Should we open a second site?

Cerno is a private AI boardroom that runs consequential decisions through a structured process — advisor challenge, a recorded vote, risk and confidence scores — and returns a brief like the one below. This example is illustrative; your own would use your numbers.

How this brief was reached

Several advisors formed independent positions first, a Devil's Advocate challenged the leading view, and the advisors voted — with the dissent kept on the record rather than smoothed over. The Chairperson then synthesised the recommendation, scored risk and confidence, and separated what's known from what's assumed. The 3–2 vote and the mid-high risk score are the honest signal: this is a conditional yes, not a confident one.

What made it conditional rather than negative was that all three conditions are cheap. A demand test costs a fraction of a fit-out. A break clause costs a negotiation. Hiring before spending costs time, not money. None of them stops the expansion; each of them converts an assumption into a fact before the irreversible spending starts.

The Risk Officer's dissent was adopted directly: the weekly bookings tripwire exists because the pre-mortem's top failure mode had no early-warning signal at all. The CFO's dissent was not adopted, but it set the shape — a staged commitment is precisely what keeps the runway from thinning past the point of recovery.

What the board weighed

A full site proves demand here, not there. 95% utilisation is a strong fact and it's a fact about one location. The customers who fill it walk past that window, live on those streets, or found you through something local. Whether the same offer pulls three miles away is a different question, and the board declined to treat the first as evidence for the second.

£180k of enquiries is not £180k. The brief lists a 30% conversion rate as an assumption, not a known fact, and that placement is deliberate. Expansion cases are usually built on a pipeline number multiplied by a conversion rate borrowed from the existing site — where you have a reputation, reviews and word of mouth that a new site starts without.

The overrun is the base rate, not bad luck. Comparable expansions run over on both time and cost with enough regularity that reference-class forecasting treats it as the expected case. A plan with no allowance for it isn't optimistic; it's using the wrong number. The CFO's five-month runway assumes the fit-out goes to schedule, which is the one thing fit-outs reliably don't do.

Nobody is running it. No site lead has been identified, which means the implicit plan is that the owner covers both sites until someone is found. That's the arrangement that quietly damages the first site — the profitable one — while everyone's attention is on the new one. Making the hire a gate protects the business you already have.

If your situation differs

If you've opened a second site before, the biggest assumptions are answered by your own record, the base rate is yours rather than borrowed, and this becomes a much more confident yes.

If the second site serves the same customers — an overflow unit, extra storage, a workshop behind the shop — the demand question largely disappears and this is really a lease decision about space, not an expansion decision about a market.

If the site lead already exists internally and wants it, one of the three conditions is met before you start, and the risk score drops accordingly.

The frameworks behind it

This decision leans on a pre-mortem (what would make the second site fail?) and reference-class forecasting (how do comparable expansions actually turn out?). It's also close to a one-way door — hence the emphasis on a break clause and a staged commitment.