Opportunity Mapping

Opportunity Mapping

> Part of the discovery-phase skill pack · synthesis group · reads discovery-context.md (run profile-builder first if missing).

Turns the top pain points into opportunities and forces multiple candidate solutions before the team locks in. Two modes depending on whether the client already proposed a solution.

Step 1 — Read discovery context + insights

Read discovery-context.md (section 2. Product / Initiative → Solution already proposed by client — drives mode selection) and insight-matrix.md (anchors the tree against ranked insights).

If discovery-context.md is missing, ask the BA inline: "did the client propose a specific solution, or are we generating cold?" — picks Generate-mode vs Validate-mode; tag output [ASSUMED] on uncertainty. If insight-matrix.md is missing, ask: "name the top 1–3 pains in one sentence each" — or proceed with every leaf tagged [NO-EVIDENCE-ANCHOR]. Never block; recommend profile-builder / insight-synthesis for high-stakes work.

Step 2 — Pick mode

| Section 2 says | Mode | |---|---| | No solution proposed | Generate mode — open brainstorm of solutions | | Solution proposed by client | Validate mode — frame the proposed solution as ONE candidate, generate 2-3 alternatives, force comparison |

Tell the BA which mode is active and why. In Validate mode, this skill's job is partly political: the team needs to walk into the proposal/SoW with evidence either supporting the client's solution OR proposing a better one — silence equals "yes" to the client's existing idea.

Step 3 — State the desired outcome

One sentence: metric + direction + timeframe. Pull from problem-canvas.md success signal, sharpened with insight-matrix evidence.

> "Reduce time-to-first-value from 3 days to under 1 hour, this quarter."

Step 4 — Hang the opportunities

Each top-3 pain from insight-matrix becomes an opportunity branch. In user language, not solution language.

• ❌ "Add bulk import"
• ✅ "I spend 2 hours/day re-typing data from CSVs"

Step 5 — Pick one opportunity to attack first

Don't attack all three in one cycle. Pick based on:

• Strongest evidence (frequency × intensity from insight-matrix)
• Highest strategic fit (per discovery-context.md section 1 Stage)
• Cheapest to test

Defer the others to a "next cycle" section.

Step 6 — Generate candidate solutions

Generate 2-4 candidate solutions for the chosen opportunity. Force diversity along these dimensions:

• Incremental — extend client's existing product
• Radical — would require new infrastructure, ML, integrations
• Buy / integrate — wrap an existing tool, pay vendor, OSS
• Do-nothing baseline — better docs, scripts, contractors, no software

In Validate mode, the client's pre-proposed solution becomes one of these. Don't rig the others to lose — generate honestly.

Step 7 — Score solutions

| Solution | Bet size (S/M/L) | Time to build | Problem-fit confidence (H/M/L) | Build feasibility (H/M/L) | Aligned with client priorities (per profile) | |---|---|---|---|---|---|

The last column is critical for outsourcing. A radical solution might fit the problem perfectly but blow up the engagement timeline.

Step 8 — Pick direction

One sentence justifying the chosen solution vs runner-ups, with citations to insight-matrix and competitive-scan (substitute risk).

In Validate mode, if the chosen direction is NOT the client's pre-proposed solution: name the gap explicitly and prepare the BA for the conversation. Pre-draft 2 sentences the BA can say to the client.

Step 9 — Design 1-3 validation experiments

Each experiment must have:

• A clear test (concierge, landing page, clickable, paper prototype)
• A leading metric
• A kill criterion — what result means "don't build"

Experiments are days, not weeks. If an experiment needs >1 week, it's a feasibility-spike (validation group).

Output

./discovery/opportunity-tree.md per ./template.md.

Append to _log.md: [opportunity-mapping | YYYY-MM-DD] mode: ; opp: ; chosen: ; experiments: .

Anti-patterns

• One solution per opportunity. If you generated one, you didn't brainstorm. Force 3+.
• All solutions same bet size. All-incremental = anchored on status quo; all-radical = ignoring delivery realities.
• Rigging Validate-mode against the client. Generate the alternatives honestly. If the client's solution is best, document why; don't manufacture loss.
• Experiments that can only confirm. A good experiment can produce "no". If every outcome leads to "build it", you haven't designed an experiment.
• Attacking 3 pains at once. Sequence them.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: stanislavnianko
• Source: stanislavnianko/product-discovery-claude-skills
• License: MIT

Install and usage instructions live in the source repository linked above.