How to use AI to build a simple weekly report from your sales data that tells you what's working and what isn't

Your sales data is already telling you what's working. You're not reading it — and you're not alone. Most small business owners make inventory and staffing calls based on gut feel rather than the numbers sitting in their Square or Shopify dashboard. This post walks you through a repeatable weekly system to analyze sales data for your small business — export your data, paste it into an AI, get a plain-language report in under 15 minutes. It works because AI can parse tabular data and return structured analysis without you needing to know a formula, a dashboard, or a chart.

What You Need Before You Start

ChatGPT{:target="_blank"} (GPT-4o): The AI you'll use to analyze your data. Free tier works with pasted data. Paid tier ($20/month as of mid-2025) supports direct CSV file upload, which reduces formatting issues. Check OpenAI's pricing page{:target="_blank"} for current plans.

Alternatively, Claude{:target="_blank"} (3.7 Sonnet): Strong at structured data and consistent output formatting. Free tier available. Particularly reliable if you want the same report structure every week.

Or Gemini{:target="_blank"} (2.0 Flash): Google's current fast model. If you already track sales in Google Sheets, Gemini integrates natively via Google Workspace — meaning you may be able to analyze your data without any export step at all.

Your sales export: A CSV or plain-text export from Shopify{:target="_blank"}, Square{:target="_blank"}, or a Google Sheet you maintain manually. Instructions for each are in the next section.

Time required: 15–20 minutes per week once the system is set up. Initial setup (first run) takes 30–45 minutes.

Skill level: No coding required. Basic comfort copying and pasting text. If you've ever exported a bank statement, you can do this.

---

Get Your Sales Data Out of Square, Shopify, or a Spreadsheet

The data you need exists. The obstacle is usually knowing where to click.

From Square (free tier):

1. Open your Square Dashboard{:target="_blank"} and click Reports in the left menu. You should see a list of report types.
2. Select Sales Summary, then set the date range to the past 7 days. You should see a table of daily totals, top items, payment types, and refunds.
3. Click Export and choose CSV. You should see a file download to your computer.

From Shopify:

1. Open your Shopify Admin{:target="_blank"} and navigate to Analytics > Reports. You should see a list of pre-built report types.
2. Open Sales by Product and set the date range to the past 7 days. You should see a table with product names, quantities sold, and revenue.
3. Click Export in the top right. Choose CSV for Excel. You should see a file download begin.

From a Google Sheet (service businesses):

If you don't use a POS system, you need four columns at minimum: Date, Service Type, Amount, Payment Method. If your sheet has these, you're ready. No export step required — you'll copy directly from the sheet. If you use Gemini, you can run the analysis directly inside Sheets without leaving the document.

This step matters because the quality of your AI sales analysis depends entirely on the quality of the data you paste. Incomplete exports produce incomplete reports. Check your date range before moving on.

---

How to Analyze Small Business Sales Data With AI: Run Your First Report

This is where the actual work happens — and it's mostly about how you ask.

1. Open your CSV file in Excel, Google Sheets, or a plain text editor. You should see rows of sales data with headers across the top row.
2. Select all the data — including the header row — and copy it. You should have the full table in your clipboard.
3. Open ChatGPT, Claude, or Gemini in your browser and start a new conversation.
4. Paste the following prompt, then paste your data directly below it in the same message:

Paste this prompt, then paste your data below it:

• I run a [type of business — e.g., retail gift shop, coffee shop, online apparel store].
• The data below covers sales from [start date] to [end date].
• Please analyze this data and tell me:
  1. Total revenue for the week and how it compares to any prior period in the data
  2. My top 5 products or services by total revenue
  3. My 3 slowest-selling products or services
  4. Which day of the week had the highest and lowest sales
  5. Average transaction value
  6. Any refunds or voids and their total value
  7. One thing I should consider changing based on this data
• Format your response as a bullet-point summary I can read in under 3 minutes.
• Do not include customer names or personal information in your response.
• [Optional: Note any promotions or unusual events that week, e.g., "We ran a 20% off sale on Thursday."]

5. Send the message. You should see a structured plain-language report returned within 30 seconds.

The specificity of the prompt is what separates useful output from a vague summary. Vague prompts produce vague reports. The seven numbered questions give the AI a clear output checklist.

---

Reading Your Report: What to Act On and What to Set Aside

The AI will return a report. Not all of it deserves equal attention.

Act on: Day-of-week patterns. If Tuesday is consistently your lowest revenue day, that has scheduling and staffing implications. Top and bottom performers by revenue. If your slowest three items appear week after week, that's a reorder and shelf-space decision.

Cross-check before acting: Revenue numbers are not profit numbers. The AI does not know your cost of goods. A high-revenue product with thin margins may matter less than a lower-revenue item with strong margins. Raise this in a follow-up prompt if you know your margins.

Set aside: One-week anomalies the AI flags as significant. If you had a large wholesale order or a holiday, that week's data will skew the averages. Tell the AI about it in your prompt — it will only factor it in if you mention it.

A dry truth about AI analysis: it will tell you what the data shows. It will not tell you whether the data is telling the whole story. That judgment stays with you.

---

When Something Goes Wrong

The AI returns a response but the numbers look wrong. Most likely cause: the formatting broke during copy-paste and the AI misread column alignment. Fix: paste your data into a plain text editor first, verify the columns look clean, then re-paste. If you're on ChatGPT paid tier, upload the CSV file directly instead — this eliminates formatting errors.

The AI says it can't read the data or returns only a general response. Most likely cause: the pasted data lost its structure and arrived as a wall of text. Fix: open the CSV in Google Sheets, copy from there rather than from the raw file, and paste again. Sheets preserves column structure better than some text editors.

The report is too general to be useful — no specific products named, just totals. Most likely cause: your export didn't include product-level detail, only daily summaries. Fix: go back to your Square or Shopify dashboard and pull the product-level report specifically (Square: "Item Sales" report; Shopify: "Sales by Product" report) rather than the summary view.

---

What to Do Next

Save your prompt as a note — in Apple Notes, Notion, or a Google Doc. Title it "Weekly Sales Report Prompt." Every Monday morning, export your data, open the note, paste the prompt, paste the data, send. The prompt is your template. The report is the output.

Once you've run this three weeks in a row, you'll have enough data to ask the AI to compare weeks — which is when the analysis starts to get genuinely useful.

If you want to extend this into expense tracking, read how to build a simple weekly financial snapshot using AI.

---

FAQ

How do I analyze sales data for a small business without an accountant? Export a week of sales data from Square, Shopify, or a Google Sheet, then paste it into ChatGPT, Claude, or Gemini with the prompt in this post. The AI returns revenue totals, top and bottom performers, day-of-week patterns, and one recommendation — all in plain language. No formulas or dashboards required.

Can I use this method to analyze sales data if I don't use Square or Shopify? Yes. Any data you can organize into rows and columns works. A Google Sheet with four columns — date, item or service, amount, payment method — gives the AI enough to produce a meaningful report. Google Sheets{:target="_blank"} is free. You don't need a POS system to use this workflow.

How do I read my Shopify sales report without getting lost in the numbers? Pull the "Sales by Product" report for the past 7 days and export it as a CSV. Paste it into the AI prompt in this post. The AI will surface the five numbers that actually matter — top products, slowest sellers, best day, average transaction, and refunds — without you needing to interpret the raw table yourself.

Is it safe to paste my sales data into ChatGPT or Claude? Product names, SKUs, revenue totals, and quantities are safe to paste. Do not paste customer names, email addresses, or any other personally identifiable information. Aggregate sales data — the kind you'd share with an accountant — is appropriate. PII is not. Both OpenAI's privacy policy{:target="_blank"} and Anthropic's privacy policy{:target="_blank"} describe how submitted data is handled.

Why doesn't AI just connect to my Shopify or Square account directly? The standard chat interfaces for ChatGPT, Claude, and Gemini do not have live access to external accounts. You export and paste manually. This is a firm limitation of the current tools, not a setup error. Some third-party integrations and API configurations can automate this, but they require technical setup beyond a standard chat interface.

How do I know if the AI's analysis is accurate? Spot-check two or three numbers against your source export before you rely on the report. Match the total revenue figure and one product's sales volume. If those align, the rest of the analysis is likely sound. If they don't, the paste formatting likely broke — see the troubleshooting section above.

What's the difference between using ChatGPT versus Claude for this? Both handle tabular data well. Claude 3.7 Sonnet{:target="_blank"} tends to produce more consistent formatted output week over week, which matters if you want the same report structure each time. GPT-4o's paid tier supports direct CSV file upload, which removes copy-paste formatting issues entirely. If you're on a free plan, either works — or try Gemini if you already live in Google Sheets. If you run into formatting problems repeatedly, GPT-4o paid with file upload is the cleaner path.