Every operator I have worked with can tell me what they made last period. Almost none of them can tell me how many eating occasions their households had, or how many of those occasions they won. That second number is the growth number. The first one is a performance read on the second.
This piece is about the unit. Not about a metric to add to the dashboard — about which unit growth is actually denominated in, and what changes in the operation once the right one is running underneath the decisions.
The Denominator Decides Everything Downstream
[Share Of Stomach] denominates in occasions. A household has roughly twenty-one eating occasions in a week. Every one of them is decided, and every one of them is decided against a field of alternatives that includes the operation, other operations, the deli case, the c-store, the desk drawer, and the household’s own refrigerator. [Share Of Stomach] is the count of occasions the operation won divided by the count of occasions that existed.
Per-person average denominates in occasions the operation already won. It is dollars per person per won occasion. That distinction is the whole architecture. Change the denominator and you change the competitor set, the growth arithmetic, the daypart strategy, and what the operator is looking at when he decides where the next dollar goes.
Both numbers are real. Both are needed. They answer different questions and they frequently move in opposite directions, and an operation that runs only the second one will read its own contraction as improvement without ever making an error.
PPA Is A Performance Unit Being Run As A Growth Unit
Per-person average tells the operator how well the operation converted an occasion it already won. That is a legitimate and important read — it grades the menu architecture, the mix, the cast’s ability to build the occasion into something worth what it costs. As a performance instrument it is doing exactly the job it was built for.
It cannot do growth. Per-person average has no denominator for loss. An occasion the operation did not win does not appear in the ratio as a zero — it is absent from the data entirely. So a base that shrinks by a fifth, spending 25% more per head, produces a rising line, and nothing in the instrument distinguishes that line from actual growth. That is [Measurement Asymmetry], and it is not a flaw to be corrected. It is the shape of a ratio built on won occasions.
The two units move against each other more often than operators expect. Per-person average rises by extracting more from each occasion won. [Share Of Stomach] rises by winning more occasions. Occasion-consolidation behavior — a household stacking three trips into two larger orders — raises the first and lowers the second at the same time. Discount-driven volume does the opposite. An operator watching only one of the two cannot tell which of those two things is happening in his own building.
The Headroom Arithmetic
This is the part that reorders priorities, and it is arithmetic anyone can run on the back of a check.
Say a household in your trade area gives you two of its twenty-one weekly occasions. That is roughly 9.5% share. Now consider the two moves available.
Squeeze per-person average 8%. That move operates on two occasions. Best case, you collect 8% more on the sliver you already hold, and you pay for it in menu engineering, upsell coaching, and whatever the Guest concludes about your prices.
Win a third occasion. That is a 50% increase in the volume you hold from that household. Fifty percent, against eight.
The asymmetry is not close, and it is structural rather than situational. Any operation holding a minority of the field has vastly more headroom in the field than in the sliver. And the lower the share, the more extreme the asymmetry gets — which means the operators with the least share are exactly the operators for whom the harvest strategy is worst, and they are the operators most likely to be running it, because a small operation reaches for the lever that fits inside its own four walls.
The harvest lever is not wrong. It is the smaller lever, and it is being pulled as though it were the larger one.
The Occasion Can Be Split
Here is the read most operators have never had to build, because until recently the field did not require it.
Fifty-three percent of grocery prepared-food buyers construct a hybrid meal — a prepared entrée from the deli case plus something finished at home. That occasion was not won and it was not lost. It was split. And the half that went to the deli case was in contention. The operation was a live candidate for it.
A whole-occasion loss and a half-occasion loss are different conditions requiring different moves. A whole loss means the household chose a different destination for the entire event. A half loss means the household wanted part of what the operation makes, and the operation was not available in a form that fit the rest of the evening. Those are two different problems and one of them is far more addressable than the other.
Neither one shows up anywhere in restaurant data. This is the sharpest current example of why the occasion has to be the unit: the field has developed a form of partial capture that the industry’s instruments cannot represent at all. When the unit is the occasion, the half-occasion is at least a visible, nameable condition. When the unit is the dollar per attendance, it does not exist.
What Actually Wins An Occasion
The field is not being won on price, and the evidence on that is unusually clean.
Thirty-seven percent of shoppers say deli-prepared food is just as affordable as restaurant options — parity, not advantage. When researchers asked what drives the purchase, the answers came back functional rather than purely economic, with convenience and time savings cited most often. Frequency in the category rose 9% year over year, 61% buy ready-to-eat at least monthly, and substitution runs inversely to age, with nearly a third of 18-to-29-year-olds substituting.
Parity price plus a non-price win is the most instructive combination in the current data, because it tells the operator exactly which variable is in contention. It is not the money. It is the fit — the shape of the occasion, the effort it demands, the decision window it has to be won inside.
That window is narrow and it is late. Deli purchases concentrate between noon and five in the afternoon. Sixty percent of people decide lunch the same day, 54% decide breakfast the same day, and about half decide dinner the same day. The occasion is not being planned. It is being resolved a few hours out, by whoever is available in a form that fits.
One note on the evidence, because the discipline applies even when the data supports my argument. The strongest body of research on that channel is published by the grocers’ trade association. It is credible and I have used it. Across six sources on the category, not one named a single weakness of grocery prepared foods, and evidence that only points one direction is telling you something about who assembled it. Use it, and name who owns it.
Building The Occasion Ledger
Five moves, in order. This is a period-cadence discipline, not a project.
One — Count occasions before dollars. Every period, occasions won goes on the page first and dollars per occasion goes second. The order is not cosmetic. It sets which number the room treats as the growth number and which one it treats as the grade.
Two — Split check movement into price and quantity. Quantity growth on flat or falling occasion counts is consolidation, and consolidation is share loss wearing a performance costume. Report the split, not the total.
Three — Build the substitute list per daypart. For each daypart, name the specific channels that could take the occasion, and name the variable each one wins on. Specific — the store two miles out, the c-store on the commute pattern, the household’s own refrigerator. Convenience-store inside sales reached $341.2 billion in 2025, a twenty-third consecutive year of growth, and 60% of c-store operators report rising foodservice sales concentrated in breakfast and lunch. That is a named competitor at two specific dayparts.
Four — Read the decision window. For each daypart, name when the household decides and whether the operation is present in that window in a form that fits. Present means findable, reachable, and executable inside the window — not open.
Five — Attribute the loss. For every occasion count that fell, produce a hypothesis with a name on it. Not macro. Not the consumer. A channel, a variable, and a daypart. The hypothesis will often be wrong. A wrong named hypothesis is a working instrument; an unnamed loss is not.
What You Do Monday Morning
Take one daypart — the weakest one — and build a single page with two lines on it: occasions won for the last twelve periods, and dollars per occasion for the last twelve periods, in that order. Nothing else on the page.
Then answer one question in writing: of the occasions that left, how many left whole and how many were split. You will be estimating. Estimate anyway, and write down what the estimate is based on. The estimate is the first version of an instrument you do not currently have, and next period you will have a better one.
Run that page every period for a quarter and the growth conversation in your building changes on its own, because the first number on the page is finally the number growth is denominated in.
The Closer
The dollar is what you collect. The occasion is what you win. An operation that measures only the first will get progressively better at collecting from a field it is progressively losing, and every instrument it owns will confirm that things are improving. Put the occasion first on the page and the rest of the architecture reorders itself behind it. This is [By Design Or By Default] at the level of the unit — you choose the denominator, or you inherit one and let it choose your strategy.
To understand what not to do, as it exists in the wild, go to Hacksterism.
Digging Deeper
Positions on the record
-
What The Guest Experience Actually Is — https://physics.jeffreysummers.com/what-the-guest-experience-actually-is
-
The Twelve Laws Of The Guest Experience — https://physics.jeffreysummers.com/the-twelve-laws-of-the-guest-experience-2/
-
Administered Pricing Without A Pricing Department — https://physics.jeffreysummers.com/administered-pricing-without-a-pricing-department-what-independent-operators-actually-face/
-
QFO: What Else Can I Do But Discount — https://physics.jeffreysummers.com/qfo-what-else-can-i-do-but-discount/
-
Your Sales Are Up Because Fewer People Came — https://hacksterism.jeffreysummers.com/your-sales-are-up-because-fewer-people-came
-
Why Too Expensive Is Never About The Price — https://hacksterism.jeffreysummers.com/why-too-expensive-is-never-about-the-price/
-
The Operator’s Contract — https://jeffreysummers.com/the-operators-contract/
Term definitions from the Knowledge Base
-
[Share Of Stomach] — https://kb.jeffreysummers.com/docs/share-of-stomach/
-
[Share Of Experience] — https://kb.jeffreysummers.com/docs/share-of-experience/
-
[Measurement Asymmetry] — https://kb.jeffreysummers.com/docs/measurement-asymmetry/
-
[The Dashboard Trap] — https://kb.jeffreysummers.com/docs/the-dashboard-trap/
-
[Pricing Substrate] — https://kb.jeffreysummers.com/docs/pricing-substrate/
-
[Relational Pricing Substrate] — https://kb.jeffreysummers.com/docs/relational-pricing-substrate/
-
[Relational Compounding] — https://kb.jeffreysummers.com/docs/relational-compounding/
-
[Lost Opportunity Tax] — https://kb.jeffreysummers.com/docs/lost-opportunity-tax/
-
[No Static Achievement] — https://kb.jeffreysummers.com/docs/no-static-achievement/
-
[The Read] — https://kb.jeffreysummers.com/docs/the-read/
-
[By Design Or By Default] — https://kb.jeffreysummers.com/docs/by-design-or-by-default/
-
[Guest Experience] — https://kb.jeffreysummers.com/docs/guest-experience/
