As of 13 August 2026, AI can forecast stock demand for your next supplier order.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Do it yourself

   Second choice

   A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open your sales or till system and export dated sales by product for the most useful recent history, keeping returns and cancelled orders identified.
   2. Export current stock, stock locations and outstanding purchase orders, including expected delivery dates, and record which incoming stock is already confirmed.
   3. Gather the supplier catalogue or latest price list and copy in lead times, order days, minimum order quantities, case sizes, pack sizes and shelf-life rules.
   4. Write down upcoming promotions, closures, seasonal events, one-off contracts, storage limits and the stock-cover or safety-stock rule you normally use.
   5. Paste the cleaned tables and notes into the prompt, replacing every bracketed slot, and ask the model to produce the low, central and high scenarios and order recommendation.
   6. Recalculate a sample of the model's demand, incoming-stock deduction and pack-size rounding in your spreadsheet, then compare every recommendation against your sales export and current stock report.
   7. Change any assumption that does not match your business, remove products flagged as unreliable from automatic ordering, and send the final quantities to the supplier only after checking availability, budget and delivery dates.

   Prompt

   Act as a cautious stock-planning analyst for my UK business. Forecast demand for my next supplier order using only the data I paste below. Do not invent missing figures, trends, promotions, seasonality or supplier information. If a required input is missing, say exactly what is missing and continue only with clearly labelled assumptions.
   
   Use these inputs:
   - Product or SKU list: [paste]
   - Historical sales by product and date or week: [paste]
   - Current stock by product: [paste]
   - Stock already ordered but not received, with expected delivery dates: [paste]
   - Supplier lead time: [paste]
   - Supplier order days or next available delivery date: [paste]
   - Minimum order quantities, case sizes and pack sizes: [paste]
   - Shelf life or expiry constraints: [paste]
   - Target stock cover, safety-stock rule or service-level requirement: [paste]
   - Known promotions, closures, events or unusual sales periods: [paste]
   - Budget or storage limits: [paste]
   
   First check the data for missing values, duplicate periods, inconsistent units and products that cannot be forecast reliably. Then produce:
   1. Expected demand for each product over the period covered by the next order and until the next realistic replenishment.
   2. The method used for each product, in plain English.
   3. A low, central and high demand scenario, with every assumption shown.
   4. Recommended order quantities after deducting usable stock and confirmed incoming stock, rounded to the stated pack or case size.
   5. Products to expedite, reduce, omit or investigate, with a reason.
   6. A short list of checks I must complete before sending the order.
   
   Use the same units as the input data. Separate calculations from judgement. Flag products with sparse, changing or unreliable history instead of presenting false precision. Do not claim to know current supplier availability, prices or delivery performance unless I provide them.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

3. Hand it to a person

   The distant third

   A person who owns the outcome does this end to end, worth it when the failure is dear.