From the book · Unleash Your Business With AI
Chapter 18: Numbers, pricing and decisions
The 4 prompts from Chapter 18, ready to copy, with the chapter's mistakes to avoid and three tasks for this week.
Ask your sales data three questions
I am giving you [file name], a cleaned sales export for [period] from [Tally / Excel / marketplace report]. It contains no customer personal data. Parties are shown as codes. Columns: [list each column and what it means] [upload the file or paste the rows] Before answering, report the number of data rows and the grand total of [net amount column], so I can match them with my books. Then answer: 1. Which [10] products sold more than in [previous period], by value? Show both figures. 2. Which products had no sales at all in this period? 3. Which party codes ordered in [previous period] but not in this one? 4. [your own question] Rules: - Calculate with code or exact formulas, not estimation. If you cannot run code, say so and give me the Excel formula or pivot table to use instead. - For each answer, show the method and the figures it is based on. - Do not explain why something sold or did not sell unless the data shows it. Mark any reason you offer as a guess. - If the data looks wrong or incomplete, tell me. Do not work around it.
What you should get back: a row count and grand total first, then four answers, each with its method.
Check: match the row count and grand total with your source. Confirm one answer with a filter in Excel. The "why" in question 1 is yours to answer. You know there was a wedding season or a stock-out. The AI does not.
That's the first one
Open the rest, free
Enter your email once and every prompt page and guide on this site opens.
- Find the stock that is not moving
- Three ways to price
- Argue against me
- Mistakes to avoid
- Do this week
Find the stock that is not moving
I am giving you our stock report exported on [date]. Columns: [for example "Item code, Description, Category, Qty in stock, Cost per unit, Last sale date, Qty sold in last 90 days"] [upload the file or paste the rows] 1. Using code or exact formulas, list items with no sales in the last [X] days, and items where stock is more than [Y] times the last 90 days' sales. 2. For each, show stock value at cost (quantity x cost per unit), and give the total for the list. 3. Group the list by category. 4. Suggest ways to clear these items that suit a [B2B / B2C / both] business, such as bundles, retailer schemes, marketplace offers, corporate gifting, return to supplier or use as samples. Note the risk of each. 5. Show your method so I can recheck in Excel. Do not suggest any price or discount percentage. What to clear, and at what price, is my decision.
What you should get back: a list grouped by category with a value against each item, a total and some options with their risks.
Check: recalculate the value of the three largest items by hand, and walk to the godown to confirm they are really there.
Three ways to price
Product: [name] Cost per unit: [landed cost, including freight in] Channel costs, from my own records: [for example "Marketplace: commission and fees per unit", "Retailer margin: X%", "Distributor margin: Y%"] GST rate: [rate, as confirmed by my accountant] Current selling price by channel: [list] Price scenarios I want to compare: [for example "current", "+5%", "-5%", "₹749"] Build a table as Excel formulas showing, for each channel and each scenario: selling price, GST, channel costs, net realisation, margin per unit and margin as a % of selling price. Add one row per scenario showing how many units I must sell to earn the same total margin as [number] units at the current price. Rules: - Give me formulas and tell me which cell each goes in, not just values. - Do not assume any fee, rate or cost I have not given. Write <<INPUT>> where a value is missing. - Do not recommend a price. I will decide.
What you should get back: a block of formulas you can paste into a sheet, with <<INPUT>> wherever you left something out.
Check: enter the fictional figures above (cost ₹400, prices ₹520 and ₹749, no other costs). The sheet should show ₹120 and ₹349 per unit, and about 34.4 units to match 100 sales at ₹520, which means 35. If it does not, the formulas are wrong. Tax questions go to your CA.
Argue against me
[paste your business note from Chapter 5] Decision I am considering: [for example "opening a second showroom", "launching a premium range", "hiring a regional sales manager"] My reasons: [list] What I know about costs and risks: [list, using ranges if you prefer not to share exact figures] Act as a sceptical, experienced business advisor who wants me to succeed but thinks this may be a mistake. 1. Give the 5 strongest reasons not to do this. 2. Pre-mortem: imagine it is a year later and the decision failed. Give the 5 most likely reasons. 3. List the assumptions I am making that I have not tested. 4. For each reason, suggest one small, low-cost way to test or reduce the risk before committing. 5. List the 3 questions I should answer with real data before deciding. Rules: - Use only the facts I have given. Do not invent market sizes, statistics or competitor details. If you need a fact, ask me for it. - Be direct. Do not soften your points. - Do not tell me what to decide.
What you should get back: objections, failure reasons, untested assumptions, small tests and three data questions.
Check: any number or market fact in the answer that you did not supply is unverified. Strike it out or confirm it. Then answer the three questions from your own books.
Mistakes to avoid
- Asking AI to total a column in its head. Ask for the formula or for code, then match the grand total.
- Comparing turnover instead of money earned over cost. A price rise can cut turnover and still earn more.
- Letting AI suggest the price or the discount. It builds the table. You decide.
- Using a fake "was" price as an anchor. It is dishonest, and consumer law treats misleading representations as an unfair trade practice.
- Treating the AI's objections as a verdict. They are a list of things to test.
Do this week
- Export last week's sales, remove customer details, and run the three-question prompt. Match the row count and grand total with your books before reading the answers.
- Take your best-selling product and work out, on a calculator, what each sale earns over cost at today's price and at one higher price, and how many sales the higher price needs to earn the same amount.
- Run the "argue against me" prompt on one decision you are considering, and write down the three data questions it leaves you with.