Opportunity Discovery: A Field Guide for When You Have No Ideas (Or Too Many Bad Ones)
The Problem in Concrete Terms
You are sitting at your desk. You have the skills, the time, and the capital to build something. But when you open your notebook or your code editor, the page stays blank.
Or worse: the page fills up with ideas you abandon by Tuesday. "Uber for dog walking." "AI wrapper for PDF summarization." "A better CRM for plumbers." You write them down, feel a spark, then spend three hours on Reddit or Product Hunt realizing the market is saturated, the margins are zero, or you simply do not care about the problem enough to grind through the boring parts.
This is the opportunity discovery trap. It manifests in three specific ways:
- The "Shiny Object" Loop: You chase trends (crypto, AI agents, no-code marketplaces) because they feel like where the money is, not because you have an edge there. You build demos, not businesses.
- The "Scratch Your Own Itch" Dead End: You build a tool for a problem you have, only to realize you are the only person who has it, or the 50 other people who have it refuse to pay for software.
- The "Research Paralysis": You read market reports, analyze TAM/SAM/SOM spreadsheets, and map competitor matrices until you know everything about the industry except what to build tomorrow morning.
The result is months of motion without progress. You are not "validating." You are stalling.
Why the Usual Fixes Fail
1. "Just talk to customers."
This advice assumes you know who the customer is and where to find them. If you are pre-idea, you don't. Cold DMing 50 strangers on LinkedIn asking "What keeps you up at night?" gets you ignored or generic answers ("I'm busy," "Hiring is hard") that lead to commodity SaaS ideas.
2. "Look for problems in communities (Reddit, IndieHackers, Twitter)."
By the time a problem is articulated publicly in a forum, it is already being solved by three funded startups and a dozen open-source projects. You are finding commoditized pain, not fresh opportunity.
3. "Follow the money / Look at growing markets."
Macro trends (aging population, energy transition, B2B SaaS consolidation) are true but unactionable. "Build something for the energy transition" is a strategy for a venture fund, not a solo founder or small team. It lacks the specificity required for a first commit.
4. "Idea generation frameworks (SCAMPER, Crazy 8s, etc.)."
These produce volume, not quality. They exercise creativity muscles but ignore the constraint that matters: your specific leverage. An idea is only good if you are the person uniquely positioned to execute it with unfair advantage.
Step-by-Step Solution: The Asset-Led Discovery Process
Stop looking for "ideas." Start auditing assets. An asset is anything you own that lowers the cost of execution or increases the probability of success for a specific problem. Assets are concrete: code libraries, proprietary data access, specific domain credentials, an audience of 500 niche professionals, a relationship with a wholesale supplier, a regulatory certification, or a unique workflow you have automated for yourself.
Phase 1: The Asset Inventory (Timebox: 2 Hours)
Create a spreadsheet with four tabs. Be ruthlessly specific. "I know Python" is not an asset. "I have a tested, documented Python library that parses messy HL7 healthcare messages into clean JSON" is an asset.
Tab 1: Code & Technical IP
- Private repos, scripts, automation tools, browser extensions, Figma component libraries, infrastructure configs (Terraform/Helm charts) you have written and maintained.
- Filter: Keep only things that solve a specific technical pain point (parsing, integration, visualization, compliance, speed).
Tab 2: Data & Information Access
- Proprietary datasets you have scraped/cleaned/licensed.
- Access to non-public APIs, industry databases, or physical archives.
- Unique data structures you have built (e.g., a normalized schema for a chaotic industry).
- Filter: Keep only data that is expensive/difficult for others to replicate and updates frequently.
Tab 3: Network & Distribution Channels
- List 20–50 specific people (names, not "VCs" or "CTOs"). Note: Role, Company, Specific Trust Level (Would they reply to a Slack in 5 mins? Would they intro a buyer?).
- Newsletters, Slacks, Discords, Forums, WhatsApp groups where you have posting history and reputation, not just membership.
- Filter: Keep only channels where you can post "I built X, who wants beta access?" without looking like spam.
Tab 4: Domain Credentials & Constraints
- Certifications, licenses, past job titles, security clearances, union memberships, languages spoken, physical location.
- Constraints are assets too: "I can only work nights," "I cannot hire," "I have $5k budget," "I must stay in healthcare compliance." Constraints narrow the search space productively.
Phase 2: The "Adjacent Possible" Mapping (Timebox: 3 Hours)
For every high-value asset (top 3–5 per tab), answer this prompt in a doc:
If I gave this asset to a competent founder in this specific niche, what is the first product they would build that sells for $500/mo or $5k one-time?
Do not answer for "the market." Answer for the specific person you listed in Tab 3.
Example:
- Asset: Parsed HL7 library (Tab 1).
- Contact: Sarah, VP Engineering at a 50-person RCM (Revenue Cycle Management) startup (Tab 3).
- Sarah's Problem: Her team spends 20 hrs/week manually mapping new client EHR feeds to their internal schema.
- Product: A self-serve "Schema Mapper" UI sitting on top of your library. Price: $2k setup + $500/mo.
Output: A list of 10–20 "Asset-Contact-Problem-Product" quadruplets. Discard any where you cannot name the contact or the specific manual workflow.
Phase 3: The "Manual Concierge" Validation (Timebox: 1 Week per Quadruplet)
Do not write code. Do not build a landing page. Sell the workflow manually.
- Message the Contact: "I've built a script that automates [Specific Step X] for [Specific Workflow Y]. It cuts the time from 4 hrs to 15 mins. I'm running a paid pilot for 3 teams this month. $500 for the month, I run the script for you every morning, you get the output in your inbox/Slack. Interested?"
- If Yes: Collect the $500 (Stripe link, Venmo, invoice). Run the script manually (or via cron job you trigger) for 4 weeks. Document every edge case, error, and feature request.
- If No: Ask "Why not?" specifically. "Too expensive?" "Wrong step?" "Don't trust the output?" Update the quadruplet or kill it.
- Constraint: You are only allowed to build software (automate the script into a UI) after you have 3 paying pilots running concurrently for 2 weeks.
This proves three things simultaneously: The pain is real and urgent, the asset actually works in production, and you have a distribution channel (the contact).
Phase 4: The "Build vs. Buy vs. Partner" Decision Matrix
For each validated quadruplet, fill this matrix before writing a single line of product code.
| Criteria | Score 1–5 (5 = Strong Yes) |
|---|---|
| Asset Moat: Can a competitor replicate my core asset (code/data/network) in < 3 months? | |
| Margin Floor: At $Price, is Gross Margin > 70% after my labor costs (valued at market rate)? | |
| Expansion Path: Does this contact intro me to 3+ similar buyers without me asking? | |
| Founder Fit: Do I want to do this workflow every day for 3 years? | |
| Capital Efficiency: Can I reach $10k MRR without hiring? |
Rule: Total score < 16 → Kill or Partner (sell the asset/lead to someone else). Score 16–19 → Build slowly, keep consulting
Top comments (0)