Opportunity Discovery: A Practical Guide for Builders Who Keep Building the Wrong Thing
Opportunity Discovery: A Practical Guide for Builders Who Keep Building the Wrong Thing
The Problem in Concrete Terms
You have technical skills. You can ship. But every project you launch gets crickets. You've built three side projects this year—each technically solid, each solving a "problem" you convinced yourself existed. None made a dollar.
The pattern is always the same: you pick an idea because it's interesting to build, not because someone is actively paying to solve it. You mistake "that's cool" for "I'll pay for this." You spend weeks on auth systems, dashboards, and perfect CI/CD pipelines for products zero people asked for.
Meanwhile, you see worse products than yours making money. They're ugly. Their code is held together with duct tape. But they found a real wound and put a bandage on it. You're still looking for a wound that matches your favorite bandage.
This isn't an ideas problem. It's an evidence problem. You're guessing instead of investigating.
Why the Usual Fixes Fail
Reading "how I found my SaaS idea" posts. Survivorship bias dressed as advice. The founder retroactively constructs a clean narrative from a messy process. They don't tell you about the six dead ends, the cold emails that got ignored, or the month they spent building for a market that didn't exist.
Brainstorming sessions. You and your cofounder sit in a room listing "problems." You generate solutions to problems you imagine people have. No customer is in the room. The output is a list of features, not opportunities.
Keyword research tools. Ahrefs tells you search volume. It doesn't tell you willingness to pay, urgency, or whether the searcher has budget authority. "How to invoice clients" has volume. "Invoice generator for freelance designers who bill weekly and hate FreshBooks" has buyers.
Building an MVP to "test the market." You build first, validate later. By the time you learn nobody wants it, you've sunk 200 hours. The sunk cost fallacy then keeps you polishing a product nobody will buy.
Asking friends "would you use this?" They say yes to be supportive. They won't say "I'd pay $50/month" because that's awkward. Politeness is not a purchasing signal.
Step-by-Step Solution
Phase 1: Pick a Pond Before You Fish
Don't start with an idea. Start with a market segment you can reach and understand.
Action: List five groups you have genuine access to. Not "SaaS founders" (too broad, too noisy). Think: "independent insurance adjusters in Texas," "wedding photographers who shoot 20+ weddings/year," "plant managers at midwest food packaging plants."
Criteria for each:
- You know 3+ people in this group personally or can get intros within a week
- They spend money on tools (evidence: they already use paid software)
- They have a recurring, painful workflow (not a one-time annoyance)
- You understand their domain well enough to spot BS
Cross off any group that fails these. Pick one. This is your pond. Stay here for the next 30 days.
Phase 2: Run Problem Interviews (Not Solution Interviews)
Goal: Find a workflow so painful they've already hacked together a workaround.
Recruit: Message your 3+ contacts. Script: "I'm researching how [role] handle [workflow area]. No sales pitch. 20 minutes. I'll send you a $20 coffee gift card." Pay them. It filters for serious people and respects their time.
Interview script (stick to this):
- "Walk me through the last time you did [workflow]." — Let them narrate. Don't interrupt.
- "What tools or spreadsheets do you use for this?" — Screenshot if remote. Photo if in person.
- "What's the most annoying part?" — Wait for the emotional language: "hate," "waste," "mess," "risk."
- "Have you tried to solve this? What happened?" — This reveals existing workarounds and budget.
- "If magic wand—what would the ideal flow look like?" — Only ask at the end.
Do not: Pitch your idea. Ask "would you use X?" Show mockups. Talk about features.
Do: Record (with permission). Take verbatim notes. Note exact language—they'll use these words on your landing page later.
Target: 8-12 interviews. Stop when you can predict their answers.
Phase 3: Score the Pain
After each interview, score the problem on four dimensions (1-5 each):
| Dimension | 1 | 5 |
|---|---|---|
| Frequency | Monthly or less | Daily/weekly |
| Urgency | "Nice to fix" | "Losing money/reputation now" |
| Budget evidence | Free tools only | Already paying for partial solutions |
| Workaround complexity | Minor annoyance | Custom scripts, VAs, spreadsheets, duct tape |
Threshold: Total score ≥ 14/20. Below that, keep interviewing or switch ponds. You're looking for a bleeding neck wound, not a paper cut.
Phase 4: Map the Current Workflow in Detail
Take the highest-scoring problem. Diagram every step of their current process—including the human workarounds.
Example output for "freelance designer invoicing":
- Complete project → 2. Email client for PO number (2-day delay) → 3. Create invoice in FreshBooks (manual line items) → 4. PDF export → 5. Email to client + CC bookkeeper → 6. Track in Notion "awaiting payment" → 7. Follow up at net-15, net-30 (manual calendar reminders) → 8. Reconcile in QuickBooks (manual entry) → 9. Chase late payments (awkward emails)
Identify: Which steps are manual? Which cause errors? Which involve context-switching? Which steps do they hate?
This diagram is your product spec. Every box is a feature candidate. The hated manual steps are your MVP.
Phase 5: Validate Willingness to Pay (Pre-Sell)
Build a one-pager. Not a landing page builder. A Google Doc or Notion page with:
- Headline in their language: "Stop manually reconciling FreshBooks → QuickBooks every Friday"
- 3 bullet pain points from interviews
- The proposed workflow (your diagram, simplified)
- Price: "$XX/month per seat"
- CTA: "Join 5 pilot partners — get 50% off lifetime + direct Slack access to founder"
Send personally to your 8-12 interviewees + 5 warm intros they gave you. Script: "Based on our conversation, I'm building this. Looking for 5 pilot partners who'll give weekly feedback. Half price for life. Interested in a 15-min walkthrough?"
Success metric: 3/10 say "yes, send me the link to pay." Not "looks cool." Not "let me know when it's ready." They click a Stripe link or reply "invoice me."
If you get 0-1: The pain isn't urgent enough, or your solution misses the mark. Return to Phase 2 with a narrower problem slice.
If you get 3+: You have an opportunity. Build only the pilot features. Nothing else.
Phase 6: Build the Pilot, Not the Product
Scope ruthlessly. One workflow. One user type. No settings, no onboarding flow, no admin dashboard, no integrations unless the pilot users demand them during weekly calls.
Ship in 2-3 weeks. If it takes longer, you're building product, not pilot.
Weekly 15-min calls with each pilot. Watch them use it. Note where they hesitate. Fix that week. Ignore feature requests—watch behavior.
Graduation criteria: 3/5 pilots use it 3+ times/week without prompting for two consecutive weeks. Then you have a product. Now you can build auth, settings, landing page, marketing.
Checklist
- [ ] Listed 5 accessible market segments with paying-tool evidence
- [ ] Selected one pond; have 3+ warm contacts
- [ ] Conducted 8-12 problem interviews (recorded, verbatim notes)
- [ ] Scored each problem on Frequency, Urgency, Budget, Workaround
- [ ] Found at least one problem scoring ≥ 14/20
- [ ] Mapped the full current workflow for that problem (diagram)