positioning

Positioning Skill

Structured positioning statement exercise for startups and teams. Combines deep competitive research with forced-choice strategic questions to produce an honest, compressed positioning statement.

Designed for accelerator portfolio companies, internal use, or any team that can't answer "why us?" in one sentence.

Philosophy

Most positioning exercises produce corporate fluff because they skip the hard questions. This skill forces honesty by:

• Researching competitors BEFORE asking questions (so you can't hide from reality)
• Making every question a forced choice (no "all of the above")
• Requiring the team to write independently, then collide answers
• Compressing the final output to under 30 words

The output is not marketing copy. It's a strategic alignment tool.

Workflow

PHASE 1: Context gathering

Action: Use AskUserQuestion to collect basic info.

Questions:
1. "What does the company do? (2-3 sentences, plain language)"
   - Free text
2. "Who is the target customer today?"
   - Free text
3. "What stage? What's the team size?"
   - Options: [Pre-seed solo/duo, Pre-seed with team (3-8), Seed, Series A+]
4. "Is there an existing website, pitch deck, or memo I should read?"
   - Free text (URLs or file paths)

If the user provides a file path or URL: Read it. Extract product description, claimed differentiators, current messaging, pricing, and team.

PHASE 2: Competitive research

Action: Before launching research, detect available research tools by calling ListMcpResourcesTool or checking which MCP tools are available. Build a tool instruction block based on what's present.

Tool detection priority (check in order, use all that are available):

MCP Tool Pattern	What it provides	How to use
mcp__perplexity__perplexity_research	Deep multi-source research	Use for the primary competitive landscape query
mcp__perplexity__perplexity_search	Fast web-grounded search	Use for quick competitor lookups
mcp__exa__web_search_exa	Semantic web search	Use for finding similar/adjacent companies
mcp__parallel-search__web_search_preview	Parallel web search	Use for batch competitor searches
mcp__parallel-task__createDeepResearch	Analyst-grade research reports	Use for deep competitor analysis
mcp__firecrawl__firecrawl_search	Web search + scraping	Use for competitor website analysis
WebSearch (built-in)	Basic web search	Always available as fallback
WebFetch (built-in)	Fetch URL content	Always available as fallback

Compose the research tool instructions dynamically. For example:

• If Perplexity is available: "Use mcp__perplexity__perplexity_research for the main competitive landscape query, then WebFetch to verify specific competitor websites."
• If Exa is available: "Use mcp__exa__web_search_exa for semantic search to find companies similar to [description]."
• If nothing extra is available: "Use WebSearch for competitor queries and WebFetch on each competitor's website."

Launch 2 parallel agents with the detected tools:

Agent 1: Direct competitors

Launch an Agent (general-purpose) with this prompt:

Research the competitive landscape for: [company description from Phase 1].

[INSERT DETECTED TOOL INSTRUCTIONS — tell the agent exactly which tools to use]

Find:
1. The 3-5 closest competitors (same customer, same problem)
2. For each: one-liner positioning, pricing, key differentiator, weakness
3. What do ALL of them say? (common claims = table stakes, not differentiators)
4. What does NONE of them say? (potential white space)

Output as a markdown table + 3 bullet summary.

Agent 2: Adjacent/aspirational competitors

Launch an Agent (general-purpose) with this prompt:

Research companies adjacent to: [company description from Phase 1].

[INSERT DETECTED TOOL INSTRUCTIONS]

Find:
1. 2-3 companies solving a related problem for the same customer
2. 2-3 companies solving the same problem for a different customer
3. Any company that this team might secretly aspire to be (the "we're like X but for Y" reference)

For each: one-liner, why they matter as context.

Launch both agents in parallel. The skill works with just built-in WebSearch/WebFetch but produces richer research when Perplexity, Exa, or other research MCPs are configured.

After agents return: Present findings to user as a compact summary. Ask:

AskUserQuestion:
- "Anything missing or wrong in this competitive picture?"
  Options: [Looks right, Let me add/correct]
- "Which competitor do founders most often compare you to?"
  Free text

PHASE 3: The four questions

Action: Present the four positioning questions using AskUserQuestion. Each question is informed by the competitive research.

IMPORTANT: Frame questions using SPECIFIC competitor names and findings from Phase 2. Do not use generic placeholders.

---

Q1: The founder/customer test

header: "The customer test"
question: |
  Your ideal customer is evaluating you against [TOP COMPETITOR 1] and [TOP COMPETITOR 2].
  They can only pick one. Why do they pick you?

  Rules:
  - Don't say "price" or "we're cheaper"
  - Don't say "better technology" without specifics
  - Don't say "customer service" (everyone says this)
options:
  - [Generate 3-4 options based on research findings, each representing a distinct strategic direction]
  - "Something else (write your own)"

Q2: The honest gap

header: "The honest gap"
question: |
  What is the single biggest reason your ideal customer says NO today?
  Be specific. Not "we're early stage" — that's a cop-out.
options:
  - [Generate 3-4 options based on competitive weaknesses found in research]
  - "Something else (write your own)"

Q3: The core bet

header: "The core bet"
question: |
  You can only invest in ONE of these directions for the next 6 months.
  Which one?
options:
  - [Generate 3-4 options based on the white space found in research,
     each with a 1-sentence description of what it means concretely]
  - "Something else (write your own)"

Q4: Market scope

header: "Market scope"
question: |
  Which market are you actually in? Pick the one that defines your roadmap.
options:
  - [Generate 3-4 market definitions from narrow to broad,
     based on the company's product and competitive research.
     Example: "Smart contract security for Solidity only" vs
     "AI code security for all languages" vs
     "Critical infrastructure verification"]

Q5: Your first draft

After the four questions, immediately ask:

AskUserQuestion:
- "Now write your positioning statement. Under 30 words. Don't overthink it — gut reaction based on what you just answered."
  Template: We help [WHO] achieve [WHAT] through [HOW], unlike [ALTERNATIVE].
  Free text

Do not skip this. Save the user's draft verbatim. In Phase 6, show it side-by-side with the final synthesized version and call out what changed and why.

PHASE 4: Team collection

Action: After the user completes all five questions, generate the team exercise.

Output a clean markdown block that the user can copy-paste to their team:

# Positioning Exercise — [Company Name]

Answer each question independently. Don't discuss with teammates first.
Spend max 10 minutes total. Gut reactions > polished answers.

## 1. The customer test
[Customized question with competitor names from research]

## 2. The honest gap
[Customized question]

## 3. The core bet
[Customized question with forced-choice options]

## 4. Market scope
[Customized question with scope options]

## 5. Positioning statement
Fill in, keep under 30 words:
We help [WHO] achieve [WHAT] through [HOW], unlike [ALTERNATIVE].

---
Return all answers to [user name] by [user specifies deadline].

Then ask:

AskUserQuestion:
- "Send this to your team and paste all their replies here when you have them.
   How many people will respond?"
  Free text

PHASE 5: Synthesis

Action: When user pastes team replies, analyze them.

Analysis structure:

1. Alignment map — Where does everyone agree? (These are real. Use them.)
2. Contradiction map — Where do people disagree? (These are the strategic decisions that need to be made.)
3. Surprise findings — Anything one person said that nobody else did but is clearly right.
4. Gap analysis — Questions nobody answered well (signals the team hasn't thought this through).

Present this analysis, then ask:

AskUserQuestion:
- "For each contradiction, which direction do you want to go?"
  [Present each contradiction as a forced binary choice]

PHASE 6: Final positioning statement

Action: Generate the combined positioning statement.

Output format:

# [Company Name] — Positioning Statement

## The customer test
[2-3 sentences. Lead with the answer. No fluff.]

## The honest gap
[1-2 sentences. Brutally honest.]

## The core bet
[2-3 sentences. What you're building, what exists today vs. planned.]

## Market scope
[1 sentence. The market you chose and what it excludes.]

## Positioning statement
[Under 30 words. Fill in the template:]
We help [WHO] achieve [WHAT] through [HOW], unlike [ALTERNATIVE].

## Your first draft vs. final
> **Your draft:** [user's verbatim draft from Phase 3 Q5]
> **Final:** [synthesized version above]
> **What changed:** [1-2 sentences on what shifted and why — e.g. "You led with technology, the team led with outcome. The final version centers the customer problem."]

Quality rules for final output:

• No sentence over 20 words
• No adjectives that can't be verified (remove "innovative", "cutting-edge", "world-class", "unique")
• Every claim must be either (a) already true or (b) explicitly marked as a bet
• If something is aspirational, say "By [date], we plan to..." — not "We are..."
• Total length: under 200 words for all sections combined

Save to: Ask the user where they'd like to save the output.

PHASE 7: Iteration (optional)

If user wants to iterate, use AskUserQuestion to identify which section needs work. Re-run only that section's question with tighter constraints.

Anti-patterns to avoid

• Don't generate positioning without research. The competitive context is what makes questions sharp.
• Don't let the user skip the team exercise. Solo positioning statements are echo chambers.
• Don't use marketing language. If it sounds like it belongs on a SaaS landing page, rewrite it.
• Don't accept "all of the above" answers. Force choices. The pain of choosing IS the exercise.
• Don't make the final statement longer than 30 words. Compression is clarity.