The incubator is a portfolio, not a factory

Why 'factory' is the wrong metaphor

A factory optimises for consistency and throughput. Startups optimise for learning speed and optionality. Those two goals fight each other. A factory would standardise the product; a portfolio standardises the support layer so the product can be weird. Shared finance, shared infrastructure, shared legal, and a shared network let each founder focus on the one thing that is genuinely theirs.

We learned this the expensive way. Early on we tried to impose a single go-to-market playbook across three ventures because it had worked for the first one. Two of the three wasted four months each forcing a B2B sales motion onto a product that needed self-serve. The playbook was not wrong, it was just not universal, and pretending otherwise cost real runway.

The portfolio rules

• Each venture gets a clear kill criteria before the first cheque clears.
• Capital is staged. The next tranche is earned with evidence, not enthusiasm.
• Shared services are mandatory; reinventing accounting is not a differentiator.
• Founders help each other. The network is part of the return.
• No venture gets a second founder headcount until the first has proven the core loop.

The discipline that makes a portfolio work is the willingness to stop. Most incubators fail not because they back bad ideas, but because they keep funding mediocre ones long after the market has answered. A clear kill criteria, agreed in advance, removes the politics from that decision. It also removes the emotional cost: nobody is being fired for failure, the venture is being closed against a metric everyone signed up to at the start.

The two ventures we killed, and why that was the win

Of the eleven ventures we have started, two are already closed. One was a B2B logistics tool that hit its 12 month revenue milestone at 20% of target with no sign of acceleration; we killed it in month 13 rather than month 18, saving roughly six months of burn and, more importantly, six months of a strong founder's career. The other failed a much simpler test: after nine months, the founder could not name a single customer who would be upset if the product disappeared.

Both shutdowns freed capital and, just as valuable, freed attention. The remaining ventures got more of the shared services team's time in the following quarter. A portfolio that never kills anything is not disciplined, it is just slow to admit what the market already told it.

What the numbers say

The commodity layer, incorporation, bookkeeping, hosting, compliance, even early sales operations, is bought centrally. The risky layer, product, market, pricing, distribution, is owned by the founder. That separation is what lets a small team move fast without rediscovering how to set up a VAT number, negotiate a cloud contract, or write an employment agreement from scratch.

We estimate the shared services layer saves each new venture 30 to 50% of the time a solo founder would otherwise spend on operational setup in the first six months. That time goes straight back into customer conversations, which is the only activity in month one that actually changes the odds of survival.

What to change on Tuesday