Venture building is usually pitched as a forward motion: find the pain, test the riskiest assumption, ship the smallest product that sells, make it investable, turn pilots into paying customers. That arc is real. Every venture I have carried moved through those five gates, and nothing scaled until the evidence at the gate before it was real.
But a portfolio is not a single venture, and a portfolio has a sixth gate that the conference-stage version of venture building skips: the decision to stop, taken early enough that stopping is cheap.
Three calls, not one
At every gate there are three honest calls: persevere, pivot or retire. I have made all three across seven ventures, and the pattern that surprised me is which call created the most value. The persevere calls produced the metrics that go on a portfolio page. The pivot calls produced the interesting stories. The retire calls produced the money: budget and team-months that did not get burned proving something the evidence had already answered.
Retire calls are rare for emotional and institutional reasons, rarely analytical ones. By the time a venture has a name, a deck and a sponsor, everyone in the room is paid, promoted or emotionally rewarded for it continuing. Nobody's incentives point at the exit. Which is exactly why the exit has to be designed in before anyone is attached.
What a cheap exit requires
Retiring cheaply takes infrastructure, the kind you put in place when optimism is highest:
- Stop criteria written at the validate gate. Before the first experiment runs, write down what result would end the venture. If you cannot name it, you are not validating, you are decorating a decision already taken.
- An evidence ledger, not a highlights reel. Every experiment logged with its cost, its result and what it changed. The ledger is what lets a room full of sponsors agree the threshold was crossed, without relitigating feelings.
- The cheapest experiment that can produce a signal. Expensive experiments create sunk cost, and sunk cost is the enemy of the retire call. If a desk study answers it, do not run a pilot. If a paid POC answers it, do not build the platform.
- Sequenced markets, not simultaneous ones. Entering one market at a time keeps the blast radius of a wrong bet small. The market map should earn its pins one by one.
- A separate word for stopped. Retired is not failed. A venture retired on evidence is a successful experiment with a negative result, and the team that ran it should be reused, not punished. If retiring is career-fatal, nobody will ever surface the evidence for it.
The uncomfortable accounting
Here is the test I hold myself to: if you cannot point at a venture you stopped, you are not running a portfolio discipline, you are running a momentum machine. The scoreboard of a venture builder is not only the ventures that reached market. It is also the capital that did not get spent past the point where the answer was known.
The five forward gates make you look good. The sixth gate makes the portfolio work.