Discovery Roadmap

Use this when a founder or team has a product idea (especially a community, social, or consumer product) but has not yet validated that the audience wants it. This skill produces a "Now / Next-Later" board of cheap experiments that reveal real interest before any platform gets built. It is the productized version of the Wingmates founder consultation: a competitive scan, a sequence of no-engineering experiments, a channel and content test library, success signals, and the bigger bets held for later.

Related skills: Pairs with /discovery-questions for the consultation itself, /discovery-debrief to capture the conversation, /jtbd-analysis for the demand behind the idea, /opportunity-solution-tree to connect outcomes to experiments, and /now-next-later-roadmap for the roadmap shape. Feed the result into /case-study and /engagement-journal.

Inputs

Ask the user to provide:

• Product idea: {{product_idea}} (what it is, who it is for)
• Target audience: {{target_audience}} (who the first members or users would be)
• Known constraints: {{known_constraints}} (founder time, budget, channels they can already reach, any hard no-gos)

If the user has run a founder consultation already, pull the notes; otherwise run /discovery-questions first.

Process

Step 1: Reframe the goal from build to learn

Open by separating the two questions: "what should we build" and "what can we learn this week for almost nothing." For an early audience-dependent product, the second comes first. State the riskiest assumption in one sentence (usually: that the target audience wants this and will show up and engage). Everything downstream is a cheap test of that assumption.

Step 2: Run a competitive / existing-community scan

List the existing products, communities, groups, or resources the target audience already uses. For each, capture in one line: what format it uses, what topics draw people in, and what Wingmates-style product could borrow or do differently. The point is to avoid reinventing what already works and to learn the audience's existing language.

Step 3: Sequence the "Now" experiments (no engineering)

Lay out a left-to-right sequence of low-effort moves the founder can start this week without building a platform. The Wingmates default sequence, adapt as needed:

1. Research existing groups and interest
2. Seed a small community through referrals
3. Open a free instance of an existing tool (e.g. Slack or Discord) to gauge topic interest
4. Share relevant content per channel to see what earns a reaction
5. Gauge success on a few key signals
6. Incentivize members to refer others

Each step gets a "to learn / to decide / to consider" note so the founder knows what each experiment is meant to reveal.

Step 4: Build the channel and content experiment library

Propose candidate channels or topics the audience might care about. For each high-potential channel, write one or two concrete tests (a poll, a prompt, a question) so "gauge interest" becomes a list of things to actually post. Keep these cheap and specific.

Step 5: Define success signals

Name the few signals that tell the founder an experiment worked. Favor leading indicators that are visible early: engagement and activity, share of organic posts from members vs. the founder, event attendance, reactions and replies. Avoid vanity totals.

Step 6: Hold the bigger bets in "Next / Later"

Put everything that costs real effort into a Next / Later column: broader promotion, virtual and in-person events, expert mentorship, integrations with other tools, and monetization. State the rule that gates each one ("once we know X about our members"). If monetization appears, always keep a free tier.

Step 7: Assemble the board

Lay the output out as Miro-ready columns so the founder can drop it onto a board. Use the structure in reference/board-template.md. Add a short cover note: the riskiest assumption, the first three moves, and what a "go" signal looks like.

Output format

Produce two things:

1. A short cover note (4-6 sentences): the riskiest assumption, the reframe from build to learn, the first three Now moves, and the signal that would justify investing more.
2. The board, in these labeled sections:
   • TODAY / NOW: low-effort things to learn about potential members (the Step 3 sequence, left to right, each with a to-learn/to-decide note)
   • Competitive / existing-community scan (Step 2, one line each)
   • Channel + content experiments (Step 4, channel then the test to run)
   • Success signals (Step 5)
   • NEXT / LATER: things to do once we know more (Step 6, each gated by a learning condition)

Rules

• No em-dashes. Use commas, colons, or en-dashes.
• Keep every "Now" move genuinely low-effort: if it needs engineering, it belongs in Next / Later.
• Tie every experiment to a thing it is meant to learn. An experiment with no learning attached is just a task.
• Always preserve a free tier if monetization appears.
• Mark anything uncertain with .
• This output is a discovery artifact, not a build plan. End on what to learn next, not what to ship.