DEV Community

Alan Descure
Alan Descure

Posted on Originally published at smtm-store.pages.dev

Opportunity Discovery: A Practical Guide for When You're Stuck

Opportunity Discovery: A Practical Guide for When You're Stuck

The Problem in Concrete Terms

You have time, skills, and maybe even capital—but you cannot point to a single specific problem worth solving. Every idea feels either too small (a feature, not a business), too crowded (ten funded competitors), or too vague ("help people be more productive"). You sit down to brainstorm and write down the same three categories: SaaS for [niche], marketplace for [fragmented industry], AI wrapper for [workflow]. Then you stare at the list and close the notebook.

The stuckness shows up in three ways:

Analysis paralysis disguised as research. You've read 40 Substack newsletters, listened to 20 podcast episodes about "how I found my idea," and saved 200 tweets about "problems worth solving." You can articulate frameworks—TAM/SAM/SOM, jobs-to-be-done, wedge strategies—but you cannot name one unsolved problem you've personally verified in the last week.

The "good idea" trap. You filter for ideas that sound impressive to peers or investors rather than problems that are painful, urgent, and frequent for a specific group. You discard a boring but real problem (e.g., "HVAC contractors lose $3K/month on missed callback scheduling") because it doesn't feel "venture-scale" in the shower-thought sense, even though it meets every practical criterion.

No feedback loop. You have no mechanism to test whether a problem actually exists before building. You either build in isolation for months or do nothing at all. There's no intermediate step between "I think this is a problem" and "I'm writing code."


Why the Usual Fixes Fail

Customer development interviews. The standard advice: "Talk to 50 people." But you don't know who to talk to, what to ask, or how to avoid polite lies. Most people ask "Would you use a tool that...?" and get "Yeah, maybe" — which means nothing. Without a specific hypothesis and a concrete artifact to react to, interviews become therapy sessions for the founder, not validation.

Scratch your own itch. This works if your itch is shared by thousands of people with budget authority. Most itches are either too niche (you're the only one) or too generic ("I want better email") to support a business. And if you're a developer, your itches skew toward developer tools—a crowded, low-budget market.

Trend riding. "AI for X," "Web3 for Y," "No-code Z." By the time a trend is legible enough for you to spot it, the easy opportunities are taken. You're competing on execution against funded teams who started six months earlier. Trend-chasing also produces commodity ideas: the 14th "ChatGPT for PDFs" wrapper has no moat.

Marketplace / directory / aggregation plays. These feel safe because supply and demand are visible. But they require solving a chicken-and-egg problem and building trust and handling payments and managing quality—all before you know if the core transaction actually creates value. Most die at the "supply side won't join without demand" stage.

Framework collecting. You learn the IDEO design thinking double diamond, the YC "what do people want?" mantra, the "jobs to be done" canvas. Frameworks organize thinking but don't generate insights. You can fill out a canvas for a bad idea just as easily as a good one.


Step-by-Step Solution

Phase 1: Build a Problem Inventory (Week 1)

1.1 List every job you've done, industry you've touched, and community you belong to.
Include: past employers, freelance clients, volunteer roles, hobbies where money changes hands, family businesses, side projects. Don't filter. A summer job at a landscape supply yard counts. Managing a Discord server for 2,000 tabletop gamers counts.

1.2 For each entry, write down 5–10 recurring complaints, workarounds, or manual processes you observed.
Use this prompt: "What did people do in spreadsheets, whiteboards, group chats, or paper that made you wince?" Be specific. Not "bad scheduling" — "dispatcher texts drivers at 6 AM with route changes, drivers miss texts, jobs get missed, owner finds out at 4 PM."

1.3 Rate each problem on three dimensions (1–5):

  • Pain: How much does it cost in time, money, reputation, or stress?
  • Frequency: How often does it occur? (Daily = 5, Quarterly = 1)
  • Budget authority: Does the person feeling the pain control spend? (Owner = 5, Junior employee = 1)

Keep only problems scoring ≥10/15. You should have 8–15 candidates.

Phase 2: Narrow to Three Testable Hypotheses (Week 1–2)

2.1 For each remaining problem, write a one-sentence hypothesis:

"[Specific role] at [specific company type] loses [specific amount/time] because [current workaround fails], and they would pay [specific price] for [specific outcome]."

Example: "Independent HVAC contractors lose $3,000/month in missed callbacks because they manage scheduling in paper notebooks and group texts, and they would pay $150/month for a mobile app that auto-logs calls, schedules follow-ups, and sends reminders."

2.2 Identify the riskiest assumption in each hypothesis.
Usually: "They would pay $X" or "They actively look for a solution" or "The workaround is painful enough to switch." Not "I can build this."

2.3 Design a one-week test for the riskiest assumption.
Tests must produce a yes/no signal, not "interest." Examples:

  • Price test: Create a one-page description with price. Send to 20 target buyers via cold email/DM. Measure: how many click "buy" or reply "sign me up"?
  • Workflow test: Ask 5 target users to screen-record themselves doing the task today. Watch for cursing, workarounds, dropped steps.
  • Switching test: Offer a free manual version of the solution (you do it by hand) for 2 weeks. Measure: do they keep using it? Do they complain when you stop?

Pick the three hypotheses with the clearest, fastest tests.

Phase 3: Run Tests in Parallel (Week 2–3)

3.1 Execute all three tests simultaneously.
Do not sequence them. You want comparative signal. Spend 2 hours/day on outreach, 1 hour on recording results.

3.2 Track in a simple spreadsheet:
| Hypothesis | Test Type | Target Contacts | Responses | Positive Signals | Decision |
|---|---|---|---|---|---|
| HVAC scheduling | Price test | 20 | 8 | 3 "sign me up" | Continue |
| Freelance designer invoicing | Workflow test | 5 | 5 | 4/5 curse at current tool | Continue |
| Restaurant inventory | Switching test | 3 | 2 | 1 keeps using after week 1 | Kill |

3.3 Kill ruthlessly.
If a test yields <20% positive signal (e.g., <4/20 signups, <1/5 clear pain), discard the hypothesis. Do not "iterate on the test." The problem either exists or it doesn't. Move to the next candidate from your inventory.

Phase 4: Deepen the Winner (Week 3–4)

4.1 Run a second, higher-fidelity test.
If the price test worked: build a clickable prototype (Figma, not code). Run 5 usability sessions. Watch where they hesitate.
If the workflow test worked: do the manual service for 5 more users, charge 50% of target price. Measure retention at 2 weeks.
If the switching test worked: add one automated feature. Measure if usage increases.

4.2 Define the "wedge"—the smallest slice you can build and charge for.
Not "full scheduling platform." "Mobile app that logs incoming calls, lets you tap 'schedule callback,' sends SMS reminder to customer." One core loop. Buildable in 3–4 weeks.

4.3 Secure 3–5 paid commitments before writing production code.
Not LOIs. Not "I'd pay." Actual money: Stripe link, invoice paid, or signed contract with deposit. If you cannot get 3 paid commitments at your target price with a prototype, the problem isn't urgent enough. Return to Phase 2.

Phase 5: Build the Wedge (Week 5–8)

5.1 Set a hard 4-week build deadline.
Scope to what you can ship in 20 working days. Cut features until it fits. The wedge is not the product—it's the proof that the loop works.

5.2 Onboard your 3–5 committed users immediately.
They are your QA, your roadmap, your reference customers. Talk to

Start here

Top comments (0)