Every operating part in your building was decided by somebody. The schedule, the review form, the receiving procedure, what a variance costs, what a cast member is allowed to fix without asking, how a pre-shift runs, what your ordering pattern is, what a Guest gets when something goes wrong.
Somebody decided each one. The only question in your operation that matters is whether that somebody was you.
This is [Designed Architecture] — an operation whose parts were decided from inside it, by the person who can see the whole thing, against a read of what this building actually does. It is the opposite of an operation assembled out of parts that arrived from somewhere else, and it is the only asset you hold that cannot be purchased, copied, or resold. Your menu can be on somebody’s board in a season. Your room cost what it cost and can be beaten by a better room. The decision about how your building runs is yours alone, and every part you buy in that category is a piece of it you hand over.
Here is how you do the work.
Two Instruments, And You Need Both
Before any of the steps, two things have to exist, because every step below runs on one or the other. Most operators have neither and try to run the work anyway, which is why it stalls.
The read. The read is what you see when you look at your own operation without deciding anything. Standing in the room at the hour that matters, watching the work happen, and taking in what is actually there rather than what the report says or what you intended. Not the P&L, which is a month late. Not the satisfaction score, which is an average of strangers. What is in front of you, right now, on this stage, with these people, at this volume.
The read is a discipline and it is learnable, and the thing that makes it hard is that you built the operation, so you are looking at your own decisions and your eye slides over them. The test of a read is whether it produced something you did not already believe. If you walked out with the conclusion you walked in with, you did not read anything.
The question. The question runs in two directions and both are required. Forward: what do we need to do to innovate our ability to build a compounding business. Current: what are we doing that is inhibiting us from building a business that compounds, that we should just stop immediately.
Forward finds what is not there. Current finds what is there and taxing you. Neither one alone gets you a whole read, because an operation can be missing the thing that would compound and carrying six things that prevent it, and those are separate lists with separate work. There is no third answer to either direction. A part that is not compounding is contracting, which is what [No Static Achievement] rules.
Two instruments. One tells you what is true. The other tells you what to do about it.
Phase One — What You Are Building
Hold it as one structure. Every part of your operation touches every other part. The schedule decides what the kitchen can produce. What the kitchen produces decides what a cast member can promise a Guest. What a cast member can promise decides what a Guest comes back for. Your review form decides which of those a cast member believes you care about. Rule one area in isolation and the untouched areas overrule you by Friday. Before anything else, hold the operation as one structure rather than a set of departments.
Name what you have been chosen for, and name the load. Guests choose you for something. Cast members stay for something. The two are connected, and if you cannot say what either one is in a plain sentence, that is the first piece of work. Then name the load — what your operation has to carry to keep delivering that thing on a Tuesday in February with two people out. The load is the real number: volume at the hour it actually arrives, the cast you actually have, the equipment you actually own, the room you actually rent. Design against the load, not against the average.
List the areas that decide how the operation runs. The areas are where decisions live. The schedule. Position descriptions and what authority they carry. The review form. Pre-shift. Receiving and ordering. Variance policy and who can call one. Training and the development path. The menu and its mix. Vendors and terms. Recovery — what happens when something goes wrong in front of a Guest. Whatever technology decides something for you. Your own calendar.
That list is your operation. Not the org chart. Every one of those areas holds decisions that are either yours or somebody else’s, and you cannot know which until you look.
Read where you are standing. Now run the read against the areas on the list. Not to sort them. Not to grade them. Just to see what is there. The sorting comes next, and it comes from what the read produced rather than from a category you brought with you.
Phase Two — What Is In The Building Now
This phase produces one thing: a sorted list of what you are actually running. It takes longer than you want and it is the phase operators skip.
Walk the areas and inventory what is in them. Every manual, form, template, position description, procedure, program, platform, subscription, and standing practice. Write it down. Include the things nobody wrote down — the way it has always been done that exists only in three people’s heads. Especially those.
Name who decided each one. Not who approved it. Who decided. You will find four kinds: you decided it, a person in your building decided it, you bought it, or nobody alive can say. The fourth kind is the biggest pile in most operations and it is the one that tells you the most.
Ask what each one was built to get around. Every part exists to get around something. A step in a procedure, a cap on somebody’s authority, a line on a form — each was a response to a limitation. Ask what it was, and ask whether it is still true here. When the answer is that nobody knows, that is [Constraint Inheritance], and you are running a workaround for a problem you may not have.
Run the question both ways on each area. Forward, then current. With the people who do that work, in the room, out loud. What do we need here to build something that compounds, and what are we doing here that is inhibiting that. Write down what comes back and do not rule on any of it in the room. Ruling in the room stops the answers.
Sort into three piles. Carries load — the part is doing real work for this operation and can say what work. Taxes the structure — the part costs you time, money, capability, or authority and returns less than it costs. Cannot say what it was built for — nobody can name the reason, which means it goes into the workshop rather than the trash, because some of those turn out to be load-bearing once somebody finally asks.
Three piles. That is the output of the phase, and Phase Three consumes it.
Phase Three — Rule It
Find your line. Before you rule anything, name the line you will not cross in this operation — the minimum a Guest gets, the minimum a cast member gets, the minimum you will accept from yourself. That line is what every ruling gets measured against. Without it, ruling turns into preference, and preference does not survive a hard Friday.
Sequence the areas. You cannot rule everything at once and you should not try. Sequence by two things: which area is taxing you hardest, and which area’s ruling unlocks the next one. Position descriptions usually come before the review form, because you cannot measure authority nobody has. The schedule usually comes before training, because you cannot develop people who are never on with the person who can develop them. Order matters more than speed.
Rule the area. Take the area at the front of the sequence and decide it — from the read, against the load, measured against the line. Say what it is. Say what it is not. Say who can decide what inside it. A ruling that does not name authority is not a ruling.
Rebuild everything that contradicts it, before you announce it. This is the step that decides whether the ruling holds. When you rule an area, go find every part of the operation that says something different — the form, the checklist, the position description, the system configuration, the sentence in the manual, the thing you said in a pre-shift six months ago — and rebuild it to match. Then announce.
Announce first and the old parts overrule you by Friday, because whatever is in somebody’s hand at seven-forty on a Friday is what the operation runs. That is [Default Gravity], and the only thing that stops it is rebuilding before you speak.
Run it across the five parties. A ruling has to hold for the Guest, the cast, the operator, the vendor, and the community. If it works for the Guest and costs the cast, it will not survive. If it works for you and costs the Guest, it should not. This is the [Five-Actor Test] and it is what separates a ruling from a good idea.
Install it through the people doing the work. A ruling installs in a person, not in a document. The document records it. The install happens when the person who does the work has the authority the ruling gives him and uses it once without asking. Until that happens, you do not have a ruling, you have a memo.
Build the load path and verify it. Name the chain: this ruling holds because this part carries it, which holds because this person has this authority, which is measured on this form, which is reviewed on this cadence. Then verify it. Go stand in the room at the hour that matters and watch whether the chain carries. If any link is you personally — your reminder, your override, your Tuesday — the part is not in the structure. It is sitting on it, and it will come off the moment you are not there.
Then take the next area in the sequence.
The Five Standing Rules
These run permanently, underneath everything above.
Design instead of repair. Repair returns the operation to how it was before it broke. Design changes what it is. Most operators spend their whole tenure in repair and call it running a restaurant, and the operation is the same operation ten years later with better repair habits. That is [Repair Work], and the identity that forms around it is [Repairman Syndrome].
Buy facts and capabilities. Never buy a ruling. Hold temps, code, wage law, tax treatment — buy them, your operation does not get a vote. Capability you cannot build — a system that does something you cannot do by hand, an accountant who knows more than you — buy that too. What you never buy is a decision about how your operation runs. The moment a purchase decides something for you, you have handed over the only asset nobody can take.
In an area you do not own, buy the capability and then rule it. You are not going to build a point-of-sale or a payroll system. Buy the tool. Then go through every setting in it and rule each one against your read, and where the tool will not do what your ruling requires, that is a known cost you accepted rather than a decision it made for you. Most operators install defaults and then live inside a configuration somebody else chose. That is [Configuration Arbitrage].
Vet the source before money moves. Before you buy anything in this category, ask how many other operations are running it. Ask what it was built against. Ask whether it would function identically in a restaurant across the country with different Guests, a different cast, and a different room. If yes, you are buying the thinnest version of the work, and the vendor will have described that as best practice.
Rebuild on contact, and re-read on a cadence. When you touch a part of the operation for any reason, rebuild it rather than patching it. And put the read on a schedule — the load moves, the cast changes, the Guests change, and a ruling made against last year’s load is a default now. The work does not end, because there is no state in which an operation holds still.
What It Costs And What It Builds
The cost is honest and I am not going to dress it up. This is slower than ninety-nine dollars. Phase Two alone will take you longer than buying a manual, and it produces no document you can hold up. Some of your rulings will be wrong and you will rebuild them. And nobody outside your building will notice for a long while.
What it builds is the only thing that compounds.
A cast that gets more capable every period, because capability is built by people making decisions inside a standard the building owns. Guests who return for something specific, because there is something specific. Vendors who treat you as particular, because you are not running the same pattern as everybody else on the route. Margin that comes from having been chosen rather than from trimming. And an operation that needs you less every year, because the load is carried by the structure rather than by you every Tuesday.
That is [Meaningfully Differentiated Value], and it has exactly one source. Somebody standing inside the building, reading it, asking the question in both directions, and ruling. There is no other way in and no way to buy it.
The full sixteen-step process, written out: The Designed Architecture
To understand what not to do, as it exists in the wild, go to Hacksterism.
Digging Deeper
Every term in my framework lives in the Knowledge Base: kb.jeffreysummers.com
Terms used in this piece: [Designed Architecture], [Default Architecture], [Default Gravity], [Constraint Inheritance], [Configuration Arbitrage], [Five-Actor Test], [Meaningfully Differentiated Value], [No Static Achievement], [Repair Work], [Repairman Syndrome], [Question Dumb Shit], [Ordinary By Design]
The arbitrage prosecuted in the wild: Hacksterism
