Idea Discovery Frameworks
Where real ideas come from — observation, not inspiration. Be a detective, not a dreamer.
Where Real Ideas Come From
Most people wait for a “eureka” moment. That's not how it works. The best ideas are found through observation, not inspiration. Your job is to be a detective, not a dreamer.
The 5 Proven Sources of Business Ideas
| Source | Core Question | Example |
|---|---|---|
| Pain Points | What repeatedly frustrates you or people around you? | Founders billing clients manually → FreshBooks |
| Trends | What's growing fast that nobody's built for yet? | Remote work surge → Loom, Notion |
| Skills | What can you do better than 90% of people? | Coding + design → productized service |
| Market Gaps | Where do existing products fail their users? | Banks ignoring young people → Revolut |
| Remixed Ideas | What works in one market but not another? | Uber for X, Airbnb for Y |
Three Mental Models Worth Internalizing
"Scratch Your Own Itch"
If you've lived the problem, you understand it at a depth no research can replicate. You know the workarounds people use, the language they use, and what "good enough" looks like. Basecamp, Dropbox, and Notion all started this way. The risk: your problem might be unique to you. Validate externally.
"Boring Businesses Win"
The flashiest ideas attract the most competition and hype — and hype doesn't pay bills. Pest control, invoice processing, laundry pickup, B2B compliance tools — ugly problems with predictable demand, low competition, and customers who actually pay. Sexy ideas get clicks. Boring ideas get revenue.
"Distribution-First Thinking"
Most founders ask: "Is this a good product?" The right question: "How does this reach its first 1,000 customers?" If you can't answer that clearly before you build, you have a hobby — not a business. Think about distribution before features.
Using Tools to Surface Real Demand Signals
Don't guess what people want. Read the evidence they leave everywhere online.
| Tool | What to Look For | How to Use It |
|---|---|---|
| Google Trends | Rising search volume on a problem keyword | Compare 5-year trend; look for consistent upward slope, not spikes |
| Raw frustration in niche subreddits | "Search: I hate that", "why is there no", "anyone else annoyed by" | |
| Exploding Topics | Emerging trends before they peak | Filter by category; look for "Early Stage" labels |
| Twitter/X | Complaints and viral pain points | Search `[problem] is so annoying` or `I wish there was a` |
| Amazon Reviews | Exact language customers use + product gaps | Read 2-star and 3-star reviews — gold is in the middle |
| App Store Reviews | SaaS product failures | Sort by "Most Critical" — every complaint is a feature gap |
The Amazon Review Trick
Idea Filtering: The Shortlist
Spend 2 hours eliminating 80% of ideas so you can properly validate the 20% that survive.
The Quick-Kill Scoring Matrix
Score each idea across 5 dimensions in under 10 minutes. Be ruthless.
| Filter | Question | Score 1–5 |
|---|---|---|
| Market Size | Can this reach ₹1Cr+ revenue at scale? | 1 = tiny niche, 5 = mass market |
| Competition Level | Is the market winnable? | 1 = dominated by giants, 5 = fragmented/open |
| Monetization Clarity | Can I charge for this on day 1? | 1 = unclear, 5 = obvious pricing model |
| Founder-Market Fit | Do I have unfair access/insight here? | 1 = total outsider, 5 = lived the problem |
| Time to First Revenue | How fast can I make ₹1? | 1 = 12+ months, 5 = within 30 days |
Red Flags That Signal a Bad Idea
"Everyone could use this."
If your customer is "everyone," your customer is no one. The wider the audience, the harder and more expensive the acquisition.
The only path to revenue requires millions of users first.
Ad-supported models at early stage = slow death. You need a customer who pays on day one.
You can't describe the problem in one sentence.
If you need a paragraph to explain the pain, either the pain isn't real or you don't understand it yet.
Your competitive advantage is "better UX."
That's not a moat. Anyone can copy UI. Your advantage needs to be distribution, data, network effects, or domain expertise.
The market exists only because of a trend that might reverse.
COVID created demand for many things that evaporated by 2022. Validate that the problem is structural, not situational.
You need regulatory approval, manufacturing, or hardware.
Not impossible — but for someone under 22 with under ₹40,000 to validate, these kill speed.
Passion Ideas vs. Pain Ideas — Which Wins?
“Follow your passion” is the worst startup advice ever given. Passion is fuel. It is not a business model.
Idea Validation Framework
7 gates. Each one is a checkpoint. If you fail, loop back — don't barrel through.
The 7-Stage Validation Pipeline
Problem Validation
Is this Real, Painful, Recurring? Spend 2 hours reading Reddit, Twitter, Discord — without posting. Find 20+ unprompted complaints.
Market Sizing
TAM → SAM → SOM. Find one public stat and work backward. Investors care about your believable SOM, not your fantasy TAM.
Competitor Audit
List 8–10 competitors. Map them on 2 axes. Read their negative reviews. Find the empty quadrant — that's your opening.
Customer Discovery
10 interviews. DM people who complained publicly. Ask about past behavior — not hypotheticals. One honest convo > 10 hours of desk research.
Riskiest Assumption Test
Write every assumption your idea needs. Find the one that kills everything if wrong. Design the smallest test to prove or disprove it.
MVP Design
Not fewer features — the minimum thing to test your riskiest assumption. Landing page + manual delivery beats a perfect product that ships in 3 months.
Pre-Sell or Waitlist
The only validation that counts: someone giving you money before the product exists. 3–5%+ conversion = signal. Below 1% = rethink.
Step 1: Problem Validation — Is This Real, Painful, Recurring?
A problem worth solving has three qualities:
- Real — People actively deal with it, not theoretically.
- Painful — The cost of not solving it is high (time, money, stress, status).
- Recurring — It happens repeatedly, not once. One-time problems make bad subscription businesses.
Go to Reddit, Twitter, Facebook Groups, and Discord servers where your target customers hang out. Spend 2 hours reading without posting.
Step 2: Market Sizing — TAM / SAM / SOM
| Level | Definition | Example: AI Study Tool for Indian Students |
|---|---|---|
| TAM (Total Addressable Market) | Everyone who could theoretically buy | 300M students in India × ₹500/yr = ₹150B |
| SAM (Serviceable Addressable Market) | The segment you can realistically reach | 10M English-medium private school students × ₹500 = ₹5B |
| SOM (Serviceable Obtainable Market) | What you can realistically capture in 3 years | 0.5% of SAM = ₹25Cr |
The Key Insight
Step 3: Competitor Audit — Find the Gap
Competitors are not the enemy. They're proof the market exists.
List every competitor — Google `[problem] + tool/app/service/software`. Check G2, ProductHunt, Capterra, and the App Store. List 8–10 competitors minimum.
Map them on two axes — Pick the two dimensions your customers care most about (e.g., price vs. simplicity). Plot every competitor. Where is the empty quadrant?
Read their negative reviews — G2, Trustpilot, App Store 1-3 star reviews. Every repeated complaint is a market gap with a name on it.
Check their pricing pages — Understand their monetization model, tier structure, and what's gated behind paywalls. This tells you what customers value.
Step 4: Customer Discovery — 10 Interviews
This is the highest-ROI activity in validation. One honest conversation with a real potential customer is worth more than 10 hours of desk research.
How to get 10 interviews fast (with zero budget):
- Post in relevant subreddits: “I'm researching [problem area]. Would anyone be open to a 20-min chat? Happy to share findings.”
- DM 20 people in LinkedIn/Twitter who've publicly complained about the problem.
- Ask in WhatsApp groups, Discord servers, Slack communities.
- Offer a small incentive (₹200 Amazon voucher, or just genuine curiosity — it works).
Interview Script — Copy This
- “Tell me about the last time you dealt with [problem]. Walk me through it.”
- “How often does this happen?”
- “What did you try to fix it? What worked, what didn't?”
- “What's the cost of this problem — in time, money, or frustration?”
- “Have you paid for anything to solve this? What and how much?”
- “If you had a magic wand, what would the perfect solution look like?”
- “Who else do you know who deals with this?”
Never ask: “Would you use/buy this?” — hypotheticals lie. Past behavior doesn't.
Step 5: The Riskiest Assumption Test
Every business idea rests on a foundation of assumptions. One of them, if false, kills the whole thing. Find that assumption and test only that.
| Assumption | Risk Level |
|---|---|
| Students struggle with JEE prep | Low — obviously true |
| They use smartphones | Low — obviously true |
| They would pay ₹499/month for AI tutoring | HIGH — this is the riskiest one |
| AI can explain concepts at JEE level | Medium — testable |
| Parents would approve the spend | Medium |
Step 6: MVP Design — What's the Minimum Version?
| MVP Type | When to Use | Example |
|---|---|---|
| Landing Page MVP | Test demand before building | "Join the waitlist" page with a Google Form |
| Concierge MVP | Deliver manually, pretend it's automated | Do it by hand first — fake the automation |
| Wizard of Oz MVP | User thinks it's automated; you're doing it manually | "AI" chatbot that's actually you |
| Pre-sell MVP | Charge before you build | Sell 10 spots at a discount; if 0 buy, stop |
| Prototype MVP | Test the UX/flow without backend | Figma clickthrough or a Notion template |
For most software ideas: A landing page + a manual delivery process is enough to test willingness to pay. Build the automation after someone gives you money.
Step 7: Pre-Sell or Waitlist Test
This is the only validation that actually counts — someone giving you money or meaningful commitment before the product exists.
The Pre-Sell Method (Budget: ₹0–₹2,000)
Build a landing page (Carrd, Framer, or even a Notion page) — free.
Write one clear headline: the problem you solve, for whom, and what result they get.
Add a "Buy Now" or "Join Waitlist" CTA.
Drive 200–500 targeted visitors (Reddit post, Twitter thread, WhatsApp group blast, ₹1,500 in Facebook/Instagram ads to a targeted audience).
Measure: If 3–5%+ click through and take action, you have signal. Below 1% — rethink the offer or the audience.
The Idea Scorecard
Score any business idea out of 100. Use this before spending more than 2 hours on any idea.
Problem Intensity is below 6 — mandatory kill regardless of total score.
1 = minor annoyance, 10 = costs real money/time daily
⚠ Non-negotiable — must score 6+
1 = <₹10L opportunity, 10 = ₹10Cr+ realistic in 3 years
1 = "I'd never pay", 10 = already paying for inferior solution
⚠ Non-negotiable — must score 6+
1 = Amazon/Google competes, 10 = fragmented + underserved
1 = no context, 10 = lived the problem, know the customers
1 = 12+ months, 10 = can earn in under 30 days
1 = no idea how to reach customers, 10 = clear channel + low CAC
1 = anyone can copy overnight, 10 = data/network/brand moat
1 = zero interest, 10 = genuinely curious about this space
Score Interpretation
| Score | Verdict | Action |
|---|---|---|
| 75–100 | 🟢 GO | Move to full validation immediately |
| 55–74 | 🟡 ITERATE | One or two criteria are weak — fix the concept, retest |
| Below 55 | 🔴 KILL | Don't polish a bad idea. Move to the next one |
Mandatory Rule
Real-World Examples
Two ideas. One that looks great but fails. One that looks boring but wins.
A social app for college students to anonymously rate their professors.
Why it looks good: Huge addressable market, relatable problem, viral potential, no huge competitor in India, technically buildable.
| Criteria | Score |
|---|---|
| Problem Intensity | 7/10 — students do want this |
| Market Size | 8/10 — millions of college students |
| Willingness to Pay | 2/10 — nobody pays for this |
| Competition Openness | 6/10 — Quora, Shiksha exist but not focused |
| Founder-Market Fit | 6/10 |
| Speed to Revenue | 1/10 — ad model requires massive scale |
| Distribution Clarity | 5/10 — viral potential but no clear channel |
| Defensibility | 3/10 — trivially copied |
Where it breaks:
- Zero willingness to pay — it's entertainment, not a must-have.
- Monetization path is ads-at-scale — you need millions of users before ₹1 of real revenue.
- Legal and institutional pushback — universities will block or suppress it.
- Anonymous rating platforms have a moderation nightmare (defamation risk).
A WhatsApp-based reminder and follow-up service for small Indian business owners to chase unpaid invoices.
Why it looks boring: Manual, unglamorous, not AI-powered, nobody tweets about it.
| Criteria | Score |
|---|---|
| Problem Intensity | 9/10 — unpaid invoices = direct revenue loss |
| Market Size | 7/10 — 63M+ MSMEs in India, millions struggle with collections |
| Willingness to Pay | 9/10 — they're losing ₹lakhs; ₹500/month is nothing |
| Competition Openness | 8/10 — enterprise tools exist; nobody serves small businesses simply |
| Founder-Market Fit | 7/10 — if you know any small business owner |
| Speed to Revenue | 9/10 — can charge on day 1, deliver manually |
| Distribution Clarity | 8/10 — target chartered accountants, business WhatsApp groups |
| Defensibility | 6/10 — data + relationships over time |
Why it works:
- The pain is financial — a business losing ₹2L in unpaid invoices will pay ₹500/month without blinking to recover it.
- No behaviour change required — customers already use WhatsApp; you're not asking them to learn a new app.
- Start manually, automate later — on day one, you can literally send the reminders yourself. No code required to validate.
- Clear distribution channel — CA communities, business owner WhatsApp groups, trade associations.
First 30 days: Sign up 5 small business owners at ₹499/month. Send their invoice reminders manually via WhatsApp. If 3 of 5 renew in month 2, you've validated the business model. Build automation only after proof.
Common Traps & Cognitive Biases
Knowing these biases doesn't protect you from them. Build safeguards into your process.
Confirmation Bias in Idea Validation
You fell in love with the idea before you started validating it. Now every signal you find "confirms" it. You remember the 3 people who said "that's interesting" and forget the 7 who said "I wouldn't pay for that."
Before every customer conversation, write down the 3 things that would make you kill the idea. Actively look for those. Ask someone you trust to steelman the case against your idea.
The "Friends Said It's Great" Trap
Your friends are not your customers. They like you. They don't want to hurt your feelings. Their "that's such a cool idea!" is worth exactly nothing as a validation signal.
Talk to strangers. People who have no social reason to be kind. If a stranger in a Reddit thread voluntarily says "I need this," that's a signal. If your best friend says "I'd totally use this," that's noise.
Building vs. Selling First
Most technical founders default to building. It feels productive. It's comfortable. It's also the most dangerous thing you can do before validation. You can spend 3 months building something nobody wants.
The right order: Sell → then build → then optimize. The minimum test: Can you get one person to give you money for a thing that doesn't fully exist yet? If yes, build. If no, figure out why before touching code.
Perfection Paralysis
Waiting until the idea is "perfect" or until you "know more" is a form of fear dressed up as diligence. You will never have enough information before talking to customers. The market will always tell you more than your head will.
Set a hard deadline. "I will run 10 customer interviews by [date 10 days from now]." Make it a constraint, not a goal.
Your Next 7 Days
From Idea to First Validation Signal
Brain-dump 20 ideas using the 5 Sources framework. No filtering yet.
Apply the Quick-Kill Scoring Matrix. Shortlist top 3 ideas.
For your #1 idea: spend 2 hours on Reddit, Twitter, Amazon reviews finding evidence of the problem.
Run the 7-step Competitor Audit. Find the gap in the positioning matrix.
Reach out to 20 potential customers. Target: book 5 conversations.
Run your first 3 discovery interviews using the exact script. Take notes.
Score the idea using the full Idea Scorecard. Decide: Go, Iterate, or Kill.
End of Day 7 Checkpoint
Built for builders who move fast and think clearly.