Gavel Gavel
Experts Skills
Cited from real sources 5 min read Updated May 2026

A framework by Garry Tan

Garry Tan's Launch-Jankiest Rule for Early-Stage Startups

The launch-jankiest rule, articulated by Garry Tan on YC Office Hours, is the single most important shipping habit for an early-stage founder: launch the jankiest version of the product that still provides minimum value, then iterate from real customer reaction. The fear of launching kills more startups than competition does. Founders polish for an imagined audience that does not exist while the small group of people who actually have a hair-on-fire problem never get a chance to use the thing.

The rule

"Launch early,
launch often."

If you give somebody a janky product and they still use it, that speaks volumes about how important the problem is that you're solving.

Garry Tan YC Office Hours Watch at 14:57

The framework

A janky MVP is a demand test, not a product test

Most pre-launch advice is about product polish. The launch-jankiest rule is about reframing what the launch actually does. A janky MVP is not asking the market whether your product is good. It is asking whether the underlying problem is urgent enough that someone will tolerate an embarrassing version of the solution. If they do, the problem is real. If they walk because of the jank, the problem was probably a nice-to-have.

Tan's filter is concrete. You are not looking for a hundred users. You are looking for five or six:

It's more about filtering down those 100 into the five or six who actually have the problem you're addressing, who have such a hair on fire problem they might use your janky MVP solution. They might even feel empathy for your company.

That is the entire game in the first 90 days. Five customers with hair on fire teach you more than five hundred curious browsers. The hair-on-fire crowd will give you feedback that is actionable. The curious crowd will give you feedback that sounds smart and means nothing.

The trap founders fall into is treating the first launch as a one-time event with a permanent record. It is not. The launch is a probe. The market does not remember most startup launches and never will. The pressure founders place on launch day is a story they are telling themselves, not a fact about the world.

How to apply it

Cut scope until launch feels uncomfortable

The right amount of polish is a little less than the founder thinks is acceptable. Optimize for the speed of the next iteration.

  1. 1

    List every feature you think you need before launch.

    Then strike through anything that does not produce minimum value end-to-end. Most pre-launch backlogs have three or four items that are actually required and twenty that are not.

  2. 2

    Make sure the core thing actually works.

    Tan's caveat to the rule: launching too early is fine unless the product fundamentally does not provide value. Janky is not synonymous with broken. The seven things you kept must complete the loop.

  3. 3

    Replace automation with manual labor where possible.

    Don't build a payment system, send an invoice. Don't build a notification engine, email the user yourself. Manual processes are reversible decisions; engineering is a one-way door.

  4. 4

    Find five hair-on-fire prospects before launch.

    If you do not personally know five people whose problem is urgent enough that they would tolerate a janky version, you are not ready to launch. You are ready to do customer research.

  5. 5

    Treat launch day as Tuesday.

    Pick a date, ship on it, watch what happens. The pressure to make launch day a coordinated PR moment is the pressure that delayed everyone else's launch. Skip it.

  6. 6

    Diagnose week by week after launch.

    If no one signs up, the messaging is wrong, the targeting is wrong, or the problem is not what you thought. Tweak one variable per week. Most pivots are misdiagnosed as pivots when they are just bad week-one messaging.

  7. 7

    Launch again every three months.

    The first launch is a probe. The second is iteration. The third is when traction starts to compound. YC startups launch on multiple platforms multiple times. The first launch is not the launch.

Launch early, launch often. The best founders use launches to diagnose problems, find their early adopters, iterate on the product based on customer feedback, and then they launch again.
Tan on the launch cadence Watch at 17:25

Boundary conditions

When it works, when it fails

Works best when

  • The market has clear hair-on-fire users (B2B saving hours per week, consumer with acute pain)
  • You are pre-product-market-fit and learning is the goal
  • The core loop can be cut to one thing that works end-to-end

Fails when

  • The product needs regulatory approval, audit, or enterprise trust to be tried at all
  • The core loop is genuinely broken (janky != broken — the product must produce minimum value)
  • You already have a brand and audience expecting polish; janky becomes a brand violation, not a probe

April Dunford would push you to also know what problem you're solving for which alternative before you launch (otherwise feedback from the early five is hard to interpret). The launch-jankiest rule and Dunford's competitive alternatives are complementary: jankiest answers when to ship, alternatives answer what to say when shipping. Both before traction.

The receipts

Where Tan discusses this

07:56
Timestamp

YC Office Hours — Launch sooner than you think

Filtering for hair-on-fire users

You're trying to filter into the five or six who actually have the problem you're addressing, who have such a hair on fire problem they might use your janky MVP solution.
14:57
Timestamp

YC Office Hours — Launch sooner than you think

Janky as a demand signal

The benefits of launching way before you even feel ready: it's the best way to test if somebody really wants it. If you give somebody a janky product and they still use it, that speaks volumes about how important that problem is that you're solving.
17:25
Timestamp

YC Office Hours — Launch sooner than you think

Launch early, launch often

I don't think you can launch too early as long as there's some minimum value that the product actually provides. Launch early, launch often. The best founders use launches to diagnose problems, find their early adopters, iterate, and then launch again.

Useful? Pass it to a founder polishing month four before launch.

Want the full playbook?

Get 329 execution & shipping frameworks.

49 frameworks 97 rules 175 heuristics & principles 7 operators

From Gretchen Rubin, Jake Knapp, BJ Fogg, and 4 more. Drop one .md into Claude, Cursor, or ChatGPT. Your AI cites practitioners, not guesses.

See the pack

Instant .md download · One-time purchase · No subscription

Related frameworks

Positioning

Competitive Alternatives Positioning

April Dunford

Start with what customers would do if you did not exist, not who you compete with.

Pricing

Gasp Pricing

Alex Hormozi

If the buyer doesn't gasp at the price, you didn't go high enough. Anchor, then walk down.

Distribution

Core Four Customer Acquisition

Alex Hormozi

Warm, cold, content, ads. Pick one. Four hours a day for a quarter.

Pricing

Willingness-to-Pay 20/80 Axiom

Madhavan Ramanujam

20% of features drive 80% of willingness to pay. Most founders give that 20% away free.