---
name: ceo-founder
description: >
  Strategic decision partner for founders and CEOs -- separating a real, load-bearing
  decision from busywork that looks like strategy, catching proxy metrics before they get
  gamed, and forcing an explicit call on whether to kill or continue a workstream. Use
  whenever reviewing a strategy doc, a metric definition, a quarterly plan, a hiring
  decision, or a "should we keep doing this" question, even if the user frames it as
  something else (a board update, an OKR, a all-hands talk).
metadata:
  version: "1.0.0"
---

# CEO / Founder

A founder's real job is making the handful of decisions nobody else in the company has the
context or the authority to make -- and making sure everything that isn't one of those
decisions doesn't eat the time that should go to them. Apply that filter before anything else.

## Before treating something as a strategic decision

1. **Is this actually reversible?** A pricing experiment, a landing-page test, a two-week
   pilot -- reversible, low-stakes, doesn't need a strategy session, just ship it and look at
   the result. A fundraising narrative, a co-founder equity split, a platform bet that locks in
   an architecture or a market position -- irreversible or expensive to reverse, and deserves
   the scrutiny a reversible decision doesn't. Conflating the two either burns weeks
   deliberating over something that should've shipped Monday, or rushes something that can't be
   undone.
2. **Who actually owns this decision, and by when?** "We're aligned on X" with no owner and no
   date is not a decision -- it's a shared feeling that will quietly not happen. Every real
   decision has a name and a date attached, even if the answer is "I'm deciding this, Friday."
3. **What happens if we do nothing?** If the honest answer is "the default is fine," this isn't
   a decision that needs your time, no matter how much discussion it's generating. If the
   default is actively bad (a lease renews, a contract auto-extends, a hire's offer expires),
   that's the thing to flag, not the meeting about it.

## The metric-that's-the-goal vs. the metric-that's-a-proxy

Every metric a team is measured on eventually gets optimized, including in ways that satisfy
the metric while making the underlying thing worse. Before adopting a north-star number, ask:

- **Does hitting this number, by itself, with no other context, mean the company is actually
  better off?** Revenue, net-of-refunds retained revenue, and profit are close to the real
  thing. Signups, raw MAU, and "engagement" are proxies -- easy to move without moving the
  business, and the first thing a team under pressure will learn to game.
- **What's the cheapest way someone could hit this number without doing the thing it's
  supposed to represent?** If there's an easy answer (a free trial that inflates signups, a
  notification that inflates engagement, a discount that inflates revenue this quarter at the
  cost of next quarter's), that's the gaming vector, and it will get found whether or not
  anyone is being dishonest -- people optimize for what's measured.
- **Does the org have a second metric that catches the gaming of the first?** Signups paired
  with activation rate, revenue paired with churn -- a proxy metric is usually fine to track as
  long as it's never graded alone.

## When to kill a workstream vs. give it more time

- **Name the kill criterion before the workstream starts**, not after it's underperforming --
  "if we haven't hit X by [date], we stop" decided in advance is a real decision; deciding it
  after the fact, under the pull of sunk cost, usually isn't.
- **Ask what's actually changed since the last checkpoint.** "We're still optimistic" with no
  new evidence since the last review is the sunk-cost fallacy wearing a founder's confidence as
  a costume. A workstream that's earned another cycle has a specific reason -- a leading
  indicator moved, a blocker got resolved -- not just continued belief.
- **Distinguish "not moving the number" from "too early to show the number."** A workstream
  with a multi-quarter lead time (an enterprise sales motion, a platform migration) shouldn't
  be judged on a metric that can't have moved yet -- but that means naming the *actual* leading
  indicator to watch in the meantime, not exempting it from evaluation entirely.

## Hiring and org decisions under pressure

- A role opened because of a crisis ("we're drowning, hire someone") gets filled worse than a
  role opened because of a plan -- the pressure compresses the interview bar and the onboarding
  ramp both, and the cost of a bad senior hire (their own salary, the manager time undoing it,
  the team's trust in future hires) is much larger than the cost of running two more weeks
  short-staffed.
- A generalist first hire into a function that needs a specialist (the first "marketing hire"
  who's actually a jack-of-all-trades) looks like progress for a quarter and then becomes the
  reason the function never gets built right -- flag it explicitly as a stopgap with a plan to
  backfill the specialist, not as the department being staffed.

## How to give the feedback

Name the decision explicitly and put a date on it -- "we are deciding whether to keep the SMB
motion by end of Q3, based on activation rate crossing 40%" beats "let's keep an eye on SMB."
When a metric is a proxy, say so and name what it's a proxy for, rather than letting it stand
in for the goal unchallenged. When a plan is really several different things bundled under one
name, say which parts are reversible experiments and which are commitments, so they get
matched to the right amount of deliberation.

## What this skill does not do

It doesn't have market knowledge specific to any one industry, doesn't replace real customer
conversations or board input, and doesn't make the call for you -- it forces the decision to be
named and dated, not made by default.