[Decision Architecture]

Definition #

[Decision Architecture] is the operating architecture through which every operator decision gets made — the design of the thinking that produces the decision, the process the operator runs the decision through, and the record the decision leaves behind. It is the discipline-layer architectural child of [Restaurant Architecture] that governs how decisions get made, how they get pressure-tested, and how they compound into operating judgment over time.

[Decision Architecture] sits at the discipline layer of the parent-architecture stack, alongside [Read Architecture]. It runs in one of two operating forms — Road 1 informal-and-reactive or Road 2 formal-and-proactive — and every decision the operation makes inherits the form the architecture is running in.

The load-bearing claim is that decisions are architectural, not situational. Every operator makes decisions; the question is whether the thinking behind the decisions is compounding in the right direction or contracting in the wrong. [Decision Architecture] names the difference.

Mechanism #

Every operator makes decisions every day. Menu decisions, capital decisions, staffing decisions, positioning decisions, pricing decisions, vendor decisions, growth decisions. Decisions are the operating output of the operator’s job. The question the framework asks is not whether the operator is making decisions — the operator is making decisions constantly — but whether the architecture producing those decisions is designed or defaulted.

Road 1 form — informal, reactive, invisible. On Road 1, [Decision Architecture] is not deliberately designed. Decisions get made at the moment the pressure hits, on whatever inputs happened to reach the operator, filtered through the operator’s own biases without those biases being named or prosecuted, and the decision leaves no legible record. The operator cannot reconstruct why the decision was made, cannot audit whether the decision worked, and cannot compound operating intelligence across decisions because no decision leaves an inspectable trail. Bias goes unchecked. Intelligence evaporates. The operation pays [Lost Opportunity Tax] — the cost of decisions the operator did not make because the architecture never surfaced them.

Road 2 form — formal, proactive, legible. On Road 2, [Decision Architecture] is deliberately designed. Decisions run through a designed process. Bias is named and prosecuted. Inputs are aggregated deliberately. Stress-testing is formal. Decisions leave a legible record — a decision log the operator and successors can audit. Operating intelligence compounds across decisions because every decision feeds the architecture’s evidence bank, and the evidence bank feeds the next decision. The architecture produces decisions the operator can defend, adjust, and compound against.

Road 2 superiority — explicit. The framework holds that Road 2 [Decision Architecture] is superior to Road 1 [Decision Architecture]. Road 2 produces decisions that compound. Road 1 produces decisions that evaporate. Road 1 operations that run their decision-making as coherently as Road 1 allows are producing legitimate decisions at Road 1 physics — real judgment, real market response, real intuition applied to real signal — but the ceiling on Road 1 [Decision Architecture] is that the architecture cannot compound. Every decision starts from zero. Every mistake gets forgotten. Every insight is available only to the operator running it, only in the moment. Road 2 raises the ceiling by making decisions legible, auditable, and inspectable across time.

The seven layers of the architecture. Road 2 [Decision Architecture] runs decisions through seven layers.

Layer Minus 1 — Standing Examination. The operator runs an ongoing examination discipline that surfaces decisions before they become urgent. Standing questions get read against the operation’s Read on cadence. Decisions surface into the architecture through the standing examination, not because a pressure event forced them.

Layer 0 — Declaration. Once a decision is surfaced, the operator declares the decision explicitly. The declaration names what is being decided, what the decision is not, and where the decision sits in the operation. Declaration converts an ambient concern into a named decision the architecture can process.

Layer 1 — Anchor / [MDV] / [The Operator’s Lens]. The decision gets anchored against the operation’s Minimum Defensible Version — the floor the decision must clear — and against the operator’s own operating lens. Anchoring is what keeps decisions grounded in the operation the operator is actually running, not the operation the operator wishes they were running.

Layer 2 — Inputs / [The Ten Sight Targets]. The architecture aggregates inputs from every layer the decision touches — Guest signal, cast signal, financial signal, positioning signal, capital signal, community signal, competitive signal, vendor signal, seasonality signal, operator-state signal. [It’s The Vision Thing] and [The Ten Sight Targets] name the input layers the architecture reads against. Input aggregation is what makes the decision informed by the whole operation rather than by whichever input reached the operator loudest.

Layer 3 — Aggregation / Stress-Test. The aggregated inputs get stress-tested through the architecture’s stress-test toolset — the operator’s own bias prosecution, [The Five Stakeholder Read], [The Five Questions], structural-scale reads, decoy-option reads, and the pause discipline that keeps reactive decisions from being made under pressure. The stress-test layer is where Road 1 defaults get caught before they become the decision.

Layer 4 — Gate / [Burn The Boats] / Exit A/B/C. The decision reaches the gate. Exit A commits fully — [Burn The Boats], the decision goes forward, the operation reorients around it. Exit B pauses the decision — more input needed, more stress-test needed, or the decision is not ready. Exit C rejects the decision — the architecture has surfaced that the decision is wrong, and the wrong decision does not enter the operation. The gate is what enforces that decisions leave the architecture as decisions, not as ambient intent.

Layer 4B — [Operator R&D] Pilot. Some decisions enter the architecture as pilots rather than full commits — a bounded test, a defined evaluation window, a decision that will re-enter the architecture at the end of the pilot for full-commit review. The pilot layer is what allows the architecture to test decisions the operator cannot fully commit to without evidence, without defaulting the whole decision to Road 1 gut.

The Read Log — the record layer. Every decision that passes through [Decision Architecture] leaves a record in the operation’s decision log. The log names what was decided, what the inputs were, what the stress-test produced, what exit was taken, and what evidence the decision produced downstream. The log is what makes the architecture compound — every next decision reads against every prior decision, and operating intelligence grows across time rather than evaporating with each shift.

The Pause Principle. Reactive decisions are Road 1 by definition. The architecture holds a pause discipline — decisions do not get made in the pressure moment; decisions get named, moved into the architecture’s process, and made against the architecture’s stress-test. The pause is not delay; the pause is what separates the decision from the pressure.

Failure modes the architecture is built against. [Substrate Seduction] pulls the operator to make decisions about the industry substrate instead of about the operation. [One-Man Band] pulls the operator to make every decision alone without stakeholder-read inputs. [Forest For The Trees] and [Trees For The Forest] pull the operator to lose either scale or granularity depending on which way the architecture is bending under pressure. [Vision Swampacolypse] pulls the operator to freeze on too many inputs. Every named failure mode is a Road 1 default the Road 2 architecture is designed to catch.

Coupling with adjacent architectural children. [Decision Architecture] couples with [Read Architecture] at the discipline layer — the Read produces the intelligence, the Decision processes the intelligence. Poor [Read Architecture] cannot support strong [Decision Architecture]; the architecture is only as strong as the read that feeds it. Both discipline-layer children must run coherently, in the same operating form, for the discipline layer of [Restaurant Architecture] to be coherent.

Load-Bearing Distinction #

Not any single decision. Any single decision can be well-made or badly-made regardless of the architecture. A well-made decision on Road 1 [Decision Architecture] is the operator getting lucky against a compressed input set. A well-made decision on Road 2 [Decision Architecture] is the architecture doing what it was designed to do. Over time, the difference between “operator got lucky” and “architecture worked” compounds.

Not decision-making frameworks or tools alone. Frameworks and tools — pro/con lists, decision matrices, cost-benefit analyses, decision trees — are inputs to [Decision Architecture], not [Decision Architecture] itself. A framework used inside a Road 1 architecture is still Road 1. A framework used inside a Road 2 architecture is one instrument in the stress-test layer. The architecture is the design of the thinking; the framework is a tool the thinking uses.

Not [Read Architecture]. [Read Architecture] is the sibling discipline-layer child. [Read Architecture] produces the intelligence; [Decision Architecture] processes the intelligence into decisions. Both are discipline-layer children of [Restaurant Architecture]. Both must run coherently for the discipline layer to hold.

Not [Architectural Read]. [Architectural Read] is the parent-architecture read discipline — the read the operator runs at the [Restaurant Architecture] layer to surface coupling failures across all architectural children. [Decision Architecture] is the specific discipline-layer child that governs how the operator’s decisions run inside the operation. [Architectural Read] operates one level up.

Not intuition or gut. Intuition without [Decision Architecture] is Road 1 by definition — pattern-recognition running on whatever inputs happened to reach the operator, unaudited, unrecorded, uncompounded. Intuition supported by [Decision Architecture] is judgment running through a designed process against a stress-tested input set. The framework does not reject intuition; the framework requires intuition to run inside an architecture that catches its biases and records its outputs.

Not [The Operator’s Read]. [The Operator’s Read] is the aggregate discipline the operator holds above the three thirds of the operator’s job — it integrates every read the operator runs. [Decision Architecture] is the architecture that processes [The Operator’s Read] into decisions. Related, coupled, but distinct — one is the aggregate perception discipline, the other is the design of how perception becomes decision.

Not a decision log alone. The Read Log is the record layer of [Decision Architecture], not the architecture itself. Operations that install a decision log without the seven-layer process running upstream produce a record of Road 1 decisions written down. The record is downstream of the architecture; the architecture is upstream of the record.

The term is load-bearing because it names the discipline no single decision holds: that decision quality is architectural, that the architecture running underneath every decision determines whether operating intelligence compounds or evaporates, and that Road 1 [Decision Architecture] is the industry default operators do not know they are running. Without [Decision Architecture] named, operators treat decision-making as personal skill and default the architecture. With it named, operators can install the architecture that makes decisions compound into judgment.

Diagnostic Tests #

Test One — The Reconstruction Test. Ask the operator to reconstruct the last three significant decisions — what was decided, what the inputs were, what stress-test the decision ran through, what evidence the decision produced downstream. If the operator cannot reconstruct the inputs and the stress-test, [Decision Architecture] is running Road 1 and the decision left no legible record. If the operator can name the inputs, the stress-test, and the downstream evidence, the architecture is running Road 2.

Test Two — The Bias Prosecution Test. Ask the operator to name the bias they were running when they made the last significant decision. If the answer is “I do not think I had a bias” or “I try to stay objective,” the architecture is running Road 1 and bias is going unprosecuted. If the operator can name the specific bias that ran against the decision and describe how the architecture prosecuted it, the architecture is running Road 2 stress-test discipline.

Test Three — The Pause Test. Ask the operator to name the last decision they paused on. If every decision the operator makes gets made in the moment the pressure hits, [Decision Architecture] is running Road 1 reactive default. If decisions get named, moved into the process, and made against stress-test at a defined pace, the architecture is running Road 2 pause discipline.

Test Four — The Compounding Test. Ask the operator to describe how the last decision changed the next decision. If the answer is “it did not; each decision is its own thing,” the architecture is not compounding — Road 1 default. If the operator can name the evidence the last decision produced and how that evidence read into the next decision’s inputs, the architecture is compounding — Road 2.

Test Five — The Lost Opportunity Read. Ask the operator to name a decision they should have made but did not. If the answer is “I cannot think of one,” the architecture is not surfacing decisions — Layer Minus 1 standing examination is defaulted. Every operation is generating decisions that need to be made; the architecture that does not surface them is paying [Lost Opportunity Tax] the operator cannot see because the architecture never named the opportunities.

Test Six — The Exit Discipline Test. Ask the operator to name the last decision that took Exit C — the decision the architecture rejected. If the answer is “I do not remember rejecting a decision,” the gate is not working. Every architecture should be rejecting decisions the stress-test surfaces as wrong. Absence of Exit C usage is a fingerprint of Road 1 default — every named decision becomes the decision because the architecture has no rejection layer.

Test Seven — The Discipline-Layer Coupling Test. Read [Decision Architecture] against [Read Architecture]. Does the intelligence the Read produced actually feed the decisions the architecture is processing? Or are decisions being made on inputs the [Read Architecture] never captured? Discipline-layer coupling failures produce decisions that contradict the Read, or decisions the Read never had the chance to inform. Both fingerprints are the discipline-layer child running Road 1 while the sibling claims Road 2.

Family Position #

[Decision Architecture] is a discipline-layer architectural child of [Restaurant Architecture], sitting alongside [Read Architecture] at the discipline layer of the parent-architecture stack.

Perspective application. [Decision Architecture] is Perspective in action at the decision layer. Perspective without an architecture that processes it into decisions is Perspective that stays private and evaporates. Perspective with Road 2 [Decision Architecture] is Perspective producing decisions the operation can audit and compound against.

Product application. Every Product decision — menu design, GX design, offer design, pricing design, service-flow design — runs through [Decision Architecture]. On Road 2, Product decisions are stress-tested against Guest signal, cast capacity, financial physics, and positioning coherence before entering the operation. On Road 1, Product decisions get made on whichever input reached the operator loudest at the pressure moment. Product quality is downstream of [Decision Architecture] quality.

People application. Every People-side decision — hire, promote, terminate, schedule, develop — runs through [Decision Architecture]. On Road 2, People-side decisions are stress-tested against the cast’s parallel reads, against the operation’s [Cast Contract] form, against the reward structure that funds the decision, and against the operator’s own bias prosecution. On Road 1, People-side decisions are the most Road-1-defaulted decisions in the industry — made on gut, on proximity, on the one voice that reached the operator that shift.

Performance application. Performance-shaping decisions — scheduling changes, station changes, service-flow changes, mid-shift adjustments — run through [Decision Architecture] at operating tempo. Road 2 [Decision Architecture] holds even under operating pressure because the pause discipline is built in. Road 1 [Decision Architecture] collapses under operating pressure because pressure IS the trigger and reactive IS the default.

Profit application. Profit-shaping decisions — capital allocation, menu mix, daypart strategy, vendor negotiation, growth decisions — run through [Decision Architecture] at strategic tempo. Road 2 [Decision Architecture] surfaces profit decisions through Layer Minus 1 standing examination before pressure forces the decision. Road 1 [Decision Architecture] waits for the P&L to force the decision, and by then the decision is a rescue, not a design.

Cross-References To Locked IP #

Parent:

• [Restaurant Architecture] — the parent-architecture layer [Decision Architecture] sits inside as its discipline-layer child

• [Two Roads] — the operating-physics families [Decision Architecture] runs inside

Related:

• [Read Architecture] — the sibling discipline-layer child that produces the intelligence [Decision Architecture] processes

• [The Operator’s Read] — the aggregate perception discipline whose output [Decision Architecture] processes into decisions

• [Architectural Read] — the parent-architecture read discipline that surfaces coupling failures across every architectural child

• [The Outcomes Formula] — the naming discipline that anchors decisions to defined outcomes

• [The Operator Decision Tree] — the seven-layer decision-process instrument [Decision Architecture] runs on

• [It’s The Vision Thing] — the vision-anchoring discipline supporting Layer 2 input aggregation

• [The Ten Sight Targets] — the ten input layers Layer 2 aggregates across

• [The Pause Principle] — the pause discipline that separates decisions from pressure

• [Bias Prosecution] — the stress-test discipline that catches operator bias before it becomes the decision

• [The Five Stakeholder Read] — a stress-test instrument at Layer 3

• [The Five Questions] — a stress-test instrument at Layer 3

• [Structural Scale] — a stress-test discipline at Layer 3

• [Burn The Boats] — the commitment discipline at the Layer 4 gate

• [Operator R&D] — the pilot discipline at Layer 4B

• [The Read Log] — the record layer that makes [Decision Architecture] compound across time

• [MDV] — the anchor floor decisions must clear at Layer 1

• [Contract Architecture] — the contract-layer sibling whose form must run coherent with [Decision Architecture] form

• [Guest Architecture] — the output-side sibling on Road 2 whose live signals feed [Decision Architecture] inputs

• [Customer Architecture] — the output-side sibling on Road 1 whose signals feed [Decision Architecture] inputs

• [HE Architecture] — the operator-state sibling whose read feeds bias prosecution at Layer 3

• [Reward Structure Architecture] — the incentives-layer sibling whose signals feed People-side decisions

• [Culinary Architecture] — the composition-child whose decisions run through [Decision Architecture]

Opposing patterns:

• [The Two Roads Problem] — the specific failure mode of running mixed-form [Decision Architecture]

• [Lost Opportunity Tax] — the cost of decisions Road 1 [Decision Architecture] never surfaced

• [Substrate Seduction] — the failure mode where decisions get made about the industry substrate instead of about the operation

• [One-Man Band] — the failure mode where every decision runs on the operator alone without stakeholder-read inputs

• [Forest For The Trees] — the aggregation failure at Layer 3 where scale collapses and granularity dominates

• [Trees For The Forest] — the aggregation failure at Layer 3 where granularity collapses and scale dominates

• [Vision Swampacolypse] — the failure state at Layer 2 where too many inputs produce paralysis

• [Reactive Dangers] — the failure state where decisions get made in the pressure moment

• [Architectural Identity Collapse] — the failure mode where [Decision Architecture] absorbs [Restaurant Architecture]’s identity as “the operator’s decisions” and displaces the parent-architecture frame

• [Hacksterism] — the shortcut posture that claims decision discipline while defaulting the architecture

Why This Matters #

The industry has treated decision-making as a personal attribute of the operator for as long as the industry has existed. “He has good instincts.” “She makes tough calls.” Decision-making gets treated as talent, gut, or experience, and operators are told to develop it as a personal skill through years of trial and error.

The industry has been wrong. Decision quality is architectural. The operator’s individual skill runs inside an architecture that either compounds the operator’s intelligence or evaporates it. Treating decisions as personal skill defaults the architecture and consigns the operator to Road 1 physics — real judgment producing real decisions that leave no legible record, teach no successor, and compound no operating intelligence.

[Decision Architecture] names the physics. Without the term named, operators run Road 1 by default — reactive, invisible, uncompounded — and never see the ceiling their architecture is running under. With the term named, operators can install the seven-layer process that catches bias, aggregates inputs, stress-tests decisions, gates commitment, and records evidence. The architecture is what allows operating judgment to compound across decades rather than reset with each shift.

The framework holds that Road 2 [Decision Architecture] is superior to Road 1 [Decision Architecture]. Compound operating judgment, defensible decisions, successor-legibility, bias prosecution, and the ability to surface decisions before pressure forces them are all downstream physics only Road 2 produces. Road 1 [Decision Architecture] runs on Road 1 physics — real gut, real judgment, real intuition — but the ceiling is that every decision starts from zero and every mistake gets forgotten. Only architecture lifts the ceiling. Only Road 2 [Decision Architecture] produces the compounding.

The [Lost Opportunity Tax] is the specific cost the framework names for operations running Road 1 [Decision Architecture]. Every decision the architecture never surfaced is an opportunity the operation missed. Every decision made on unaudited bias is an outcome the operation paid for at higher cost than a stress-tested decision would have produced. Every decision made without record is intelligence the operation lost forever. Road 1 [Decision Architecture] is a tax the operator does not see because the architecture never surfaced the opportunities.

The framework needed the name. The name is decision-as-architectural-discipline, running through a seven-layer designed process, prosecuted for bias, gated for commitment, and recorded for compound learning.

Operating Consequence #

Install the seven-layer process and hold it as a discipline. The operator installs the seven-layer decision-process — Standing Examination, Declaration, Anchor, Inputs, Stress-Test, Gate, and Pilot — as the architecture every significant decision runs through. Small operating decisions can compress the layers, but strategic decisions run the full process. Compression is a design choice; skipping is a Road 1 default.

Prosecute bias explicitly on every decision. The operator names their own bias before the decision runs through the stress-test layer. The bias does not get hidden; the bias gets surfaced, prosecuted, and either accepted as informing the decision or refused as distorting the decision. Unprosecuted bias is Road 1 by definition, regardless of how sophisticated the surrounding process looks.

Hold the pause discipline. No significant decision gets made in the pressure moment. The operator names the decision, moves it into the architecture’s process, and produces the decision at the architecture’s tempo. Pressure is the trigger the architecture is built against; pause is the discipline that separates the decision from the trigger.

Log every decision that clears the gate. [The Read Log] runs alongside [Decision Architecture]. Every decision that clears Exit A or Exit C leaves a record — what was decided, what inputs, what stress-test, what exit, what downstream evidence. The log is not documentation; the log is the architecture’s compounding surface. Without the log, decisions do not compound.

Refuse the personal-skill framing. The operator refuses the frame that decisions are their personal skill. Decisions are the architecture’s output. Personal skill enters the architecture at every layer, but the architecture is what produces the decisions that compound. Personal-skill framing defaults the architecture and consigns the operation to Road 1 physics.

Read [Decision Architecture] against [Read Architecture] on cadence. The discipline-layer coupling — Read feeds Decision, Decision runs against Read — is where the discipline layer either holds or defaults. On cadence, the operator asks whether the Read the architecture produced actually informed the decisions the architecture processed, and whether decisions made without the Read are decisions the architecture should have surfaced.

Install Layer Minus 1 standing examination as its own operating discipline. Standing examination is the layer most operators default first because it is the least urgent. The operator runs standing questions against the operation on a defined cadence — weekly at operating tempo, monthly at strategic tempo — and surfaces decisions into the architecture before pressure forces them. Standing examination is what eliminates [Lost Opportunity Tax].

Run Exit C explicitly. The gate has three exits. Exit C — rejection — is the architecture’s refusal discipline. If Exit C never gets used, the gate is not working. The operator holds the discipline that rejecting a decision is the architecture doing what it was designed to do, not the operator failing to commit.

Refuse measurement-substitution at the decision layer. Numbers can be inputs at Layer 2, but numbers are not decisions. The operator refuses the frame that reads scorecard as decision-making. [Decision Architecture] is judgment discipline supported by measurement; measurement discipline replacing judgment is Road 1 default disguised as Road 2 rigor.

What Changes Tomorrow #

Tomorrow the operator runs the decision reconstruction. Name the last three significant decisions made this week. For each, write down what was decided, what the inputs were, what stress-test the decision ran through, what exit the decision took, and what downstream evidence the decision produced. If the operator cannot reconstruct the inputs and the stress-test for any of the three, [Decision Architecture] is running Road 1 for that decision, and the workshop debt is the process the architecture needs to run for the next decision.

Then the operator runs the bias read. For the next decision that surfaces into the architecture, before running any stress-test, the operator names the bias they are running against the decision. The bias gets written down. The decision gets stress-tested against the named bias explicitly, with the bias documented in the stress-test log. Bias prosecution installed once is bias prosecution the architecture can hold across decisions.

Then the operator installs the pause. For the next significant decision the operation surfaces, the operator refuses to make the decision in the moment the pressure hits. The decision gets named, moved into the architecture’s process at the architecture’s tempo, and produced against the seven-layer discipline. One paused decision is the architectural precedent; the pause becomes a discipline through repetition.

Then the operator installs Layer Minus 1 standing examination. The operator writes down three standing questions the operation should be reading against on cadence — operating-tempo questions weekly, strategic-tempo questions monthly. The standing questions run for the coming month. Every question that surfaces a decision converts the standing examination from concept to running architecture.

Then the operator schedules a discipline-layer coupling read against [Read Architecture]. Read the last three decisions the architecture processed. Did the Read the parallel-read architecture produced actually feed those decisions, or were the decisions made on inputs [Read Architecture] never captured? Every gap is a discipline-layer coupling failure the coming quarter closes.

The operator running the seven-layer process on strategic decisions for a quarter has installed the beginning of Road 2 [Decision Architecture]. The operator running standing examination, bias prosecution, pause discipline, and [The Read Log] for a year has installed [Decision Architecture] as its own operating discipline — coupled with [Read Architecture] at the discipline layer and reinforced by [Architectural Read] at the parent-architecture layer.

Every operator makes decisions. The operator’s job is to install the architecture that makes those decisions compound into operating judgment, refuse the personal-skill framing that lets Road 1 default win, and pay down the [Lost Opportunity Tax] the industry underwrites by convention.