As of 13 August 2026, AI can analyse sales by product.

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 30 minutes until you can act on the result.

   How to actually do it

   1. Open your sales or order system and export the relevant transactions as a CSV or spreadsheet, including the product identifier, order date, quantity, price, discounts, refunds, VAT status and costs where available.
   2. Remove customer names, addresses, payment details and other personal data that the product analysis does not need, then keep a copy of the unedited export.
   3. Open the export and record what each column means, the date range, currency, whether prices include VAT, and how cancelled orders, returns and discounts appear.
   4. Paste or upload the cleaned file with those definitions into an AI data-analysis tool, then use the prompt to request product totals, rankings, comparisons and a reconciliation to the source data.
   5. Compare the reported all-product totals and formulas with spreadsheet totals made from the original export, and manually recalculate three products including one with a return or discount.
   6. Ask the AI to correct only errors you can identify from the source data, then save the final tables, charts, definitions and verification notes with the export.
   7. Send the checked analysis to the person responsible for pricing, stock or sales decisions and label any explanation or recommendation as an interpretation rather than a measured fact.

   Prompt

   Analyse the sales data I paste or upload below by product. First describe the columns you found and identify missing, duplicated or suspicious rows without silently correcting them. Ask questions before calculating if the date range, currency, VAT treatment, returns, discounts, product identifiers or sales measure is unclear.
   
   Use the source data only. Do not invent figures, products or assumptions. Report the number of units sold, order count and sales value by product where the columns support them. Keep gross sales, discounts, refunds, net sales and cost of goods separate, and state the exact formula used for every calculated measure. Show the total for all products and reconcile it to the source total where possible.
   
   Rank products by [sales value / units sold / gross margin]. Include the top and bottom products, each product's share of the chosen measure, and any meaningful change between [period 1] and [period 2] if the data supports a valid comparison. Use pounds and retain the source precision. Flag products with low volumes, missing values or results that should not be compared directly.
   
   Give me a concise findings section, followed by tables and charts if useful. Separate observations from possible explanations. Do not claim that one product caused a change unless the data proves that. End with a verification checklist listing the source columns, filters, formulas, totals and three product rows I should check manually.
   
   Data:
   [PASTE OR UPLOAD SALES DATA]
   
   Business definitions and context:
   [DESCRIBE SALES MEASURE, RETURNS, VAT, COSTS, PRODUCT GROUPS AND PERIODS]

   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.