Category

F01: Perspective

Definition
Every operator runs R&D. You have been running it since the day you opened.
Every menu change, every training adjustment, every hospitality pilot, every daypart decision, every service model tweak — that is R&D. The activity is not the problem. The problem is that nobody named it, nobody logged it, and nobody built a formal instrument to evaluate it. The process is invisible. And an invisible process cannot be examined, cannot be improved, and cannot be defended against the bias running inside it.
The independent restaurant operator runs the highest-stakes R&D in the industry. No safety net. No portfolio of locations to absorb a failed test. No corporate process to catch what [Informal Instrumentation] misses. The enterprise R&D leader has a named process, a budget line, a review committee, and a scorecard. You have feel, history, and a P&L that shows up thirty days after the decisions that produced it. The mandate for formal R&D thinking is stronger at your level, not weaker.
Gartner studied 53 enterprise R&D organizations and found that traditional selection processes institutionalize incrementalism — not because leadership lacks vision or courage, but because the instrument was built to produce incremental results. Balanced scorecards weighted on historical success guarantee the selection of what worked before. The process selects against transformation before the decision-maker ever enters the room. The independent operator’s version of this is more acute. At least the enterprise has a process it can examine and reform. You have an intuition running an invisible scorecard — built on past victories, weighted toward what cleared the threshold before, and structurally blind to any value the standard instrument cannot see.
You are not choosing incrementalism. [Informal Instrumentation] is selecting it. The distinction matters because one is a failure of courage and one is a failure of frame. You do not lack courage. You lack a legible instrument for the work you are actually trying to do.
What the Informal Process Looks Like
You run a test. A new menu item, a hospitality pilot, a lead on the stage, a pacing change during peak production. The test runs for a week, maybe two. At the end you check the numbers — did covers move, did food cost hold, did comp sales tick. The numbers do not move. The test dies.
What you never logged: the three tables that came back the following week specifically because of the connection the lead made. The two Guests who asked for the same cast member by name. The stage read on Friday that showed the cast running at a different energy level. The table that lingered thirty minutes past their check because the conversation was that good. That evidence existed. It evaporated because it was never logged, never named, never carried into the evaluation. [Informal Instrumentation] only counted what the historical scorecard was built to see. Everything else disappeared the moment the shift ended.
This is the mechanism behind [Static Decline]. The operator running an informal process against a historical scorecard is not choosing to stay still. The process is selecting stillness for him. Every Road 2 move that cannot clear a Road 1 instrument gets killed in evaluation before it ever proves itself. The relational bet loses not because the market rejected it but because the instrument never gave it a hearing.
The Formal Process
[Operator R&D] runs only when Exit C triggers at Layer 4 of the Decision Tree — when the operator has sufficient sight to know a decision is required but insufficient evidence to gate it soundly. The pilot builds the evidence the gate needs. Ten steps. No shortcuts. The discipline is in the completion.
Step 1 — What Do I See
[The Read] applied formally to the pilot context. Discovery phase. The operator walks into the situation — the stage, the kitchen, the numbers, the cast, the Guest behavior — and reads what is actually there. Not what he expects. Not what he hopes. What is actually happening. This is [The Read] in its formal application — not the instinct that evaporates after the shift but the disciplined observation that feeds the log. The Read provides inputs. It is not yet evidence. Evidence requires prosecution.
Step 2 — Where Is the Bias
Declaration revisited on the pilot-specific read. Before the evidence is assembled, the operator names his priors. What do I already believe about this situation. What has worked before that I am likely to weight too heavily. Which bias position is active — confirmation, [Golden Rule Bias], [Disruption Bias]. The bias has to be named before the prosecution begins or the prosecution has no defendant. This is the step most operators skip. It feels unnecessary when the read feels clear. That clarity is the signal the step is most needed. A read that feels obvious has almost certainly already been corrupted by a prior assumption. Name it before it names the verdict.
Step 3 — Just the Facts
[Bias Prosecution] runs on the pilot-specific read. The operator runs his read against his declared priors. Where does the evidence confirm the prior — note it but do not weight it yet. Where does the evidence contradict the prior — that contradiction is the signal. Strip the preference. Strip the historical scorecard. Strip the disruption lean. What does the evidence actually say about this operation, this Guest, this production, this moment. Not in general. Not based on what worked elsewhere or failed somewhere else. Here. Now. For this operator. Nothing is universally transferable. The prosecution standard is always operator-specific, market-specific, moment-specific.
Step 4 — What Does It Mean
[Five Stakeholder Read] runs. Business, cast, Guest, community, operator. All five surfaces, all weighted, simultaneously. Not lagging metrics alone. Not feel alone. The Guest surface and the business surface run together because Road 2 produces business outcomes through Guest relationships — you cannot evaluate one without the other. No single surface kills the pilot alone. A move that wins on the business surface but loses on the cast surface is a named trade-off, not a clean win.
Step 5 — Is There MDV Here
The gate inside the pilot. Before a test is initiated, the operator answers one question: is this deep enough to produce [Meaningfully Differentiated Value].
Meaningful — does it matter to the Guest in a way that registers, changes behavior, generates return, produces a connection moment that would not have existed otherwise.
Differentiated — does it deliver something the market is not already delivering, something a Road 1 operator structurally cannot replicate, something that is visible and valuable to the Guest on its own terms.
If the answer is no on either dimension, the test does not run. Not because the idea is bad but because a test that cannot produce MDV is a test that burns time, energy, and cast capacity on an incremental result. The formal process protects against that by naming the standard before the test begins. If the answer is yes on both — the test runs.
Step 6 — [R&D Brief]
The hypothesis is stated. Both instruments declared upfront before the test runs — the lagging metric that will be watched AND the GX indicator that will be logged alongside it. The test window is defined. Not “until it feels right.” A named period — one period minimum, two periods for anything that touches Guest behavior, which lags by nature. The brief is the document that makes the initiation real. Without it the test runs informally against an undeclared standard and [Informal Instrumentation] reclaims the evaluation the moment results become ambiguous. The declared criteria in the brief are the protection against the informal process reasserting itself mid-pilot.
Step 7 — Test Runs
Inside the declared window. No early termination because the lagging metric did not move in week one. No extension because the operator likes the idea and does not want a verdict yet. The window was declared. It runs. [Read Log] runs continuously during the test window — every production surfaces something, every notable read gets logged before it evaporates.
Step 8 — Results
Both instruments read at close of window. Lagging metrics and logged GX indicators weighed simultaneously. Neither instrument kills the pilot alone.
A Road 2 move that underperforms on the lagging metric but shows strong GX indicator movement is not a failed pilot. It is an early-stage relational investment that needs a longer window, a scaled approach, or a cast development component before the lagging metric will reflect what the GX indicators are already showing. The formal process holds that distinction. [Informal Instrumentation] cannot — it only sees the lagging metric and calls the test dead.
The question at this step is not “did it work.” The question is: are the results indicative of the MDV target. If yes on both instruments, scale. If strong on GX indicators but weak on lagging metrics, extend with a declared second window and stated re-evaluation criteria. If no on both instruments, kill.
Step 9 — Rollout
The test cleared both instruments. The MDV case is made. The discipline now shifts from evaluation to execution — and execution has its own formal requirements. Rollout is not “tell the cast we’re doing this now.” Rollout is a staged deployment with a named owner, a training component, a cast buy-in process, and a feedback mechanism built in from day one.
Post-rollout, the [Five Stakeholder Read] runs again. Business, cast, Guest, community, operator. What is the evidence saying now that the idea is running at scale. What did the test not show that full deployment revealed. What is the cast saying that the stage read is confirming or contradicting. What are Guests doing that the lagging metrics have not caught yet. This is not a satisfaction survey. It is a formal intelligence gather from every surface the rollout is touching. The feedback feeds the refinement.
Step 10 — Refine. The Loop Continues.
The formal process is not a one-time event. It is a compounding discipline. Each refinement pass produces a better read, a sharper prosecution, a more precise MDV test, and a deeper body of logged operational intelligence.
The informal process burns intelligence every shift. The formal process banks it. The compounding is slow at first and then it is not. The operator two years into [Operator R&D] is making decisions that the informal operator cannot make — not because he is smarter or more experienced but because he has evidence the informal operator destroyed. The reps were always there. The formal process makes them legible. The intelligence was always being gathered. The log makes it durable. The bias was always running. The prosecution makes it visible.
After Step 10, the pilot verdict and evidence re-enter the Decision Tree at Layer 3. Aggregation runs again — now on evidence, not assumption. [Bias Prosecution] runs again against the pilot evidence. The gate runs. Exit A or Exit B from here. No second Exit C on the same decision. The pilot built the evidence. The gate decides on it.
What This Produces
The operator who builds this discipline does not become a researcher. He becomes an operator who can defend his decisions with evidence instead of feel, who can protect a relational investment from a transactional scorecard long enough to see what it actually produces, and who accumulates a body of logged reads that compounds into genuine operational intelligence over time.
None of this requires a research budget, a consulting firm, or a corporate process. It requires the discipline to name what you are doing and the honesty to prosecute what you already believe. That is the work. It was always the work. Now it has a name.
What Changes Tomorrow
Name one test you have run in the last six months that died because the numbers did not move in the first two weeks. Run it backward through Steps 1 through 5. What did the read show that you never logged. Which bias was running in Step 2 that you never named. What did the [Five Stakeholder Read] show across all five surfaces that the lagging metric alone missed. What GX indicators were present that the informal scorecard never counted. The test may not have failed. The instrument may have failed the test. Run the backward pass today and find out which one it was.

Explanation
See Definition.