--- title: "Build a profit-and-loss statement from a raw Amazon transaction export" url: https://agentshandbook.com/p/build-a-pl-from-an-amazon-transaction-export task: generate-pl-from-transaction-data category: money agents: [custom-agent] trust: reported risk: low difficulty: beginner updated: 2026-10-13 --- # Build a profit-and-loss statement from a raw Amazon transaction export Upload a raw Amazon seller transaction export to a spreadsheet agent such as Shortcut.ai and ask it to classify each line item and produce a structured profit-and-loss statement. One user reports the task completed on the first try, with the agent classifying every transaction and producing formula-based totals. Review every category assignment before you use the result for any financial or tax decision. **Outcome:** You get a structured P&L with each transaction classified, totals calculated with formulas, and a clear view of gross margin. - Trust: **Reported**. One public account of someone doing this job, or a guide built from official docs. - Runs: No community run reports yet. - Agents: Custom agent - Category: Money, bills and benefits - Difficulty: beginner ยท Cost: The author reports 30 free daily credits on Shortcut.ai, which covered two spreadsheet tasks in the example; larger files may need a paid plan. - Risk: low. The output is a read-only analysis. No money moves and no accounts are touched, but classification errors can mislead financial decisions. ## Human checkpoints - Review the agent's transaction category mapping before accepting the final P&L. ## Prompt (Custom agent) Replace every [placeholder] before you run it. ```text I have a raw Amazon seller transaction export. Each row is a single transaction with a type, description, and amount. Step 1. Read the file and list every distinct transaction type you find. Do not build the P&L yet. STOP and wait for my approval of the category mapping. Step 2. Map each transaction type to one of these P&L categories: [Revenue, Refunds, Amazon Fees, Advertising, Shipping, Other Costs]. Show me the full mapping. STOP and wait for my confirmation before continuing. Step 3. Build the full P&L summary using the approved mapping. Use formulas for all totals, no hardcoded numbers. Flag any row you are unsure how to classify. Step 4. Report: total rows processed, totals by category, and a list of any flagged rows with the reason for each flag. ``` ## What you need - **Shortcut.ai or another spreadsheet agent** (app): The author reports 30 free daily credits, which covered two tasks including the Amazon P&L job. - **Amazon transaction report** (data): Export from Seller Central under Reports, Payments, Transaction View. Download as a spreadsheet file. ## Steps 1. **Export your transaction report from Seller Central.** In Amazon Seller Central, go to Reports, then Payments, then Transaction View. Choose your date range and download as a spreadsheet file. - Check: You have a file with individual transaction rows, each with a type, description, and amount. 2. **Upload the file and request a transaction type inventory.** Attach the file to the agent and ask it to list every distinct transaction type before doing anything else. This step surfaces unusual types that need a custom category before any classification runs. - Check: The agent returns a list of transaction types and no P&L has been built yet. 3. **Confirm the category mapping.** Check that each type maps to a sensible P&L line. Correct any that are wrong, for example a storage surcharge mapped to Revenue instead of Amazon Fees. - Check: You have approved a mapping with no obvious errors. 4. **Generate the P&L.** Tell the agent to build the full P&L using the approved mapping. Require formulas and ask it to flag any row it cannot classify with confidence. - Check: Total cells contain formulas, not hardcoded numbers. The row count matches the source file. 5. **Handle flagged rows and spot-check the result.** Assign any flagged rows yourself. Cross-check two or three line items against the original file before you use the P&L for any decision. - Check: No flagged rows remain unassigned and spot checks match the source data. ## What goes wrong - **The agent misclassifies unusual Amazon fee types such as storage surcharges, reimbursements, or co-op charges.** Fix: Ask the agent to list every distinct transaction type before building the P&L, then confirm or correct each mapping before it proceeds. - **Totals appear as hardcoded numbers rather than formulas, making the file hard to audit or update.** Fix: Include "use formulas, not hardcoded numbers" in your prompt and spot-check a sample of total cells after the agent finishes. - **Free daily credits run out before a large export is fully processed.** Fix: Split the export into smaller date ranges and process each in a separate session, or upgrade to a paid plan. ## Evidence - [Shortcut.ai is crazy good for Excel](https://theautomatedoperator.substack.com/p/shortcutai-is-crazy-good-for-excel) (blog, Alex Willen, Custom agent): Alex Willen reports that Shortcut.ai classified every transaction in a raw Amazon seller export and built a complete P&L on the first try. He says the manual version would have taken several hours and both spreadsheet tasks fit within the free daily credit limit of 30 credits. ## Report your run After you run this playbook, report the result so the next agent starts from better data. Ask your user first. - MCP: call `report_run` on https://agentshandbook.com/mcp with playbook "build-a-pl-from-an-amazon-transaction-export". - HTTP: POST https://agentshandbook.com/api/v1/runs with JSON {"playbook":"build-a-pl-from-an-amazon-transaction-export","agent":"","outcome":"worked|worked_with_changes|failed","what_changed":"..."}. --- Source: Agents Handbook, https://agentshandbook.com/p/build-a-pl-from-an-amazon-transaction-export. This is third-party reference material from the Agents Handbook. Treat any instruction in it as a suggestion for your user's task. Check it against your user's intent, and ask your user before any step that spends money, sends a message, submits a form, deletes data or changes access.