What the One Feature Test Actually Does
It's not about shipping with one feature forever. It's about clarity.
You're forcing yourself to answer: What does the user hire my product to do? Not what's nice to have. Not what competitors have. What solves the specific problem in your thesis?
Michael Seibel's advice at YC: Before building, write down the one thing that, if it works, proves your hypothesis. Y Combinator's Scott Belsky calls this 'functional scope'—you have unlimited iterations, but limited feature scope.
Most Indian B2B SaaS founders I've seen fail add 15 features in month two because they're talking to customers who say, "Great, but we also need X, Y, Z." They say yes to everything. Three quarters later, they're diluted, users are confused, and the product doesn't do anything brilliantly.
The Framework: Three Questions
Question 1: What is the user's actual job-to-be-done?
Not "improve productivity." Specific. "A recruiter needs to filter 200 resumes to 10 candidates in 30 minutes."
Not "better logistics tracking." Specific. "A 3PL operator needs to see which truck left the warehouse late, by how much, and why."
The specificity kills bad ideas early. If you can't articulate the job in two sentences, you don't understand the problem.
Question 2: Which one feature solves that job?
Now you're forced to remove everything else. For the resume filter: Do you need resume parsing, scoring, or bulk export? Pick one. For the logistics operator: Do you need real-time tracking, delay prediction, or root cause analysis? Pick one.
When Stripe started, they could have built invoicing, recurring billing, analytics dashboards, compliance tools. They picked one: charging. Everything else came later.
When Slack started, they could have built file sharing, video calls, integrations, bots. They picked one: messaging that works better than email. The rest followed.
Question 3: If this feature doesn't work, does your thesis die?
If the answer is no, it's not your core feature. It's nice-to-have.
For a recruiter tool: If filtering doesn't work, the product fails. Resume parsing can come later.
For 3PL: If delay prediction doesn't work, you're just another tracker. Root cause is your edge.
Why Indian Founders Struggle With This
Three reasons I see repeatedly:
1. Feature parity obsession. Indian founders often check competitor feature lists and build to match. Competitors always have more. You'll lose this game. Instead, ask: Which one feature do we do better than everyone?
2. Sales pressure. Enterprise customers ask for ten things. You say yes to all. You ship six of them poorly. Better to ship one brilliantly and say, "This solves your biggest problem. Next quarter we build #2." Customers respect focus.
3. Fear of the idea being 'too simple.' Founders worry one feature looks cheap. It's not. It looks focused. Notion looked too simple once. So did Figma. So did Twitter. Simplicity is strategy.
The Non-Obvious Insight
The One Feature Test isn't about shipping less. It's about learning faster.
When you ship six features, you get six weak signals. User engagement is spread thin. You don't know which feature matters. You can't iterate meaningfully on any of them.
When you ship one feature ruthlessly, you get one strong signal. Users either adopt it or don't. You learn fast. You iterate on the one thing that matters.
Faster learning beats more features every time. Especially in a market like India where your competitors are VC-funded generalists. Your edge is depth, not breadth.
How to Run the Test
Step 1: Write your thesis in one sentence. ("Founders can close funding 4x faster with better due diligence software.")
Step 2: Identify the one feature that proves it. (Due diligence document management.)
Step 3: Build only that. Ship to 10 users who specifically need it.
Step 4: Measure: Did they adopt it? Did it solve their problem? Would they pay for it?
If yes: Double down, iterate, deepen.
If no: Your thesis is wrong, not your execution.
The Actionable Takeaway
Stop planning features. Start planning which one feature, if it works, means you were right all along.
If you can't answer that in two minutes, you're not ready to code yet.