| name | research-market-problem |
| description | Investigate and document a customer or market problem using current, sourced research and authorized connected information. Use when validating a problem hypothesis, learning who experiences it, understanding triggers and consequences, comparing current alternatives, or producing a research record that can inform future personas, offers, messaging, and campaigns. |
Research Market Problem
Turn a problem hypothesis into a sourced, reviewable problem record. Separate observation, inference, and open questions.
Resolve relative paths from this skill directory. The MarketingOS root is ../...
1. Define the research question
Require:
- the problem hypothesis;
- market, geography, or category scope;
- relevant customer roles when known;
- the decision the research should inform;
- recency or deadline requirements.
Read existing files under ../../research/, ../../personas/, ../../offers/, and ../../evidence/. Read ../../../business/objectives/objectives.md when the research is connected to a company outcome. Do not assume an existing persona or offer is correct merely because it exists.
2. Plan the evidence search
Identify questions about:
- who experiences the problem and in what situation;
- triggers, frequency, severity, and consequences;
- language customers use to describe it;
- current tools, services, manual processes, and the option to do nothing;
- evidence that contradicts or narrows the hypothesis;
- unresolved segments, contexts, or alternatives.
Use authorized internet, browser, file, and connector access. Do not bypass authentication, access controls, robots restrictions, or platform limitations.
3. Gather credible sources
Prefer direct customer language, primary research, official data, first-party product information, and reputable current sources. Use secondary commentary to discover leads, not as the sole support for important claims.
For every source, record:
- title, publisher or speaker, and URL or local path;
- publication date and capture date;
- what was directly observed;
- the question the source helps answer;
- limitations, audience, and potential bias.
Use short excerpts only when necessary. Summarize instead of copying long source passages.
4. Test the hypothesis
Actively seek both supporting and contradictory evidence. Distinguish:
Finding — directly supported by a source;Inference — a reasoned interpretation of findings;Hypothesis — a proposition that still needs evidence;Unknown — information not established by the available research.
Do not invent market size, frequency, costs, customer quotations, or confidence. Do not upgrade a problem to Validated solely because several articles repeat the same unsourced assertion.
5. Write the problem record
Create or update ../../research/problems/<problem-slug>.md using ../../templates/problem-record.md.
Include:
- a concise problem statement;
- affected roles or segments;
- situations and triggers;
- consequences and current alternatives;
- supporting evidence and counterevidence;
- source links and confidence;
- open questions;
- possible implications clearly labeled as unapproved.
Use status Hypothesis, Supported, Validated, or Rejected. Require human acceptance before setting Validated. Link persona, offer, or evidence IDs only when those records exist.
6. Preserve research inputs
Store permitted original source files under ../../research/source-material/. Store market-wide findings under ../../research/market/ and alternative or competitor findings under ../../research/competitors/ when separate records are useful. Do not duplicate material already available through an authorized connector unless durable local preservation is necessary and permitted.
7. Report the result
Return:
- current problem status;
- strongest supporting findings;
- meaningful counterevidence;
- remaining questions;
- files created or updated;
- the next research or human decision.
Do not create public claims, finalize personas, define offers, or publish content as part of this skill.
