Chapter 06
Product-minded founders do not only need more demos. They need more evidence: what evaluators were trying to do, what confused them, what alternatives they considered, and what problem was urgent enough to ask about.
Stand Guidebook
Chapter 06 of 09
Field guide
Stand turns website chat into a continuous customer-learning loop. The Product Discovery skill is the structured version: help first, ask consent, drill into reality, and preserve evidence with links back to the conversation.
Use this chapter when your website traffic can teach you about positioning, product-market fit, pricing, migration blockers, documentation gaps, or the language prospects use before they become fans.
Help the visitor before asking discovery questions.
Ask permission before the Product Discovery skill switches into interview mode.
Aim the conversation at up to three current research questions.
Review evidence cards with links back to the original chat.
Why this matters
Founders usually hear from friendly customers, sales calls, and support tickets. That leaves a gap: evaluators, skeptics, confused visitors, and people comparing alternatives before they are ready to commit.
Those visitors are often honest in a way fans are not. They ask what your page failed to explain. They reveal what they are comparing you against. They describe the job they actually need done.
Example
Example
Interview discipline
The Product Discovery skill should not ask leading roadmap questions. It should ask about the last time the problem happened, what the visitor tried, who was involved, what broke, what it cost, and what happens if nothing changes.
The agent should stop when the visitor becomes rushed, frustrated, terse, or asks for direct help. The goal is learning with consent, not extracting a survey.
Transcript study
This example starts with a practical Copilot-to-Figma question, asks consent, then follows the visitor into team structure, handoff pain, workarounds, and a support-worthy next step.
Conversation ended
Question design
A discovery question is only useful if the answer could change something: the headline, the pricing explanation, the docs path, the onboarding promise, the roadmap priority, or who the founder interviews next.
This is where Stand can be unusually valuable. It lets you ask one or two timely questions at the exact moment the visitor is wrestling with your category, not weeks later in a scheduled interview.
Setup
Do not ask the agent to learn everything. Give it the current questions that would actually change your next product or positioning decision.
Good research questions are specific enough to guide follow-ups but broad enough that a real visitor can answer from experience.
Review
A good discovery report should not sound like an oracle. It should cite the conversations behind each finding, identify caveats, and keep weak signals separate from strong ones.
This is what lets a founder act without fooling themselves. You can change a headline after five similar confusion signals. You should not rewrite the roadmap after one enthusiastic feature request.
Digest section
Questions
No. It adds a stream of grounded evidence from visitors who are already evaluating, confused, or comparing. Use strong signals to plan deeper interviews.
The agent helps first, asks permission, asks one question at a time, and stops when the visitor wants direct help or seems done.
Continue the guide