Stats Compass

Know what's real.
Act on what matters.

Most ecommerce analytics tell you what happened.
Stats Compass tells you whether it means anything.

Every number carries a verdict: expected, real change, noise, or not enough data yet. So you know what changed, what your promotions actually earned, and where every figure came from.

Stats Compass Commerce is free during early access.

Stats Compass Commerce result for the SUMMER40 discount code, 14 May to 3 June 2026, 1,353 orders. Revenue +32%, real change: £9,144.36 more than expected, likely between +13% and +51%. Contribution margin -44%, real change: £5,920.83 less than expected, likely between -63% and -25%. New customers +153%, real change: 714 more than expected, likely between +131% and +175%.
On our demo store, SUMMER40 lifted revenue 32% and cut contribution margin 44%. Both real changes.
Works with Shopify Claude ChatGPT VS Code Cursor

Four verdicts, not a wall of charts

A number on its own can't tell you whether to act. Stats Compass judges each one against what your own history says is normal for this time of year, and says so in plain words.

Expected

Within the range this time of year usually brings. Nothing to chase.

Real change

Outside the usual range, and too large to be chance. Worth acting on.

Noise

It moved, but by no more than week-to-week swings already do.

Not enough data yet

Too little history to judge, and we tell you how much more it needs.

What changed

A weekly report with a verdict on every headline figure: revenue, orders, average order value, margin, new and returning customers. "Nothing unusual happened" is a complete report.

What your promotions actually earned

Each discount's lift in revenue and contribution margin, measured against the sales that were coming anyway, with a range and a verdict. A busy week isn't a profitable one.

How Stats Compass compares with Shopify Sidekick

Where every figure came from

Every number opens "How this was calculated": the definition, its version, and how many orders it was computed from.

Real statistics for Shopify brands

Connect your Shopify store and Stats Compass reads your orders, products and customers. It never writes to your store. Your seasons, your promotions and your customers are judged against your own history, not an industry average.

Stats Compass Commerce promotion card for SUMMER40, headed 'Needed 2.37× · Got 1.32×': to break even it needed about 2.37 times the revenue it would otherwise have had, and it brought 1.32 times. It gave away about 34% on average, and anything above about 23% lost money. Revenue +32%, contribution margin -44% and new customers +153%, each a real change with its range, above a chart of daily sales against the sales expected without the promotion.

Promotions, measured

Each promotion's lift in revenue, contribution margin and new customers, measured against the sales that were coming anyway, with the range each could fall in. The headline says what it needed to break even and what it brought, and the chart shows daily sales against what was expected without it.

See how Stats Compass compares with Shopify Sidekick on the same question.

Stats Compass Commerce weekly overview for 21 to 27 September 2026, headed 'New customers have fallen 5 weeks in a row': revenue £5,927.06 and orders 162 both marked expected, 90-day repeat rate 26.2% marked real change, and a chart of weekly revenue over 24 months against the range this time of year usually brings.

The week, judged

Each headline figure with its verdict, and this week against the range this time of year usually brings. When something is worth a look, it leads the report: here, new customers falling five weeks in a row while revenue stays within its usual range.

Stats Compass Commerce seasonality view: the underlying sales trend with the seasons taken out, up about 20% on a year ago, beside a chart of how far each month usually runs above or below the yearly average, with ranges.

Your seasons, forecast

Your underlying trend with the seasons taken out, how each month usually runs against your yearly average, and a revenue forecast for the next 90 days, each with the range it could fall in. So a busy November reads as November, not as growth.

Get early access

Free during early access. Email us and we'll get your store connected.

Use Stats Compass with your AI assistant

Ask Claude about your store's orders in conversation. Set it up once, in four steps.

  1. 1

    Connect your store

    In the Stats Compass app in your Shopify admin, click Connect an AI assistant, on the overview or in Settings. Sign in when asked, with Google or email.

  2. 2

    Add Stats Compass to Claude

    On claude.ai, add a custom connector with:

    • Name: Stats Compass
    • URL: https://mcp.statscompass.io/mcp
  3. 3

    Sign in the same way

    When Claude asks you to sign in, use the same sign-in as in step 1, for example the same Google account. A different sign-in is a different account, and won't see your store.

  4. 4

    Ask about your store

    Start with "Use stats compass to…", for example:

    • "Use stats compass to load my store's orders and chart weekly revenue for the past year."
    • "Use stats compass to find which discount codes brought in the most revenue last quarter."
    • "Use stats compass to compare last week's revenue with the same week last year."

Using ChatGPT, VS Code or Cursor? Add Stats Compass with the install options below, then sign in the same way as in step 1.

Your data, everywhere

Commerce is built on stats-compass-core, our open-source statistics library. The same engine plugs into your AI assistant through the Model Context Protocol: 50+ data tools to load, clean, chart and model any CSV, spreadsheet or Google Sheet, from Claude, ChatGPT, VS Code or Cursor.

Load data

CSV, Excel, JSON, Parquet

Visualize

Charts, histograms, heatmaps

Transform

Filter, merge, pivot, aggregate

Export

CSV, Excel, charts as images

50+ tools at your AI's fingertips

Data loading

Load CSV, Excel, JSON, and Parquet files. Get instant summaries, column info, and data type detection. Works with local files or URLs.

Visualization

Create line charts, bar charts, scatter plots, histograms, box plots, heatmaps, and more. Export as PNG or SVG.

Data transformation

Filter rows, select columns, merge datasets, pivot tables, group and aggregate, handle missing values, and more.

Analyse your Google Sheets

Ask your AI assistant to analyse a spreadsheet and pick it from your Drive. No downloading, no re-uploading, no hunting for links. Results can be written straight back to a new sheet.

1

Ask

Tell your assistant you want to analyse a Google Sheet. It hands you a link to Google's own file picker.
2

Pick

Browse your own spreadsheets and choose one or several. Every tab loads as a separate dataset, ready for charts, statistics and models.
3

Save back

Write cleaned data, summaries or predictions back to Google Sheets when you're done.

Try it

"Use stats compass to let me pick a churn dataset from Google Sheets, then build a model to predict customer churn"

Stats Compass can only open the spreadsheets you hand it in the picker, and that limit is enforced by Google, not just promised by us. It cannot list, search or open anything else in your Drive. You can disconnect at any time by asking your assistant, or from your Google Account connections. See our Privacy Policy for how this data is handled.

See it in action

Creating visualizations with Claude

Data analysis in VS Code

How to use

Start your message with "Use stats compass to...". This tells the AI to use Stats Compass tools instead of writing code or using shell commands.

# Load data Use stats compass to load ~/Downloads/sales.csv and show me a summary Use stats compass to find my CSV files in the Downloads folder # Explore and clean Use stats compass to run EDA on the dataset Use stats compass to clean the data and handle missing values # Analyse Use stats compass to test if scores differ significantly between group A and B Use stats compass to create a histogram of the price column # Machine learning Use stats compass to train a classification model to predict churn

Without this prefix, some AI clients may write Python code or use shell commands instead of the Stats Compass tools, especially for tasks like finding files on your machine.

Get started in 2 minutes

Option 1: Claude

Add Stats Compass as a custom connector on claude.ai:

1

Open Connectors

Click Customize in the sidebar, then select Connectors.

2

Add a custom connector

Click + and enter:

  • Name: Stats Compass
  • URL: https://mcp.statscompass.io/mcp
3

Log in and start analysing

Sign in with Google or email when prompted. You're ready to go.

Option 2: VS Code / Cursor

Install the extension from the Marketplace (no Python or dependencies needed):

Option 3: ChatGPT

Connect the hosted server directly from ChatGPT settings. Requires a ChatGPT paid account (Plus or above).

1

Enable developer mode

Go to Settings → Apps → Advanced settings and enable developer mode.

2

Create a new app

Go to Apps → Create app. Set the MCP Server URL to https://mcp.statscompass.io/mcp and authentication to OAuth.

3

Authorise and log in

Grant access, log in with Google or email, and you'll be redirected back automatically.

Option 4: Claude Desktop

Add Stats Compass from the connector store (no local setup required):

Claude Connector Store

Open source / self-hosted

Run Stats Compass locally with unlimited free usage. Requires Python 3.11+ and UV.

Claude Code (CLI):

claude mcp add stats-compass -- uvx stats-compass-mcp run

Claude Desktop (manual config):

{
  "mcpServers": {
    "stats-compass": {
      "command": "uvx",
      "args": ["stats-compass-mcp"]
    }
  }
}
Full documentation on GitHub

Data tools pricing

For the AI assistant data tools. Start for free, locally or in the cloud. Upgrade when you need more.

Stats Compass Commerce is free during early access.

Local

Self-hosted on your machine

Free
forever

  • Unlimited tool calls
  • Runs entirely on your machine
  • Works with Claude Desktop, VS Code & Cursor
  • All 50+ analysis tools
  • Open source
Install locally →

Hosted Free

Cloud-hosted, no setup needed

$0
/ month

  • 100 tool calls per month
  • Works with ChatGPT, VS Code & Cursor
  • All 50+ analysis tools
  • No installation required
See setup instructions ↓

Hosted Pro

Popular

For regular data work

$10
/ month

  • Unlimited tool calls
  • Everything in Hosted Free
  • Priority support
  • Cancel any time
Upgrade to Pro →

Billing handled securely by Stripe. Cancel any time.