Customer Research

Part of Marketing & Sales Workflow Skills. Each Skill can be used on its own.

Give your agent a research question that can change your next decision

Prepare questions about people's actual experiences, or work through research notes you already have. Customer Research guides your agent to separate reported behavior, opinions, requests and commitments, then connect the evidence to the decision you are trying to make.

For a founder exploring a product idea, that might mean deciding which part of a workflow to investigate before building. The method helps organize what you know and what would change your mind. It does not supply real participants, conduct interviews or create evidence on their behalf.

What you receive and can produce

The download contains readable Markdown instructions, linked method references and a blank input template, together with the licence and buyer guidance. Use it with your own agent for interview preparation, analysis of supplied material or a research review. This is a method-file download rather than a hosted research service or subscription.

Depending on the task, you can ask for a focused interview outline, an evidence record linked to source notes, or a research summary showing what supports or contradicts each hypothesis. The output distinguishes remaining gaps and proposes a small next check with a completion condition. You review the evidence and decide whether to take that step.

Start with a decision and the material you may use

Give your agent the current decision, the product or idea if one exists, the group or situation you want to understand, and any relevant constraints. For analysis, add the notes or other material you have permission to use, with source locations and the context needed to interpret them. You can use de-identified notes; the method does not require personal identifiers.

The instructions focus questions on recent experiences, current workarounds, priorities and constraints. They keep problem research separate from pitching a solution. When evidence already answers a question, it can be reused rather than starting another interview cycle.

An illustration: a compliment and a concrete incident

Illustrative fictional construction, written manually for this page. It has not been executed or tested through an agent or the Skill and is not a customer result.

Imagine a founder considering a shift-handover product supplies two invented note entries. Note A says, in summary, that a person liked the idea and thought they might use it. Note B records that the same fictional person recalled copying an unfinished-task list from a spreadsheet into a team message before a recent shift change. No spreadsheet, message, payment or real conversation has been supplied.

A compact illustrative evidence record could distinguish:

• Note A: an opinion and hypothetical intention; it does not establish adoption.
• Note B: a self-reported past behavior; it is more specific, but has not been independently checked.
• Still unknown: why that copying step was needed, how important it was, and whether the current process was satisfactory.
• A useful follow-up question: what prompted the copying step on that occasion, and what happened next?
• A possible decision: investigate that step before treating the compliment as a reason to build a feature.

The example shows how to preserve an evidence limit while finding a more useful question. It proves no customer problem, market demand or product benefit. Missing payment evidence alone would not establish that a need is absent.

Use the method with your agent

1. Extract the download into its own folder, keeping the relative file layout intact. Read the included licence and buyer guide.
2. Follow your host's supported file-import or Skill-installation procedure. If local Markdown files are supported, explicitly give your agent the folder's SKILL.md and the task, and let it read the linked references as needed.
3. Ask for preparation, analysis or a review. Check the output against the original material, particularly who said or did what, what was only reported, and what remains uncertain. Conduct any proposed research only with the necessary permissions.

Selected book methods, independently implemented

This original workflow is informed by selected methods in Rob Fitzpatrick's The Mom Test: asking about concrete experiences, investigating the reasons behind requests, and distinguishing compliments or hypothetical intentions from meaningful evidence. The source text used is recorded as v1.04, August 2013; equivalence to other editions was not verified.

The implementation adds its own evidence records, separate evidence-type and verification fields, hypothesis-by-hypothesis findings, client-context separation and handoff boundaries. It treats payment as additional evidence, without making payment a prerequisite for useful research. It is not a complete book conversion, an official adaptation or an author- or publisher-endorsed product. The book is not included or required.

When this fits

Use Customer Research when preparing a focused learning conversation, analyzing material you are allowed to use, or reviewing conflicting feedback. It can preserve contrary cases and mixed findings rather than forcing every note into a positive verdict. A prepared outline is a plan; a proposed next step is not an interview, trial or sale already completed.

The workflow does not recruit participants, contact customers, record conversations, verify product performance or prove that a small sample represents a market. Deal negotiation and live sales qualification belong to a different task. Loading this method does not authorize messages, data uploads, publication or spending.

Language, host and costs

Some instructions and references are Chinese or mixed Chinese/English. You can ask your agent to explain them and prepare the output in your preferred language.

Prior validation covers limited Codex on Windows discovery and synthetic use. Other agents and hosts remain unverified. Use your host's supported import procedure and confirm access to the linked files; no universal installer is included. The platform's buyer download-to-import chain has not been independently verified.

Your own agent or model's charges and terms are separate. The package does not charge for each local invocation or enforce technical copy protection or prompt secrecy. Neither a research finding nor a readable method guarantees adoption, sales or other business results.

Purchase and licence

USD9.90 buys this single current-version Skill once, with no fixed expiry and no required bundle. The licensor is Jason Huang, an individual. The purchaser personally, or one institution identified for the purchase, may use and adapt this original implementation internally across multiple agents and devices. Lawful outputs may be used commercially, including in clients' businesses.

You may not resell, publicly redistribute or sublicense the original Skill package, or pass that package to clients. The licence grants no source-book rights and implies no author endorsement. No lifetime updates, service-level agreement or compatibility with every agent is promised.

Installation help and refunds

For one installation issue reported within 30 days of purchase, email [email protected]. Support is limited to 15 minutes. Refunds follow the actual applicable Capafy download-product rules for your order; no separate 14-day refund promise is added.

Our purpose

We build practical tools to help you succeed in your work and business.

Buy Customer Research — USD9.90 once for the current version.