Open a Second Restaurant Every 18 Months
Rewrite your operating systems on a cadence so a forced pivot is just another rewrite.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 70%
The mechanism treats internal reinvention as a scheduled launch rather than an occasional initiative. On a fixed cadence — roughly every eighteen months — the systems inside the business are rewritten as thoroughly as if you were opening a new location, and the rewriting is done by the people who run those systems, from the newest hire to the twenty-year veteran. Alongside that, you run deliberately extreme experiments away from the core floor: pop-ups, staff events, one-off concepts pushed further than commercially sensible, which act as the Formula 1 car to your production sedan. Every experiment is required to return something to the core product — a menu item, a system, a revised belief about the customer. The compounding output is not the experiments themselves; it is an organisation for which a forced overnight pivot is simply the next rewrite.
Origin
Mark Canlis explained on Acquired's Adapting Episode 1 why a 70-year-old fine-dining restaurant could open three new concepts in a week: Canlis had, in effect, been opening a second restaurant every eighteen months for years through system rewrites, parking-lot pop-ups and staff-party experiments.
Core principles
- 01There is probably a better way to do it than the way you are doing it.
- 02Relevance is earned continuously; it is not a state you reach.
- 03Push the extreme version of an idea — the Formula 1 car — and let it inform the everyday product.
- 04The people closest to a system should rewrite it, whether they arrived last month or twenty years ago.
- 05Success is what makes a company stop rewriting itself.
- 06Not everything old is obsolete; some systems still have useful life left.
How to run it
- 1
Set the rewrite cadence
Choose an interval — around eighteen months works for a business with annual seasonality — and commit to rewriting a meaningful set of internal systems each cycle. Treat the cycle as seriously as opening a new site.
Pro tip Naming it out loud ('we open a second restaurant every eighteen months') makes it a commitment rather than an aspiration.
Watch out A cadence with no named owner and no launch date degrades into a suggestion box within one cycle.
- 2
Let the operators do the rewriting
Assign each system to the people who actually run it, deliberately mixing new hires with long-tenured staff. New arrivals see what is arbitrary; veterans know what is load-bearing.
Pro tip Pair one of each on every rewrite so the arbitrary gets cut and the load-bearing survives.
Watch out A rewrite designed only by leadership produces a document, not a changed system.
- 3
Run the Formula 1 version
Each cycle, build one experiment pushed far past what the core business would sensibly do, staged away from the main product. The point is to see what happens at the extreme, not to make the experiment profitable.
Pro tip Physically separate it — the car park, an off-site, a shipping container — so the core operation is never at risk.
Watch out If the experiment has to justify itself on margin, it will be watered down into something you already knew the answer to.
- 4
Apply the willingly-attend test
For internal events used as test beds, hold the standard that your team would choose to come even with no connection to the company. It is a cheap, honest quality bar for anything you build.
Pro tip Explicitly tell people it is a real day off with no expectation to attend; the turnout is then real signal.
Watch out Mandatory attendance destroys the signal and turns the test bed into a morale tax.
- 5
Route every experiment back into the core
Require each project to return something concrete: an item on the menu, a system that stuck, or a revised belief about how you serve people. This is what converts play into capability.
Pro tip Close each experiment with a written note on what it changed about what you believe, not just what it earned.
Watch out Experiments that never feed back become a separate hobby business competing for the same attention.
- 6
Audit for the two failure modes
Each cycle, check both directions: which systems have gone unquestioned because they are working, and which changes would tear down something that still has useful life. Rewriting is not the same as churning.
Pro tip Ask of any system you are about to remove: how many more good years does this actually have?
Watch out It gets easy to devalue continuous improvement precisely when the business is performing well — that is when calcification starts.
In the wild
Canlis tested ideas on staff parties, held to the standard that staff would come even with no obligation to the restaurant. One party involved covering the concrete car park in sod, a pig roast, volleyball, and an above-ground pool bought on Amazon for about two thousand dollars because someone said the beach needed water. It worked, so the team threw the staff party for the city: a multi-storey hangout above the car park with a 24-foot pool, tiki huts, a pizza operation, a DJ on top of a shipping container and roasting pigs.
→ Around 1,200 people a night attended, with hour-long queues, against an expectation of a few hundred — and elements of the format fed back into the restaurant's own menu and beliefs about hospitality.
Long before a forced pivot, the fine-dining restaurant ran a Shake Shack pop-up in its own parking lot — the kind of project a Michelin-adjacent kitchen would normally consider beneath it. Treated as the extreme version rather than a stunt, it taught the kitchen and floor how to serve volume outside the dining room, and confirmed that guests would come to a fine-dining address for something informal. Years later, when the dining room had to close overnight, the drive-through burger concept was not a leap into the unknown — it was a format the building and the team had already rehearsed.
→ The organisation entered a genuine crisis with an existing, tested playbook for high-volume informal service on its own premises.
Common mistakes
Only experimenting when you are in trouble
Reinvention attempted for the first time under pressure is slow, contested and usually wrong. The cadence exists so the muscle is already built when you need it.
Tearing down systems that still have life
Rewriting on a cadence can tip into churn, removing systems with years of useful life left. Ask what remains good about a system before replacing it on schedule.
Keeping the experiments off the core product
An experiment that never returns anything to the main business is entertainment. Require a concrete transfer — an item, a system, or a changed belief — from every project.
Is it for you?
Best for
Owner-led businesses with a stable core product and a long time horizon that want the capacity to adapt fast without waiting for a crisis to force it.
Not ideal for
Pre-product-market-fit startups still searching for a first working system, or organisations where change is already constant and the problem is focus, not calcification.
From the transcript
“one of the reasons that I think this transition has been possible for us is that we practice this kind of thing”
“we open a second restaurant about every 18 months I would say that's how often the systems inside of Canlis are changing”
“it's kind of like the formula 1 version of putting out a sedan it's like why don't we go push this as far as we…”
From the episode
Adapting Episode 1: Canlis