Opportunity Discovery: A Field Guide for When You’re Stuck Guessing
The Problem in Concrete Terms
You have capacity—time, capital, or a team—but you don’t know where to point it. Every direction looks equally viable or equally risky. You’re not lacking ideas; you’re lacking a filter. The backlog is full of "maybe later" features, the competitor landscape is a blur of noise, and customer interviews feel like polite conversations that yield zero actionable signal.
Concretely, the stuck state looks like this:
- The Spreadsheet Trap: You have a prioritization framework (RICE, ICE, WSJF) but every score is a guess because the inputs—reach, impact, confidence—are unknowns.
- The "Strategic" Vacuum: Leadership asks for "big bets," but the only data you have is last month’s retention curve and a handful of support tickets.
- The False Positive Loop: You ship something based on a loud customer request. Usage spikes for two weeks, then flatlines. You realize they asked for a faster horse, not a car.
- Analysis Paralysis: You’ve read the market reports, scanned the G2 reviews, and mapped the TAM. You still cannot write a single sentence defining why now for any specific opportunity.
The core issue isn't a lack of creativity. It’s that you are trying to select an opportunity before you have discovered the evidence that makes one obvious. You are trying to decide in the dark.
Why the Usual Fixes Fail
1. Customer Interviews (Done Wrong)
Most teams run "discovery interviews" that are actually validation sessions. You pitch a concept, ask "would you use this?", and hear "yes." That isn't discovery; it’s politeness. Real discovery requires the customer to teach you their workflow, not react to yours. If you leave the call with a feature list instead of a workflow map, you wasted the hour.
2. Competitive Audits
Mapping competitor feature matrices tells you where the market was two years ago. It reveals table stakes, not gaps. Chasing parity puts you on a treadmill; it never reveals a wedge. The opportunity is almost always in the workflow step the competitor ignores because it doesn't scale or doesn't fit their architecture.
3. Internal Brainstorms / Hackathons
These generate volume, not signal. They optimize for internal excitement (cool tech, shiny UI) rather than external urgency (burning hair, budget allocated). The best opportunities usually look boring internally—"automate this manual CSV upload"—and get killed in a voting round because they aren't "strategic" enough.
4. Top-Down Mandates ("Go build AI")
A theme is not an opportunity. "AI," "PLG," or "Enterprise" are constraints, not starting points. Starting with the solution space (LLMs) forces you to retrofit a problem. You end up building a chatbot nobody uses because the real problem was a missing API endpoint.
5. Quantitative Dashboards
Dashboards show what happened (churn, drop-off). They rarely show why. You see a 40% drop-off at step 3. The dashboard cannot tell you if it’s a UX bug, a pricing shock, a missing integration, or a change in the user’s boss’s priority. You have to leave the dashboard to find out.
Step-by-Step Solution: The Evidence Stack
This process moves you from "I have no idea" to "I have three validated problem spaces worth betting on." It requires zero budget, only calendar control and a notebook.
Phase 1: Define the Hunting Grounds (30 Minutes)
Stop looking at "the market." Look at your adjacent possible.
- List your current top 3 customer segments (by revenue or strategic value).
- For each segment, write down the core workflow they use your product for. Be specific: "Upload CSV -> Map Columns -> Run Validation -> Export JSON," not "Data Onboarding."
- Identify the immediate upstream step (what they do 10 mins before logging in) and the immediate downstream step (what they do 10 mins after logging out).
- Constraint: You are only allowed to hunt in these 3 upstream/downstream zones. This kills the "boil the ocean" problem.
Phase 2: The "Shadow & Scribe" Sessions (2 Weeks, 6–8 Sessions)
Recruit 2–3 users per segment. Do not interview them. Watch them work.
- The Ask: "Can I screenshare for 45 minutes while you do [Core Workflow]? I won't talk. I just want to see where it gets messy."
- The Rule: Mute yourself. Do not help. Do not explain the UI. If they curse, struggle, switch tabs, open Excel, or screenshot an error—that is data.
- The Scribe Method: Take timestamped notes in a two-column table.
- Column A (Action): "User exports report to CSV -> Opens Excel -> VLOOKUPs against master sheet -> Copies result -> Pastes into Email."
- Column B (Friction Signal): "Sighs at VLOOKUP error. 'This breaks every Tuesday.' Switches to Slack to ask colleague for formula."
- Output: A "Friction Map" for each user. Highlight every tool switch, manual calculation, copy/paste, communication hop, and wait time.
Phase 3: The "Money & Mandate" Filter (1 Hour)
Review your Friction Maps. For every high-friction step, answer three questions using only evidence from the session (no guessing):
- Budget Owner: Who pays if this step goes wrong? (Name a title: "Director of RevOps," not "the team.")
- Existing Spend: What tool/service/headcount is currently budgeted to handle this step? (e.g., "They pay a contractor $2k/mo to clean these CSVs.")
- Trigger Event: What specific change makes this urgent right now? (e.g., "New regulator report due Q3," "Acquisition doubled data volume," "Key Excel expert quit.")
Discard any friction point lacking a Budget Owner, Existing Spend, or Trigger Event. These are annoyances, not opportunities. You are left with Validated Problem Spaces.
Phase 4: The "Wizard of Oz" Test (1 Week per Space)
Take your top 2 Validated Problem Spaces. Do not build. Simulate the solution manually.
- Define the "Magic Moment": The single output the user wishes appeared automatically. (e.g., "A clean, validated JSON file in their S3 bucket every morning at 6 AM.")
- Recruit 3 Pilot Users: From your shadow sessions or waitlist. Pitch: "We will deliver [Magic Moment] manually for 2 weeks. Free. You give us 15 mins feedback daily."
- Execute Manually: Use scripts, Zapier, VA labor, or your own hands. Deliver the output exactly as promised.
- Measure Only Two Things:
- Adoption: Did they use the output in their downstream workflow without prompting?
- Complaint Velocity: When you are late or wrong, how loud/fast do they complain? (Silence = low value. Anger = high value.)
Phase 5: The Bet Document (30 Minutes)
Write a one-pager for the winning space. No vision fluff. Only:
- The Workflow Step: [Specific step from Phase 1]
- The Evidence: [Quote + Timestamp from Phase 2]
- The Payer: [Title + Current Spend from Phase 3]
- The Trigger: [Event from Phase 3]
- The Validation: [Adoption % + Complaint Signal from Phase 4]
- The Wedge: The smallest technical build that replicates the Wizard of Oz output.
- Kill Criteria: "If we don't get [X] pilot users paying [Y] by [Date], we archive this."
Short Checklist
- [ ] Hunting Grounds Defined: 3 segments + Upstream/Downstream steps mapped.
- [ ] Shadow Sessions Booked: 6–8 sessions on calendar with real users (not champions).
- [ ] Scribe Template Ready: Two-column (Action / Friction) doc open during sessions.
- [ ] Zero Pitching: No concepts shown, no "would you like" questions asked during shadows.
- [ ] Friction Maps Complete: Every tool-switch, manual step, and workaround documented.
- [ ] Money/Mandate Filter Applied: Every candidate problem has a Budget Owner, Existing Spend, and Trigger.
- [ ] Wizard of Oz Designed: "Magic Moment" defined
Top comments (0)