# Reference Customer Discovery > A high-touch discovery framework to achieve product-market fit by co-creating solutions with a specific number of "reference customers." Use this when launching a new product, entering a new market, or when you need to validate deep customer value before scaling engineering. - Skill: `majiayu000/reference-customer-discovery` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds add majiayu000/reference-customer-discovery` - Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/reference-customer-discovery/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Product & Planning - Author: majiayu000 (https://skillmd.com/u/majiayu000) - Updated: 2026-09-09 - Page: https://skillmd.com/skills/majiayu000/reference-customer-discovery --- # Reference Customer Discovery The "Holy Grail" of product work is creating a solution that customers love so much they are willing to put their reputation on the line for it. This process involves immersing yourself in the customer's environment to solve a problem manually before building scalable technology. Success is measured by a "certificate of appreciation": revenue, engagement, or a public reference. ## Core Principles - **Value over Usability:** Just because someone *can* use a product doesn't mean they will *buy* or *choose* it. Discovery must solve for Value (will they buy?) and Viability (does it work for our business?). - **The Reference Commitment:** A customer is only a "reference" if they have used the product, love it, and will tell others about it. - **Do Things That Don't Scale:** Recruit, interview, and even perform the service manually to understand the nuances of the problem before writing a single line of code. ## The Discovery Process ### 1. Define the Target and the Goal Identify the specific problem you are solving and the "evangelist" persona who has it. - **B2B Goal:** Find 6 to 8 reference customers. - **B2C Goal:** Find 15 to 25 reference customers. *Why these numbers?* Fewer than this might be an outlier; achieving this number indicates true product-market fit. ### 2. Recruit Early Adopters (Evangelists) Look for people who feel an "irrational" need to solve the problem. - **Technologists:** They love new tech for its own sake. - **Evangelists:** They have the problem so badly they are already trying to hack together their own solutions. - **The Pitch:** "I am trying to build the best [solution] for [problem]. If I build something you love, are you willing to give me a video testimonial/5-star review?" ### 3. Immerse in the Problem (The "Out of the Building" Rule) Don't rely on user interviews in a lab. Go to where the problem happens. - Spend time on-site with the customer. - Observe the "no-shows," the workarounds, and the failures in their current process. - Act as the "manual engine" for the solution (e.g., using spreadsheets and phone calls instead of an automated system) to find where the process breaks. ### 4. Iterate Toward a Single Solution If one customer wants a feature but the others don't, **do not build it.** - Your goal is to find the common denominator that all 6-8 (B2B) or 15-25 (B2C) customers need. - Tweak the "menu" or the process until every single reference customer says, "This is exactly what I need." ### 5. Secure the Reference Once the manual/MVP solution works, ask for the "payment": - For B2B: A signed contract or a public testimonial. - For B2C: A commitment to post a 5-star review on launch day. - If they hesitate, you haven't solved the value risk yet. Use their hesitation as data for the next iteration. ## Examples **Example 1: B2B High-Volume Hiring Product** - **Context:** A company needs to hire 800 people quickly for a new bakery acquisition. - **Input:** No software exists. The PM, Designer, and Engineer go to the site. - **Application:** They manually recruit via Craigslist and flyers, call every candidate to remind them to show up, and sit in on interviews. They realize the "no-show" rate is the real problem. - **Output:** They develop a scheduling and reminder tool based on the manual "reps" they did for McDonald's and Starbucks. The product launches with $32M in sales because it was built to solve the specific no-show pain points discovered on-site. **Example 2: B2C App Launch** - **Context:** Launching a new niche social or utility app. - **Input:** A prototype or beta version. - **Application:** Instead of a "quiet launch," the PM finds 25 people who have the specific pain point. They work with them for weeks, refining the UI and features based on real usage. - **Output:** On Day 1 in the App Store, the app already has 25 5-star reviews from the people who helped build it. The "marketing" is just the exact words those 25 people used to describe why they love it. ## Common Pitfalls - **The "Nice" Bias:** Users often say they like a product just to be polite. Avoid this by requiring a "reputation stake" (the public reference). If they won't give a reference, they don't actually love it. - **Premature Scaling:** Building automation before you’ve manually solved the problem for at least 6 customers. - **Customizing for Outliers:** Building "one-off" features to please a single loud customer. If it doesn't serve the whole reference group, it's a distraction. - **Relying on PRDs:** Writing long requirements documents instead of involving engineers and designers in the actual "on-site" problem-solving. True innovation happens when the engineer sees the customer struggle.