) before building — if the customer benefit won't write cleanly, the bet isn't clear yet.
Lafley & Martin · Playing to Win
Gut-check any aspiration/vision: a winning aspiration names who you win with (customers) AND who you win against (competition) — playing-to-participate has neither and just "serves a segment." Two tells you're only playing to play:
(1) It's an internal metric ('grow 25%', 'be the #1 player') — a scoreboard, not a strategy. (2) It has no competitive dimension — it survives the substitutability swap (paste a rival's name in and it still reads true), e.g. 'be the preeminent brand'.
Fix: state winning specifically (leading share of segment X; highest NPS in Y) and name the competitor you take it from. Don't anchor the aspiration on the North-Star number — that's the proxy, not the goal.
The reinforcing cascade
Where Cagan gives the bet portfolio, Playing-to-Win gives the spine each bet must satisfy — plus the canon each strategy leans on and the exec memo that carries it to a board. HIGH
Winning aspiration
What does winning look like — for whom, against whom.
Where to play
Chosen together with how-to-win — a market you can enter but not win is the wrong where.
How to win
Low-cost OR differentiation — and low-cost ≠ low price (keep the margin).
Core capabilities
A reinforcing configuration of activities — a system, not a list of functions.
Management systems
The rules/measures/structures that sustain the strategy — skip them and it's a wish list.
Lafley & Martin
It's a reinforcing cascade (top choices frame the ones below; lower refine the ones above), not 5 independent boxes. Three things people get wrong: (1) where-to-play and how-to-win must reinforce each other — choose them together. (2) The two boxes everyone drops — capabilities (a configuration, not a function list) and management systems (the rules/measures that sustain it). (3) How-to-win is low-cost OR differentiation — and low-cost ≠ low price.
Pressure-test coherence: do all five read as one whole? It nests down org levels (build a sub-cascade only if meaningfully different from the one above). Sits beside Cagan's sequence of bets: Cagan is the bet portfolio; P2W is the where/how/capabilities/systems spine each bet must satisfy.
Biddle · the defensibility spine
A board-grade product-strategy memo is built on DHM moat decomposition: Delight (the customer value), Hard-to-copy (the actual moat — network effects / data / brand / ecosystem / tech; the leg teams drop), Margin (does the model improve unit economics?). Sequence the arc with GLEe — Get-big → Lead → Expand, progressing point tool → platform → system-of-record and sizing bottom-up TAM/SAM/SOM at each step.
Classify each bet Type-1 (irreversible) vs Type-2 (reversible) and attach explicit kill criteria. Exec-memo cue: lead with the bet → DHM moat → GLEe position → reversibility + kill criteria, then evidence.
The metric/risk spine shifts by context — public co (durable margin + guidance), regulated/FDA (evidence + approval gates over speed), hardware (irreversible releases → higher pre-commit bar), marketplace (GMV / take-rate, two-sided liquidity).
Contrarian: a standalone CPO/product-board memo not tied to company strategy can itself signal strategic drift — if it reads as a product wishlist rather than a sequence of bets serving the company thesis, that's the tell.
Canon pointers · use-when, and the one calibration it gets wrong
| Framework | Use when | Gets wrong |
|---|---|---|
| DHM (Biddle) | Gut-check any strategy | Skipping Hard-to-copy — defensibility is the leg people drop |
| GLEe (Biddle) | Sequencing a 10-15yr arc | It's an order (Get-big→Lead→Expand); don't expand before you lead — sequence point-tool → platform → system-of-record, sizing TAM/SAM/SOM at each step |
| SWOT | 5-min initial scan | Unprioritized list, no "so what," no action |
| Porter 5 Forces | Industry attractiveness | Static snapshot; ignores complementors & ecosystem shifts |
| PESTEL | Macro / regulated-entry scan | Produces trivia instead of implications-for-us |
| Ansoff | Naming a growth move | Understates diversification risk (new product × new market) |
| Value Chain | Where value is created/captured | Drawn for your firm, not the customer's |
| Product-Market matrix (Aumayr) | Coverage gaps across segments × products | Confuses coverage with actual demand |
Segmentation & market entry
Aumayr · non-prior B2B-industrial canon
Segment by geography, demographics, psychographics, behavior, needs, and firmographics (B2B). Product portfolio: ABC analysis on revenue AND contribution margin (not revenue alone) to rank priorities; review age structure across the portfolio.
Function-Technology matrix: map customer functions/needs × available technologies to surface innovation whitespace.
Market-entry sequence · entering a new market or segment
Attractiveness
Size bottom-up TAM/SAM/SOM (not top-down %), scan macro with PESTEL.
Competitive intensity
Porter applied to entry, not the whole industry.
Beachhead / wedge
Moore beachhead test + Where-to-Fish quadrants (High-Evidence/High-Need first; avoid Low-Ev/Low-Need), Blank "get out of the building" IRL to validate, require ≥3× Zone-of-Benefit.
Build–Partner–Buy
Build = max control, slow, best when core/differentiating; Partner = fast, low control, when speed > ownership and non-core; Acquire = fast+control, costly/integration-risk, best when time-to-market is decisive and a target exists.
Go / No-Go
Minimum-viable-entry criteria, first-90-day milestones, explicit kill criteria, ring-fenced resources so the bet isn't starved by the core.
Platform & architecture strategy
MIT Designing Product Families · Meyer & Lehnerd
A platform is intentional sharing of parts/code/process across variants to deliver variety on a smaller cost base. Decide in order:
(1) intent — cost-savings (leverage one core) or new-markets/revenue (reach segments you couldn't serve one-off); name which. (2) segment the variant grid (price tier × user/job), one variant per cell you'll serve + a volume mix. (3) share-vs-vary by hierarchy level — choose at what level sharing happens (whole-product / assembly / module / component / code) and tag each module common / configurable / unique; differentiate only where the customer feels it — the rest is hidden commonality that benefits only you. (4) horizon — share against a stable core; let fast-refresh modules vary on their own clock.
Two architecture levers: modular (loosely-coupled, swappable) buys variety/reuse/parallel teams/cheap upgrades; integral (tightly-coupled) buys technical performance (latency/footprint) — tight constraints force integral, and modularity is a spectrum (Höltta-Otto & de Weck 2007). Predict what must change for future segments and isolate it behind stable interfaces so variation can't cascade (Martin & Ishii 2002).
⚠️ Commonality erodes by default — 'divergence' (Boas, Cameron & Crawley 2013; mgmt levers: Cameron et al., EMJ 2017): variants drift to less sharing than planned (variant 1 pays, later variants reap), so hold it with explicit governance. Price in commonality risk (shared-part failure = broad blast radius) and the capability penalty (an over-spec'd common part overcharges the cheap variant).
8-decision template, lever menu, GVI/CI, PCI, postponement, design-for-changeability → PRO: strategyChristensen, Verlinden & Westerman · Industrial and Corporate Change, 2002
Set architecture and value-chain boundaries on customer satisfaction, not core-competence instinct: where functionality is underserved → an interdependent/integrated architecture, own the stack (tight coupling wrings out performance); overserved → modular, disintegrate (speed/customization now decide).
An interface can modularize only when the customer can specify it, metrics measure it, and the customer understands the interdependencies — until then keep it inside the firm. The sharper outsourcing rule: outsource only overserved interfaces — never the still-performance-limiting one, because that's where attractive profit migrates (profit sits at whatever stage is not-yet-good-enough; as a stage overshoots it commoditizes and margin moves on — the dynamic engine under Porter's static map).
Overshoot decision rule, three-condition gate, profit-migration + disk-drive/PC cases → PRO: strategyPlatforms, innovation & commoditization
Eisenmann/Parker/Van Alstyne, HBR 2006 · Parker & Van Alstyne, Mgmt Sci 2005
A platform links ≥2 user groups whose value to each other rises with the other side's size. Beyond steady-state liquidity, four moves the single-product playbook gets wrong:
Network-effect TYPE. Cross-side (more A pulls B — the engine) vs same-side (more A helps/hurts A). Same-side is often negative — sellers want fewer rivals (Covisint stalled on it). Diagnose both.
Price the SIDES, not the product. Pick a subsidy side (≤ cost, build the volume) and a money side (pays a premium for access). A side can rationally run free forever (Adobe: readers free, writers pay).
Winner-take-all? One platform tends to win iff: high multi-homing cost on ≥1 side · strong positive cross-side effects · no strong demand for special features (else niches survive — Amex keeps fat margins at ~5% of Visa, directional). If WTA, fight-vs-share is a bet-the-company call.
Envelopment is the threat DHM moat-decomposition misses: an adjacent platform sharing your user base bundles your function for free (Microsoft→RealNetworks), not a same-category rival.
Which side + 4 failure cases, envelopment defenses, openness governance, cold-start → PRO: strategyDoblin/Deloitte · Keeley, Pikkel, Quinn & Walters, 2013
Innovation isn't only product features. Doblin's empirical taxonomy sorts every move into ten types across three groups — Configuration (Profit Model, Network, Structure, Process), Offering (Product Performance, Product System), Experience (Service, Channel, Brand, Customer Engagement).
It's a diagnostic, not a sequence: map yourself AND competitors across all ten to find errors of omission (the types your category ignores), then combine several into one integrated "play" (each leg copyable, the combination not). Load-bearing: product/tech innovation is the easiest to copy, so a Product-Performance-only roadmap is the default trap — top innovators integrate ~2× the types (directional; can't credit innovation alone).
Ambition dial: Core (few types) · Adjacent (3-4) · Transformational (5+); new-market entry needs adjacent-or-above.
The ten types, the diagnose→play method, ambition dial → PRO: strategyCarr · "IT Doesn't Matter," HBR 2003
Distinguish proprietary tech (ownable → can sustain advantage) from infrastructural tech (worth more shared → commoditizes through proprietary → buildout → commodity, its power to differentiate any one firm declining as it becomes accessible to all). When a capability becomes essential to competition but inconsequential to strategy, its risks outweigh its advantages → shift offense to defense: spend less · follow don't lead · focus on vulnerabilities, not opportunities.
⚠️ Apply "follow, don't lead" only to the commoditized layer — for a still-generative tech (frontier AI) the proprietary window is the practice/insight you build on top, not the model you rent.
Lifecycle test, three rules, buildout evidence, the AI-trap → PRO: strategyTry it
Two market-entry instruments, live — bottom-up TAM/SAM/SOM built from the segments up, and the weighted Build · Partner · Buy decision matrix.
The TAM/SAM/SOM builder needs JavaScript to run. Run it on the Tools page.
The Build·Partner·Buy matrix needs JavaScript to run. Run it on the Tools page.