Bin less, prove it, keep the money: 10 AI skills for waste and running cost

food-waste-count

measure what you throw away, by reason, in a week

How the two work together

Claude thinks it through. Paste the Claude prompt into Claude Code, or drop the folder into your skills folder. Claude does the judgement: what to look for, what is worth doing, what is right.

Codex gets it done. At the hand-off point Claude runs Codex on your machine with one command and passes it the Codex prompt. Codex does the mechanical part and hands the result back. Claude checks it before you see it.

No API key to set up: Claude calls the Codex you already have installed. If Codex is not installed, Claude does that half itself and tells you.

Prompt for Claude

---
name: food-waste-count
description: Runs a seven day count of everything your kitchen throws away, split by the four reasons WRAP uses, weighed rather than guessed, priced against what you actually paid for the ingredients, and written up as a baseline you can repeat. Use before you change a menu, before you argue with a supplier about pack sizes, or when you suspect the bin is costing more than the staff rota.
---

# One week of weighing, and the first honest number for what your bin costs you

You give this seven days of your kitchen's time, a set of scales, four containers and your last month of invoices. You get back one dated file: every item thrown away, weighed, split into spoilage, preparation, plate and other, priced at what you paid for it, totalled by day and by reason, with the three biggest lines named and the questions each one raises. It will not tell you how you compare to other venues, because no published figure exists for a single independent site, and it will not act on the result. Acting is the next week's job.

## What it does

1. **Set a baseline before changing anything, because a count taken after a fix cannot measure the fix.** WRAP's Guardians of Grub programme is built on Target, Measure, Act in that order, and its own FAQ defines the first step plainly: "A baseline is the amount you are throwing away before you take any action. In the 'Guardians of Grub' starter pack we encourage you to establish a baseline before you take any action to give a realistic starting point." So the week you count is a normal week. No new portion sizes, no supplier change, no special board. If something has to change mid-week, note the day it changed and treat the week as two halves rather than pretending it was one.

2. **Use the four categories exactly as WRAP names them, because the split is the finding, not the total.** The Guardians of Grub FAQ defines them: "Spoilage: Food that is damaged or out of date such as spoilt vegetables. Preparation: Food that is thrown away during preparation, such as offcuts, and food that is served but not saved. Plate: Food that is left on customers' plates such as chips and garnishes. Other: Food thrown away, such as food that is ready to serve e.g. buffet, but not eaten." One total tells you nothing you can act on. A week that is mostly preparation is a knife and specification problem, mostly spoilage is an ordering and rotation problem, and mostly plate is a portion and menu problem, and each goes to a different person.

3. **Put a container under each category where the waste is actually made, and pick scales that suit the container.** The FAQ is practical about both: "You can use any type of container suitable for holding food. It is recommended to use small containers if on the counter, or buckets if on the floor", and "If you use counter top containers, any kitchen scales should be suitable. If you use buckets and/or transparent liners it may be easier to use luggage scales." Transparent containers or liners are worth the trouble: the FAQ notes that "often using transparent containers or sacks to see where the waste is coming from can be an eye opener". Write the tare weight of every empty container on its side in marker before the week starts, because a bucket weighed with its liner is the commonest way a count goes wrong.

4. **Choose a seven day window that contains your worst day and your quietest one.** The FAQ is explicit that a convenient week is the wrong week: "Ideally the period that you track should capture both the busy and quiet periods in your week, as you may find on busy days you have more plate waste and on quiet days you have more preparation waste." For most venues that means the window has to include the Saturday and the Monday, not the two Tuesdays that were easy to staff. Write the trading pattern of each of the seven days next to its row, because a day's weight means nothing without the covers, the service and the weather beside it.

5. **Weigh everything, and write the weight down at the bin rather than at the end of the shift.** The number that gets reconstructed at ten o'clock is a guess wearing a decimal point. Weigh at each service change, record kilograms to two places, and empty and re-tare. Where a single item is too awkward to weigh, count it instead and record the unit weight from the delivery note, which is a measurement rather than an estimate. The FAQ's advice for the hardest category is to reduce the burden rather than skip it: for plate waste when busy, "the key is to make it as easy as possible by working out the best place to put the bins, weighing food when you don't have a rush on and keeping notes about what is wasted to a minimum".

6. **Name the item and the reason in the fewest words that still tell you something next month.** "Veg" is not a finding. "Spinach, wilted, bought Tuesday, used Friday" is one, because it contains a shelf life and a rotation failure. Keep the note to a handful of words, but always include the item, the state and, for spoilage, the delivery date. Carcasses and bones have their own home: the FAQ puts them in preparation, and offers the two questions that follow from them, which are whether you could be "using bones in stocks" and whether you could "talk to your meat supplier about supplying meat off the bone".

7. **Price every line at what you actually paid, not at menu price and not at a market rate.** Take the unit cost off the invoice for that delivery, not an average, and show the invoice reference beside it. Menu price is the wrong number twice over: it counts profit you never earned, and it makes the total large enough that nobody believes it. Where an item came from a prepared dish rather than a raw ingredient, price the ingredients in it at invoice cost and say so. The purpose is a figure the owner will defend to a supplier, and a defensible small number beats an impressive large one.

8. **Read the week as three separate questions rather than one total.** Which single line is biggest by cost, which by weight, and are they the same line? A heavy cheap line and a light expensive one need different actions and the total hides both. Then: which category dominates, and does that match what the kitchen expected before the week started? Then: how much of the week sits in spoilage, which is the category that is almost always recoverable by ordering differently rather than by working harder. Set the target against your own baseline, in kilograms or pounds off this venue's own number, and never against a figure from anywhere else.

9. **Date the file, keep the sheet, and book the repeat.** The count is only a baseline if there is a second one. Book the same seven days in the calendar for the following month or quarter, with the same containers, the same categories and the same scales, and record any change of method as a break in the series. GOV.UK's food and drink waste hierarchy puts measurement first among the prevention actions, saying "It is good practice to measure and record your food and drink surplus and waste", and names Guardians of Grub as a route to doing it. Store the completed sheets with the build date on the front, because a count with no date will be quoted for years.

## Then it checks

1. All seven days were counted, the window includes at least one of the venue's busiest days and one of its quietest, and any uncounted service is named rather than left as a gap in the totals.
2. Every entry carries a weight in kilograms taken from scales, or a count multiplied by a unit weight read from a delivery note, and no entry is recorded as an estimate without being labelled as one.
3. Every entry is in exactly one of spoilage, preparation, plate or other, using WRAP's definitions, and no entry sits in two or in none.
4. Every priced line names the invoice or delivery note the unit cost came from, and no line is priced at menu price.
5. Container tare weights were recorded before the week began and were subtracted, with the tare for each container shown in the file.
6. The file states no comparison with any other venue, no per cover benchmark and no industry average, and every sector figure quoted is attributed to its publisher and labelled as sector wide.

Any check fails: name it, redo that step once. Failed twice: say what is wrong and stop.

## Rules
- Public information only.
- Never invent a fact, a number or a quote.
- Anything sent in someone's name says whose name it is. A summary shown to the team goes out over the name of the person who counted, not over the chef's unless the chef wrote it.
- Never state a benchmark, a target percentage, a typical waste per cover or an average for a venue of this type. No published figure exists for a single independent UK site, the numbers that circulate are sector totals, and an owner who sets a target from one will judge their own kitchen against a fiction.
- Never quote a sector figure without naming who published it and saying that it describes a whole sector. WRAP's Guardians of Grub states that "Every year, 1.1million tonnes of food is thrown away across our sector. 75% of the food wasted could have been eaten, costing us £3.2billion", and its resources page describes a campaign asset carrying a "20% of food purchased in the sector is binned" figure. Those are sector totals published by WRAP. Neither is a target for one kitchen and neither may be presented as one.
- Never price waste at menu price, never gross it up to lost revenue, and never annualise a single week into a yearly figure. One week times fifty two is not a year, it is one week with a multiplication error.
- Never present a count as a performance measure for an individual. The moment a weight can be traced to a person's shift the weighing stops being honest, and the count is worth less than the trust it cost.
- Refuse to report a total for a week in which a category was skipped. Say which category and which days were missed and report the rest as partial, because a total with a hole in it will be quoted as though it were whole.
- This output is a working document prepared for the owner to check against their own invoices, covers and rotas before it is relied on, and to discuss with their head chef and their supplier. It measures what this kitchen threw away in one week. It is not a benchmark, a compliance record or a food safety assessment.

## Built from
- WRAP, "Frequently Asked Questions for the Hospitality and Food Service sector", Guardians of Grub, https://guardiansofgrub.com/faqs/, no publication date shown on the page, copyright notice reads "WRAP 2024", read 16 September 2026: the four category definitions in step 2, the container and scales advice and the transparent container point in step 3, the baseline definition in step 1, the busy and quiet week rule in step 4, the plate waste practicality in step 5, and the carcass and bones questions in step 6.
- WRAP, "Take action to achieve your food waste prevention goals", Guardians of Grub, https://guardiansofgrub.com/, no publication date shown on the page, copyright notice reads "WRAP 2024", read 16 September 2026: the Target, Measure, Act ordering behind step 1, and the sector totals quoted under Rules, being "1.1million tonnes", "75%" and "£3.2billion".
- WRAP, "Resources", Guardians of Grub, https://guardiansofgrub.com/resources/, no publication date shown on the page, read 16 September 2026: the existence of the 7 Day Tracking Sheet, the Food Tracking Calculator, the bin labels "to add to your four food waste buckets" and the Stock Control Sheet, which is why the method in steps 3 and 9 is four containers and a repeatable weekly sheet, and the sector figure quoted in the Food Waste Action Week pack description.
- GOV.UK, Department for Environment, Food and Rural Affairs, "Food and drink waste hierarchy: deal with surplus and waste", https://www.gov.uk/government/publications/food-and-drink-waste-hierarchy-deal-with-surplus-and-waste/food-and-drink-waste-hierarchy-deal-with-surplus-and-waste, updated 1 January 2024, read 16 September 2026: "It is good practice to measure and record your food and drink surplus and waste" in step 9, and the placing of measurement inside prevention rather than beside it.
- GOV.UK, Department for Environment, Food and Rural Affairs, "Simpler recycling: workplace recycling in England", https://www.gov.uk/guidance/simpler-recycling-workplace-recycling-in-england, published 29 November 2024, last updated 3 October 2025, read 16 September 2026: the definition of what counts as food waste, including "waste from food preparation, such as onion skins" and "coffee grounds and tea bags", which sets the scope of what gets weighed, and the storage warning that food waste should not be kept for more than 2 weeks.

This is a working document prepared for the owner to check against their own invoices, covers and rotas before it is relied on. It measures one week in one kitchen. It is not a benchmark and it is not a compliance record.

Prompt for Codex

# food-waste-count

## You are given
Seven days of raw weighing sheets from one hospitality kitchen, as supplied files or transcribed rows. Each row holds a date, a service, an item description in the kitchen's own words, a gross weight in kilograms, the container it was weighed in, and a short reason note. Alongside them: the tare weight of every container, recorded before the week began; the venue's covers and trading pattern for each of the seven days; the delivery notes and invoices covering the period, with unit costs and delivery dates; Claude's assignment of each row to one of the four categories `spoilage`, `preparation`, `plate` or `other`, with the wording of WRAP's definitions and the URL and read date of the page they came from; a note of any service that was not counted and why; and a note of any change of method mid-week. Where a value is missing it arrives as the word `not supplied`.

## Produce
Write into a `./food-waste-count-output/` folder:

1. `entries.csv` with these columns in this order: `entry_ref`, `date`, `day_of_week`, `service`, `item_verbatim`, `category`, `reason_note_verbatim`, `container`, `tare_kg`, `gross_kg`, `net_kg`, `weight_basis`, `unit_cost_gbp`, `cost_source_reference`, `cost_source_date`, `line_cost_gbp`. `entry_ref` is E001 upward. `category` is exactly one of `spoilage`, `preparation`, `plate`, `other`, and is copied from the input and never changed. `weight_basis` is exactly one of `weighed`, `counted and multiplied by delivery note unit weight`, `estimated by kitchen`, `not supplied`. `net_kg` is `gross_kg` minus `tare_kg`, computed, never copied. `line_cost_gbp` is `net_kg` multiplied by `unit_cost_gbp`, computed.
2. `daily-totals.csv` with columns: `date`, `day_of_week`, `covers`, `trading_note`, `spoilage_kg`, `preparation_kg`, `plate_kg`, `other_kg`, `total_kg`, `spoilage_gbp`, `preparation_gbp`, `plate_gbp`, `other_gbp`, `total_gbp`, `services_counted`, `services_not_counted`. Seven rows plus a `TOTAL` row. Any day with a value in `services_not_counted` is listed in `gaps.md` and the `TOTAL` row carries the word `partial` in `trading_note`.
3. `category-totals.csv` with columns: `category`, `entries`, `total_kg`, `share_of_kg`, `total_gbp`, `share_of_gbp`, `largest_single_item_by_kg`, `largest_single_item_by_gbp`. Four rows plus a `TOTAL` row. Shares are percentages of this week's own total and are never compared to anything outside this file.
4. `top-lines.csv` with columns: `rank`, `ranked_by`, `item_verbatim`, `category`, `occurrences`, `total_kg`, `total_gbp`, `reason_notes_verbatim`, `delivery_dates_where_known`. Two blocks of ten rows, one with `ranked_by` of `kg` and one with `ranked_by` of `gbp`, so the heavy cheap lines and the light expensive ones are both visible.
5. `pricing-audit.csv` with columns: `entry_ref`, `item_verbatim`, `unit_cost_gbp`, `cost_source_reference`, `cost_source_date`, `priced_at_invoice_cost`, `issue`. `priced_at_invoice_cost` is `yes` only where an invoice or delivery note reference was supplied. `issue` is empty or exactly one of `no cost source supplied`, `cost source predates the week`, `cost source postdates the entry`, `menu price supplied instead of invoice cost`.
6. `baseline.md` - the dated baseline record in plain sentences: the seven dates counted, the busiest and quietest day by covers, the four category totals in kilograms and pounds, the single largest line by each measure, the container tare weights used, the counting method, any mid-week change of method, and the date the repeat count is booked for. No comparison to any other venue, no per cover benchmark, no annualised figure, no target.
7. `gaps.md` - a numbered list of: any service not counted, with its date; any entry with `weight_basis` of `estimated by kitchen`; any entry with no tare weight for its container; any entry with no cost source; any entry whose reason note is three words or fewer; any day with no covers figure; any mid-week change of method; and the total kilograms and pounds affected by all of the above together.

## Rules
- Codex measures, computes, totals and checks. It never invents, never reassigns a row to a different category than the one supplied, never rewords a kitchen's item description or reason note, and never sets the venue's target.
- Never write a benchmark, an industry average, a target percentage, a waste per cover comparison or a figure for how other venues perform. None is to appear in any file.
- Never annualise. A weekly total is never multiplied to a month, a quarter or a year in any file.
- Never price a line at menu price or gross a line up to lost revenue. Where only a menu price was supplied, `unit_cost_gbp` is left empty and the row is flagged in `pricing-audit.csv` and in `gaps.md`.
- Never attribute an entry to a named member of staff, a shift leader or a rota position, and never write a per person total, even where the source sheet names someone.
- Net weights are computed from gross minus tare, never copied from a supplied net figure. Where a tare is missing the row's `net_kg` is left empty rather than assumed to be zero.
- A total for a week with an uncounted service is labelled `partial` wherever it appears, including in `baseline.md`.
- Quote any sector figure only with its publisher named and the words "sector wide" beside it, and never in the same sentence as this venue's own number.
- Use British English, GBP for money, kilograms to two decimal places for weight, and DD Month YYYY for every date. No em dashes in any file you write, and any supplied text containing one is recorded verbatim and flagged in `gaps.md`.
- Every file ends with this line: this is a working document prepared for the owner to check against their own invoices, covers and rotas before it is relied on; it measures one week in one kitchen and it is not a benchmark.

## Return
The absolute path of each file written, the row count of each CSV, the seven dates counted and any service not counted, the total net kilograms and the total pounds for the week, the four category totals in kilograms and pounds with each category's share of the week's own total, the single largest line by kilograms and the single largest by pounds with both named, whether those two are the same line, the number of entries with `weight_basis` other than `weighed`, the number of entries with no cost source, the number of entries missing a tare weight, whether the week is `complete` or `partial` and why, the repeat date written into `baseline.md`, and the `gaps.md` item count.

Built from the best public work on this

Sources for food-waste-count

Everything below was opened and read on 16 September 2026. Nothing is cited that could not be loaded.

1. WRAP, "Frequently Asked Questions for the Hospitality and Food Service sector", Guardians of Grub

https://guardiansofgrub.com/faqs/, no publication date shown on the page; the footer copyright notice reads "WRAP 2024", read 16 September 2026.

The single most useful thing published on measuring food waste in a working kitchen, and useful precisely because it is not a strategy document. It answers the questions a chef asks on day one, and most of this skill's method is lifted from it directly.

The four categories in step 2 are quoted verbatim because paraphrasing loses the boundaries: "Spoilage: Food that is damaged or out of date such as spoilt vegetables. Preparation: Food that is thrown away during preparation, such as offcuts, and food that is served but not saved. Plate: Food that is left on customers' plates such as chips and garnishes. Other: Food thrown away, such as food that is ready to serve e.g. buffet, but not eaten". Two details earn their place. "Food that is served but not saved" sits under preparation rather than plate, which is where most people would put it, and "Other" is defined by an example rather than as a residual, which stops it becoming the bin everything awkward goes into.

Step 1's baseline definition is the page's own: "A baseline is the amount you are throwing away before you take any action ... This could be week 1 using the 7 day tracker and then the calculator tool for a month."

Step 3's equipment advice is the kind of specific that saves a wasted week: "It is recommended to use small containers if on the counter, or buckets if on the floor. If you use counter top containers, any kitchen scales should be suitable. If you use buckets and/or transparent liners it may be easier to use luggage scales." The reason for transparency is given elsewhere on the page: "often using transparent containers or sacks to see where the waste is coming from can be an eye opener."

Step 4 exists because of one answer. Asked about peak days, the page says "Ideally the period that you track should capture both the busy and quiet periods in your week, as you may find on busy days you have more plate waste and on quiet days you have more preparation waste." A count run on a convenient week is not wrong by a little, it is wrong in a specific direction.

Step 5 takes the page's realism about the hardest category: asked what to do when plate waste tracking is hard during service, it answers "the key is to make it as easy as possible by working out the best place to put the bins, weighing food when you don't have a rush on and keeping notes about what is wasted to a minimum." That is why the skill asks for a short reason note rather than a description. Step 6's carcass questions are also from here: bones sit in preparation, and the page offers "using bones in stocks, or if you don't use stock you could talk to your meat supplier about supplying meat off the bone."

Where the skill departs: the page recommends appointing a Guardians of Grub Champion and handing a baton on when they are away. The skill keeps the rota discipline and drops the title, because a named owner of the count is also a named owner of the number, and in a small kitchen that is one step from the count becoming a performance measure, which is where honest weighing stops.

2. WRAP, "Take action to achieve your food waste prevention goals", Guardians of Grub

https://guardiansofgrub.com/, no publication date shown on the page; the footer copyright notice reads "WRAP 2024", read 16 September 2026.

The campaign's front page, and the source of the sector context the skill is careful about. It states: "Every year, 1.1million tonnes of food is thrown away across our sector. 75% of the food wasted could have been eaten, costing us £3.2billion." It describes the programme as "delivered by environmental action NGO WRAP" and frames the approach as embedding "Target-Measure-Act within their everyday food and financial processes", which is the ordering step 1 enforces.

These are sector totals for UK hospitality and food service. They are quoted in the skill's rules with WRAP named and the words "sector wide" attached, and the rule below them forbids turning them into a target for one kitchen. That restriction is why the figures sit in the rules rather than in the method: an owner who reads "75% could have been eaten" and sets a 75 per cent reduction goal has invented a target out of a national total.

Where the skill departs: the page's first suggested step is the Savings Calculator, used "to take a look at how much money you could save by simply reducing your food waste". The skill does not use it or quote its outputs, because a modelled figure for a venue is not a measurement of that venue, and this skill's whole value is that every number in it was weighed or read off an invoice.

3. WRAP, "Resources", Guardians of Grub

https://guardiansofgrub.com/resources/, no publication date shown on the page, read 16 September 2026.

A catalogue of free downloads, cited for what it reveals about the intended method rather than for prose. Three entries settle design decisions. The bin labels are described as labels "to add to your four food waste buckets", which confirms that four physical containers, one per category, is the format the categories were designed for rather than one bin and a clipboard. The "7 DAY TRACKING SHEET" helps "food businesses to easily monitor and measure food thrown away over a 7 day period", which is where the seven day window comes from. The "Stock Control Sheet" helps "co-ordinate stock checks and purchasing", which is the link between a spoilage-heavy week and an ordering fix rather than a kitchen fix.

The page also carries two sector figures in its descriptions. The Meat in a Net Zero World Toolkit entry states that "WRAP research shows that over 50,000 tonnes of meat is wasted by the UK Hospitality and Food Service sector every year". The Food Waste Action Week pack entry describes "a creative image sharing the '20% of food purchased in the sector is binned' stat", which the page presents as the content of a campaign image rather than as a finding, and which the skill's rules describe the same way. Both are sector figures published by WRAP. Neither is offered as a benchmark and neither may be used as one.

Where the skill departs: it does not reproduce or depend on WRAP's tracking sheet and calculator files, which are copyright and which the site's terms state must not be adapted or used to create derivative works. It describes a method of the same shape and tells the owner where the originals are.

4. GOV.UK, Department for Environment, Food and Rural Affairs, "Food and drink waste hierarchy: deal with surplus and waste"

https://www.gov.uk/government/publications/food-and-drink-waste-hierarchy-deal-with-surplus-and-waste/food-and-drink-waste-hierarchy-deal-with-surplus-and-waste, updated 1 January 2024, read 16 September 2026.

Cited for one short passage that does a lot of work, and for where it sits. Measurement appears inside option 1, prevention: "It is good practice to measure and record your food and drink surplus and waste ... You can also use the Food Waste Reduction Roadmap or Guardians of Grub resources to help you."

Two things follow. Government guidance names Guardians of Grub as a route, which is why this skill builds its method on a campaign site rather than a statute. And the page is explicit that the top of the hierarchy is not the legal part: "Options 1 to 4 offer you guidance on how to prevent food waste occurring in your business. Options 5 to 8 are required by law." Measuring is not a legal duty. It is the only way to find out which of the eight options you are on.

Where the skill departs: the rest of the page is about what to do with surplus once it exists, which is the subject of the use-it-up-menu skill in this pack and is not restated here. One fact, one home.

5. GOV.UK, Department for Environment, Food and Rural Affairs, "Simpler recycling: workplace recycling in England"

https://www.gov.uk/guidance/simpler-recycling-workplace-recycling-in-england, published 29 November 2024, last updated 3 October 2025, read 16 September 2026.

Cited to set the scope of what gets weighed, because the legal definition of food waste is wider than a chef's. The page lists food waste as including "food leftovers, including from customer or staff meals", "banana skins and apple cores", "coffee grounds and tea bags", and "waste from food preparation, such as onion skins". Coffee grounds and tea bags are the ones that get left out of a count, and in a cafe they are not a rounding error.

It also supplies the practical constraint on running the count at all: "It is not recommended that you store food waste for more than 2 weeks." A count is a week, not a fortnight of accumulation, and a kitchen planning to save everything up to weigh at the end has a pest problem rather than a baseline.

Where the skill departs: the page's subject is the separation duty, which belongs to the waste-separation skill in this pack. The count described here is a measurement exercise and is not evidence of compliance with anything, and the skill says so in its closing rule.

Best public prompt we found for this job

The closest public artefact is the `analyze` skill in Anthropic's `knowledge-work-plugins` repository, raw source at https://raw.githubusercontent.com/anthropics/knowledge-work-plugins/main/data/skills/analyze/SKILL.md. The repository has 24,123 stars, read from api.github.com on 16 September 2026. Its best section is the one most analysis prompts leave out, a validation gate placed before presentation rather than after it:

**Row count sanity**: Does the number of records make sense?
**Null check**: Are there unexpected nulls that could skew results?
**Magnitude check**: Are the numbers in a reasonable range?
**Trend continuity**: Do time series have unexpected gaps?
**Aggregation logic**: Do subtotals sum to totals correctly?

That is the ancestor of this skill's checks and of `gaps.md`. "Trend continuity: do time series have unexpected gaps" became the rule that a week with an uncounted service is labelled partial wherever it appears, which is the single most common way a kitchen count misleads: three services missed on the Saturday and a total that reads like a good week.

What we did not copy. It offers three output tiers ending in a formal report with recommendations, which is one tier too many for seven days of bucket weights and would invite the model to decide what the kitchen should change. This skill stops at the finding, because the questions in step 8 have answers depending on things no sheet records, such as which supplier will actually change a pack size. We dropped its instruction to visualise where helpful, because a chart of four categories over seven days is four numbers wearing a costume. And we added the constraint the source had no reason to carry: no comparison to anything outside the file. A data skill's instinct is to benchmark, and benchmarking against a number that does not exist for a single venue is the main way this week gets wasted.

This is a working document prepared for the owner to check against their own invoices, covers and rotas before it is relied on. It measures one week in one kitchen. It is not a benchmark and it is not a compliance record.

Want this running in your business?

I optimise how businesses run — your sales, your visibility, your social media — and build bespoke software where nothing off the shelf fits. The first conversation is free. Work starts from £150 a day.