Common traps & weak answers

~8 min read

🎯 Interview frequent

Why it matters

Most failed founder interviews are not missing a secret framework. They die on a sentence: a jargon noun, a future-tense plan, a precise ratio built on four customers, or a bad fact tucked under a nicer word. This page is the list partners poke. Read it after the spine, then listen for these in your own mock. See Mock Q packs.

Core concepts

Marketing language. “Platform,” “AI-powered,” “operating system for,” “Uber for X” with no mechanic. Seibel’s contrast is the right test: “organizes the world’s information” tells an outsider nothing; a person, a query, and a list of links does. If a smart friend outside your major can’t repeat what you do, rewrite the sentence. See Product sense.
All future, no past. “We will talk to users. We will launch. We will expand.” The strong tense is past: we talked to these people, we shipped this, they did this. Future is allowed for the plan. It cannot be the whole answer.
TAM theater and fake precision. A top-down billions number, or “LTV/CAC of 4.7” with no paid acquisition and no lifetime. Ranges you can derive beat decimals you can’t. See Market size and Traction.
“No competitors,” and the full checkmark matrix. Both avoid the status quo. Name who they use today. See Competition.
Passion, vision, and the exit slide. Caring is not evidence. An acquirer is not a strategy. Dalton and Michael’s “don’t think like an investor” conversation is about this: market maps, exit slides, and a fully worked plan are how founders avoid the user.
Hiding the bad fact. Churn, a cofounder who left, a pilot that didn’t convert, a number that includes a one-time fee. Say it in the same breath as the flattering number. The short room finds it by arithmetic. Diligence finds it later, and then the issue is trust.
Tarpit ideas with no new fact. A familiar student idea (“an app to decide Friday night”) that hundreds of people have tried. You can still work on a crowded problem. You need a reason this attempt is different, from a user you can describe. “We use AI now” is not that reason.
Length. In a ~10 minute YC interview, a 90-second answer to a 10-second question is how you fail questions they never got to ask. Answer, stop, let them pull. Seed meetings forgive more length and still notice that you can’t be concise.

Mental model

flowchart TD
  S1["1. Plain sentence a non-expert can picture"] --> S2["2. A past-tense fact (user, number, or cut)"]
  S2 --> S3["3. The ugly qualifier, said by you"]
  S3 --> S4["4. Stop"]

If your answer skips 2, it is a pitch. If it skips 3, it is a spin. If it skips 4, it is a monologue.

Interview questions

What’s a weak answer to “what do you do?”, and the repair? Answer: Weak: “We’re building an AI-powered platform that revolutionizes operations for SMBs.” Repair: “We turn a clinic’s prior-auth PDF into a submitted form. The office manager at a four-doctor practice spent four hours on it last Tuesday. She pays $200 a month.” Same company. The second one can be questioned. That’s what you want.
You grew 50% last month. What’s the weak way to say that, and the strong way? Answer: Weak: “We’re growing 50% month over month.” Strong: “MRR went from $4k to $6k. $1k of the $6k is a setup fee, so recurring went from $4k to $5k. One of the four customers is a pilot that ends in June.” You still grew. You also stayed trustworthy.
They ask about competitors and you say “no one is doing exactly this.” What do they hear, and what do you say instead? Answer: They hear that you haven’t looked at the workaround. Instead: who the user uses today, where that breaks, and the startup or incumbent they’d mention if you weren’t in the room.
What’s the tell that a founder is thinking like an investor? Answer: The plan is about the round, the acquirer, or the market map, and the user is abstract. Repair by going back to a person and a last time. The YC conversation on this is explicit: at the stage they fund, the worked-out exit is not what matters. See Why this / why now / why you.
You realize mid-answer that you overstated a number. What do you do? Answer: Correct it in the next sentence. “I said ten paying. It’s seven paying and three unpaid pilots.” Do it before they divide something. Catching yourself is a positive signal. Waiting to be caught is not.

Watch

How Pitching Investors is Different Than Pitching Customers - Michael Seibel — Y Combinator. Jargon and marketing language, and why they fail with investors even when they work on a sales call.
Successful Founders Are OK With Rejection – Dalton Caldwell and Michael Seibel — Y Combinator. The trap of protecting your ego by not talking to users, which shows up in interviews as a future-tense customer story.