Build a cost-per-kilo sheet for a small cashew processing unit

A small cashew unit buys raw nuts once, then pays for drying, shelling, peeling, grading and packing, losing weight at every step. You are building a sheet that works out the cost of one kilo of packed kernels from labelled inputs, using made-up practice data, and finding which input moves that cost the most.

• What is the unit of output: one kilo of packed kernels, or one kilo of a particular grade?
• Which steps add cost, and at which steps does weight drop?
• How will you handle the fact that whole and broken kernels sell differently?
• Which inputs are you least sure about, and how will you test them?

A tidy sheet with an inputs block, every input in a labelled cell and marked as made-up practice data, a calculation that follows the steps in order, a cost per kilo of packed kernels, and a small table showing how that cost moves when each key input changes.

Dividing total costs by the raw nut weight. Weight falls sharply between raw nut and kernel, so the cost must be spread over the kernels you actually pack.

• A list of processing steps from a public cashew processing guide
• Made-up practice inputs for each step, clearly labelled as such
• If possible, a conversation with a unit owner about which steps cost most, without asking for their figures

AI must not present any input as a real market price or real yield. Every number in the sheet is your labelled practice assumption, and the sheet says so at the top.

1. 3.1 Check the step logicCLAUDE CHAT

   Why this tool: Agreeing the steps and where weight is lost before building stops errors spreading through the model.

   List your steps and ask Claude whether anything is missing or out of order. Do not ask it for numbers.

   Prompt to copy · fill in the [brackets]Copy

   I am modelling the cost per kilo of packed cashew kernels for a small unit. My steps are: [list steps]. At which of these steps does weight usually drop, and is any cost step missing or in the wrong order? Do not give me any numbers, only the logic.

2. 3.2 Build the modelCLAUDE FOR EXCEL

   Why this tool: Building in a workbook keeps every assumption visible and every formula checkable.

   Make an inputs block at the top with labelled cells, then one row per step carrying weight and cost forward. End with cost per kilo of packed kernels. Ask Claude for Excel to explain any formula you did not write yourself.

3. 3.3 Save a sensitivity SkillCLAUDE SKILLS

   Why this tool: A saved Skill runs the same one-at-a-time test every time you change the model.

   Write a short Skill: given your workbook's input list, it suggests a low and high value for each key input and lays out a table for you to fill by changing the sheet. You do the changing; the Skill keeps the method consistent.

   Prompt to copy · fill in the [brackets]Copy

   Create a Skill called cashew-sensitivity. Given a list of model inputs and their current practice values [paste input list], it proposes a low and a high test value for each, explains why, and produces an empty table with columns input, low, base, high, cost per kilo at low, cost per kilo at high. It must not calculate results itself.

   Before you let it act: Before using the table, check that every input listed matches a labelled cell in your sheet and that the Skill has not filled any results itself.

4. 3.4 Write the findingCLAUDE CHAT

   Why this tool: A plain sentence on what moves the cost is what a unit owner can act on.

   Paste your completed sensitivity table and ask Claude to help you say which input matters most, then rewrite it in your own words.

   Prompt to copy · fill in the [brackets]Copy

   Here is my completed sensitivity table for a cashew unit model built on made-up practice data: [paste table]. In two sentences, which input moves the cost per kilo the most and what might a unit owner do about it? Remind the reader the data is practice data.

• No number is typed inside a formula; every input sits in a labelled cell
• The sheet says clearly at the top that all inputs are made-up practice data
• You recomputed the cost per kilo by hand for the base case
• Weight is carried forward step by step, not assumed constant
• No number typed inside a formula: every input is a labelled cell
• Recompute the headline result by hand

AI often offers typical yields or raw nut prices as if they were facts. Do not use them; your inputs are labelled practice assumptions.

• Step list checked
• Inputs block with labelled practice data
• Cost per kilo calculation
• Sensitivity table
• Two-sentence finding

Show the inputs block and the final cost per kilo on one screen, then the sensitivity table with the most important input highlighted.

Explain why spreading costs over raw nut weight gives the wrong answer, and what your model showed about the input that matters most.