Idea to wedge

~8 min read

Why it matters

An idea is a hypothesis. A wedge is the smallest version of that hypothesis a specific customer will actually adopt, which can later open a bigger product. First-time technical founders usually start with the big product (“a platform for…”) and never get the first yes. The wedge is how you earn the right to build the rest.

You should already have repeated stories from Problem & customer. If you don’t, this page will just help you decorate a guess.

Core concepts

Why this. You, specifically, can reach these users or understand the workflow better than a random team. “I’m a CS student and the market is large” is not a why. “I was the treasurer and I already have the group chat” is.
Why now. Something changed that makes the old workaround worse or a new approach possible: a regulation, a new default tool everyone already uses, a cost curve (models, payments, distribution), or a behavior shift you can point at. “AI” is not a why-now. “Their current tool just priced them out” is.
The wedge is a job, not a feature list. One user, one painful step, one outcome. Airbnb’s wedge was airbeds during a sold-out conference, not “the travel platform.” Stripe’s wedge was “payments an engineer can install this afternoon,” not a bank.
Sequence matters. The wedge should make the second product easier: you now have the data, the relationship, the workflow, or the trust. If the wedge is a dead-end tool with no path to a bigger job, it might be a feature, not a company. That can still be a fine project — just don’t raise around it yet. See Fundraising map.
Narrow beats clever. A wedge you can explain in one sentence to the person who has the problem will outrun a clever architecture. Jargon (“OS for X”, “copilot for Y”) hides the fact that you haven’t picked the job.
Switching cost is the real competitor. You are not competing with a startup on Twitter. You are competing with the spreadsheet, the intern, the group chat, or doing nothing. Your wedge has to beat that, this month.

What to do next

Write three lines, out loud, to a friend who is not a founder: who it is for, the job, why you will win the first ten of them.
Write the why-now in one sentence that names a change with a date or a fact. If you can’t, you have a preference, not a timing argument.
Draw the sequence: wedge (this semester) → next job you earn the right to do → what you will not build yet.
Kill one bigger idea you are emotionally attached to. Write it down as “not the wedge” so it stops sneaking into the spec. That list is an input to MVP scope.
Check the wedge against the calls you already did. If none of those people would start with this exact job, the wedge is still a pitch.

Mental model

  big idea (someday)          wedge (this month)
  ----------------------      ---------------------------
  "platform for campus        "treasurers reconcile
   groups to run money"        dues without a spreadsheet"

  why you: you were the treasurer / you can text 20 of them
  why now: the old tool died, or Venmo isn't a ledger
  competitor: Sunday-night spreadsheet, not a startup

  wedge must:
    1. be usable by one person without a rollout
    2. beat the workaround on the job they already do
    3. leave you with a relationship or data for job #2

If the wedge needs the whole platform to be valuable, it is not a wedge.

Common failure modes

Starting from a technology. “We’ll use LLMs” is a means. The wedge is still a job someone has this week.
Why-now that is really “I’m free this semester.” Your calendar is not a market change. Use it as a deadline, not as the insight.
A wedge so small it teaches you nothing. A landing page with no workflow can test a message. It cannot test whether you can do the job. Know which question you are answering.
Copying a successful company’s current product. You are seeing chapter 10. Their wedge was chapter 1, and it looked embarrassingly small.

Watch

How to Get and Evaluate Startup Ideas | Startup School — Y Combinator. How to generate and grade ideas, including what “good” looks like before there is a company.
Lecture 1 - How to Start a Startup (Sam Altman, Dustin Moskovitz) — YC Root Access. The “why now” and “great idea” sections are the part to steal; the rest of the course is optional this week.
Competition is for Losers with Peter Thiel (How to Start a Startup 2014: 5) — Y Combinator. Monopoly language is easy to overlearn; use it to ask “where are we the default choice for one narrow job?” not to write a manifesto.