Competitive Research

You help produce a sourced competitive matrix for a focused research question. The output is a markdown table plus sourced observations, landed in research/.

Input

The user provides via $ARGUMENTS:

• A short research topic (e.g., "how competitors handle first-time user onboarding"), OR
• A path to a research-prompt file (e.g., samples/sample-competitor-prompt.md)

Workflow

1. Read the prompt

If $ARGUMENTS is a file path, read it. Otherwise use the prompt directly.

Extract (or ask in one round):

• Research topic - the question.
• Scope - which competitors (direct / adjacent / aspirational), any exclusions.
• Questions - 3-7 specific things you want answered.
• Deliverable format - matrix (rows = products, columns = questions) by default.
• Timeline and non-goals.

If any of these are missing, ask the user in ONE AskUserQuestion round. Skip any that are clear.

2. Pick the competitor set

If the user provided competitors, use them. Otherwise suggest a set and confirm:

• Direct: 2-3 products in the same pricing tier and feature surface as your own (list your company's known competitors from CLAUDE.md or ask).
• Adjacent / aspirational: 1-2 products in related spaces known for excellent practice in the research topic.
• Exclude: tools clearly out of scope.

Confirm the list with the user before proceeding.

3. Gather observations (sourced)

For each product, for each research question, gather observations. Use available tools:

• WebFetch / WebSearch for public product docs, blog posts, marketing pages, changelog posts, and product-hunt-style coverage.
• Any documentation MCP your environment has configured (check the MCP Servers Available section of CLAUDE.md).
• Internal research notes in Loose Notes/Work/ that mention the competitor by name.

Rules for observations:

• Sourced: every observation has a URL or an internal note reference. No "I think they probably...".
• Observed vs. inferred: mark each cell as observed: (you can point to a specific page/screenshot) or inferred: (your reasoning based on other evidence).
• Date-stamped: competitive landscapes move fast. Note the date of the source when it's older than 6 months.

4. Build the matrix

Produce a markdown table:

Cells can be long. Keep them scannable but informative.

5. Summarize the landscape

Below the matrix, write a 3-5 paragraph synthesis:

• Common patterns: what do all / most of the competitors do similarly?
• Divergences: where do they differ meaningfully, and why?
• Outliers: any surprising approaches worth investigating further.
• Implications for us: how this should inform our decision (without prescribing a specific solution yet).

6. Note methodology and limits

A short section at the end:

• Sources consulted: docs, blog posts, pricing pages, external analyses.
• Gaps: what you couldn't find out (and would need a customer conversation or a demo account to verify).
• Confidence: where you're confident in observations vs. where you're inferring.

7. File location and naming

• Folder: research/. Create if it doesn't exist.
• Filename: YYYY-MM-DD - Competitive - [Topic slug].md. Example: 2026-04-20 - Competitive - Onboarding UX.md.
• Frontmatter: tags: Research, Competitive, created: YYYY-MM-DD.

8. Link the research

• Daily journal: under ## Notes.
• Related epic (if any): add a link in that epic's ## Breakdown > Explore section.

9. Output summary

✅ Research: [[YYYY-MM-DD - Competitive - Topic]]
✅ Linked in today's journal

Products compared: [N]
Questions answered: [N]
Sources cited: [N]
Gaps flagged: [N]

💡 Key pattern: [One-line top insight]

Next step: run `/opportunity-solution-tree` to convert insights into opportunities.

Notes

• Sourced over speculative. If you can't find evidence, say so.
• A matrix is a structure, not a verdict. Don't prescribe "we should do X because Competitor Y does it."
• Competitive research ages fast. Add the source date for anything older than 6 months. A year-old blog post about a product's onboarding is probably wrong now.
• The Extended Opportunity Solution Tree framework pairs well with this - use /opportunity-solution-tree next to turn patterns into opportunity hypotheses.