B2B positioning strategy

Develop clear, defensible positioning for B2B SaaS products. Synthesizes ICP + competitor + brand context into a binary strategy choice (category vs problem), primary/secondary anchors, defensible differentiators, and a positioning statement that survives the 5-second clarity test.

When to run

Invoke when the user says:

• "Help me position [product/company]" / "Positioning strategy for [company]"
• "What category should we be in?" / "How do I differentiate from [competitor]?"
• "Our homepage isn't converting" / "Our messaging is confusing"
• "Should we niche down or go broad?" / "How do I explain what we do?"
• "Multi-product positioning help"
• "Give me positioning options / scenarios" / "What are the decisions to make on positioning?" → scenarios mode (N fully-worked bets + canvas + recommended pick)

Do NOT invoke when:

• User wants landing page copy directly → landing-page-copy
• User wants full messaging library → product-messaging (run after positioning)
• User wants competitor research only → competitor-research
• User wants ICP research only → icp-behavioural

Workflow sequences:

• Greenfield: positioning → product-messaging → landing-page-copy
• Research-informed: icp-behavioural → competitor-research → positioning → product-messaging
• Audit fix: website-pm-score → win-loss-analysis → positioning → landing-page-copy
• Decisions to make: competitor-research → positioning --scenarios → (client picks) → product-messaging

Inputs

Required:

• Company/product name and basic description
• Website URL (accessible, fetchable)

Recommended (improve quality):

• icp-behavioural output → sharpens persona + problem anchors
• competitor-research output → enables competitive alternative mapping
• win-loss-analysis output → real decision criteria, objections
• Sales feedback → what's working in conversations
• website-pm-score output → clarity gaps to fix

If inputs missing: Ask for website URL. Offer to run website-pm-score first to assess current clarity. Identify founder type early (customer-focused / technology-focused / competition-focused) — drives the approach sequence.

Steps

Mode: Phase 3 presents the top-N anchor combinations as positioning options + a comparison canvas (the scenarios / decisions-to-make step), then collapses to the option the client picks and runs the validation steps on it. --scenarios[=N] emits just the options + canvas standalone (a competitor-research deck closer); default N=3. The single-committed-position outcome and the downstream contract are unchanged once a pick is made. → see the premium reference

1. Phase 1 — Discovery & current state. Fetch website (homepage, pricing, about, product). Apply 5-second clarity test on H1+H2. Identify founder type. Extract current category / differentiation / ICP / problem claimed. → see the premium reference
2. Phase 2 — Anchor & alternative mapping. Map 6 anchor types (Activity, Product Category, Use Case, Problem, Persona, Competitive Alternative). Map 5 alternative types (Manual/DIY, Legacy incumbent, Direct competitor, Adjacent tool, Status quo). Run Esner Decision Tree (Framework 13) to get a starting hypothesis. Score top combinations on Clarity / Differentiation / Relevance. Classify market maturity (emerging vs mature). → see the premium reference
3. Phase 3 — Present scenarios, decide, commit. Take the top-N scored anchor combinations from Step 2.4 (default 3) and expand each into a fully-worked positioning option: concrete hero H1/H2, primary + secondary anchors with type labels, value triad (category / problem / differentiation), secondary angle archetype (Framework 14: Niche / Low-Cost / Premium / Unique Attribute / Lite — max 2), thesis, and risks. Each option must shift its primary anchor or binary strategy — no two options that are the same bet reworded. Assemble the comparison canvas (one column per option; confidence dot per cell mapped to VERIFIED/INFERRED/ESTIMATED/UNAVAILABLE). Mark a recommended pick and name the trade-off each option forces. Present as decisions to make → client picks one. Then collapse to the chosen option: lock its binary strategy (category vs problem), anchors, and 2-3 defensible differentiators (delivery / guarantee / focus / proprietary); determine market focus with TAM math; pick the primary competitive alternative. → see the premium reference + the premium reference
4. Run clarity ladder. (On the committed option once a scenario is picked.) Compress positioning to 1 word, 1 phrase, 1 sentence, 1 paragraph. If any level fails, revisit anchor or differentiator selection.
5. Run guarantee test. For each differentiator: would you guarantee with money on the line? Verdict Pass / Fail / Partial. Cross-reference proof type per anchor (Framework 15: Activity → Before/After, Use Case → Workflow + Outcome, Category → Comparative, Competitive Alt → Switch/Upgrade).
6. Map strategic implications. Document what each decision implies for messaging direction, ICP sharpening, competitive counter-positions. If competitor-research available, include voice calibration. These feed product-messaging, icp-behavioural, tov-guidelines.
7. Compose output. Use template at the premium reference (scenarios mode / --scenarios: use the premium reference). Anchor-specific positioning statement template:
   • Activity: "We help [persona] [do activity] — replacing [manual process] with [product mechanism]."
   • Use Case: "We help [persona] [accomplish use case] without [key friction] — by [differentiated approach]."
   • Product Category: "We are a [category] that [key differentiator] — unlike [competitive alternative] which [weakness]."
   • Competitive Alternative: "We're a [leader]-alternative that [key upgrade] — for [persona] who [unmet need]."
8. Channel emphasis guidance. Same positioning, different facet leads per channel: homepage → primary anchor; thought leadership → Activity/Use Case; comparison pages → Competitive Alternative; sales discovery → Use Case/Activity; case studies → proof type matching primary anchor.
9. Self-evaluation. Run completeness, evidence, guardrail checks. Surface improvements. → see the premium reference
10. Review gate (Level 2). Present full strategy + decisions. Approve or iterate.
11. Suggest chain. If approved: "Want me to run product-messaging next?" Tight coupling.

Frameworks reference (15 total): all detailed in the premium reference. Anti-patterns + gotchas: the premium reference. Changelog + MCP integration: the premium reference.

What good looks like

Evaluations

• 5-second clarity test passes (H1+H2 answer "which tools" or "which tasks")
• Decision tree hypothesis recorded AND validated against scoring matrix
• Secondary angle archetype selected from Framework 14 (max 2)
• Clarity ladder passes at all 4 levels (word / phrase / sentence / paragraph)
• All differentiators have guarantee test verdicts (Pass / Fail / Partial)
• Proof types align to primary anchor per Framework 15
• No "platform for X" alone, no invented categories, no "AI-powered" as primary differentiator
• ≤3 differentiators (focus discipline)
• Activity anchor used only when no software category exists
• TAM math is realistic (ACV × customers needed) — not hopeful thinking
• Strategic implications mapped for messaging, ICP, competitive (and voice if competitor data available)
• Scenarios mode: exactly N options (default 3, range 2-4), each with a distinct primary anchor or binary strategy (no bet reworded)
• Scenarios mode: every option has a concrete hero H1+H2 (not a placeholder), a thesis, and a risks list
• Scenarios mode: canvas present, one column per option, a confidence dot on every cell mapped to VERIFIED/INFERRED/ESTIMATED/UNAVAILABLE (no all-green canvas)
• Scenarios mode: recommendation framed as decisions to make with the trade-off per option named — not a forced single answer

Final ship gate

Run /premortem --output before ship. See /premortem skill for the 5 execution domains (will-it-resonate / will-it-convert / will-it-stay-on-brand / will-stakeholder-push-back / will-it-degrade-over-time) and output template.

Trivial-case escape: ## Premortem\nNo failure modes — trivial change satisfies the contract for genuinely trivial outputs.