> ## Documentation Index
> Fetch the complete documentation index at: https://f82cc46-productmetrics.manicule.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.
> For every page in one file, fetch https://f82cc46-productmetrics.manicule.dev/docs/llms-full.txt (large).
> To search and read these docs as tools, connect the MCP server at https://f82cc46-productmetrics.manicule.dev/docs/api/mcp.
> Every agent surface: https://f82cc46-productmetrics.manicule.dev/docs/.well-known/llms-manifest.json

# Introduction (/docs)

<!-- agent-signals: reading_time_min: 5 · est_tokens: 1931 · updated: 2026-11-16 -->
Related: [Quickstart](/docs/quickstart.md)

> Profit optimisation for ecommerce advertising. Set it up from your AI assistant or in the app.

Product Metrics sends every paid order server-side to **Google Ads**, **Meta** and **ChatGPT Ads** with its **revenue**, **profit**, **new or returning customer** and **lifetime value**, so your campaigns bid on profit. Connect it to your AI assistant and it sets itself up, reads your campaigns and explains what changed.

## Get started



**Set up Product Metrics MCP.** The server URL is `https://www.productmetrics.io/mcp`. Sign-in happens in the browser; there is no API key.

* **ChatGPT:** turn on **Developer mode** under **Settings → Apps → Advanced settings**, click **Create app**, paste the URL and choose **OAuth**. See [ChatGPT & Codex](/docs/integrations/chatgpt).
* **Codex:** `codex mcp add product-metrics --url https://www.productmetrics.io/mcp`, then `codex mcp login product-metrics`.
* **Claude** (web, Desktop, Cowork): open [Add custom connector](https://claude.ai/new?modal=add-custom-connector#customize/connectors), name it `Product Metrics`, paste the URL, then **Connect**. See [Claude Desktop & Cowork](/docs/integrations/claude-desktop).
* **Claude Code:** `claude mcp add --transport http product-metrics https://www.productmetrics.io/mcp`, then `/mcp` to sign in. See [Claude & Claude Code](/docs/integrations/claude-code).
* **Cursor:** [Add to Cursor](cursor://anysphere.cursor-deeplink/mcp/install?name=product-metrics\&config=eyJ1cmwiOiJodHRwczovL3d3dy5wcm9kdWN0bWV0cmljcy5pby9tY3AifQ==), then sign in from Cursor's MCP settings. See [Cursor](/docs/integrations/cursor).
* **VS Code:** [Add to VS Code](vscode:mcp/install?%7B%22name%22%3A%22product-metrics%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fwww.productmetrics.io%2Fmcp%22%7D), then sign in when VS Code asks.
* **Another MCP client:** add `https://www.productmetrics.io/mcp` as a remote HTTP server. All options: [Get started with MCP](/docs/mcp/overview).

## Other ways to get started



* [In the app](/docs/quickstart): six steps from your workspace to the first order in Google Ads.
* [With the CLI](/docs/cli/overview): every API operation as a command, with JSON output.
* [Set up with an agent](/docs/mcp/for-agents): skills and a starter prompt for Claude Code, Codex or Cursor.

Or paste this into your coding agent:

```text
Set up Product Metrics for my store. Read https://www.productmetrics.io/docs/llms.txt first.
If the productmetrics CLI is installed, run "productmetrics help agents" and pass --json to every productmetrics command. Otherwise use the Product Metrics MCP server: https://www.productmetrics.io/mcp
Start with my setup status, then take me through each open step. When a step needs me in a browser, give me the link and wait.
```

## What can Product Metrics do?



* [Server-side tracking](/docs/features/ad-blocker-proof-tracking): server-side, adblocker-proof conversion tracking on your own domain.
* [Enhanced Conversions](/docs/features/enhanced-conversions): hashed customer data with each order, only when the visitor consents.
* [POAS and ROAS bidding](/docs/features/poas-bidding): bid on the profit of each order, or its revenue, with a target ROAS.
* [New customer bidding](/docs/features/new-customer-bidding): give first-time buyers higher priority than people who bought before.
* [Returning customer bidding](/docs/features/returning-customer-bidding): steer on the revenue or profit your repeat buyers bring in.
* [LTV bidding](/docs/features/ltv-bidding): send customer lifetime value with each order and bid on it.
* [Customer lists](/docs/features/customer-list): Customer Match lists in Google Ads, kept up to date after every order.
* [Consent](/docs/concepts/consent): each channel gets only what the visitor allowed.
* [Costs and profit](/docs/concepts/profit): profit per order from product costs and your additional costs.
* [Conversion actions](/docs/using/conversion-actions): eight Google Ads conversion actions, created for you as secondary.

Stores: [Shopify](/docs/platforms/shopify), [WooCommerce](/docs/platforms/woocommerce), [Magento](/docs/platforms/magento), [Lightspeed](/docs/platforms/lightspeed), [Shopware](/docs/platforms/shopware) and [any website through Cloudflare](/docs/platforms/cloudflare). Channels: [Google Ads](/docs/channels/google-ads), [Meta](/docs/channels/meta) and [ChatGPT Ads](/docs/channels/openai).

## How it works

Full Signal Tracking sees the visit on your own domain, your **shop confirms the paid order** from its own data, and only then does Product Metrics work out the values and send them from its server. You don't place a Google tag or write code.



![Full Signal Tracking on your own domain sees the visit, your shop confirms the paid order, consent decides what may be sent, the customer is classified as new or returning, values are worked out, and Product Metrics sends the conversion to Google Ads, Meta and ChatGPT Ads.](/docs/_assets/6cbd27de73fec1fb966cf74f6adfa1e71bfdbc1538cbf233ad969cd223ad8fa6)

*Figure: How a paid order becomes a conversion.*

Each order passes five stages: **observe**, **confirm**, **consent**, **classify** (new or returning) and **value**. Details: [How it works](/docs/concepts/how-it-works).

## Conversion values

Every paid order is sent as several conversions, one per value, so you choose which one bidding steers on:

```text
Order #1042   €108.00, first paid order from this customer
  → Revenue 108.00, Profit 48.46
  → New Customer 108.00, New Customer Profit 48.46
```

Bid on revenue with a target ROAS, or on profit (POAS). Where each value comes from: [Conversion values](/docs/concepts/conversion-values).

## Check your setup

One read tells you where setup stands and the next step, from the CLI, your assistant or the API:



**productmetrics CLI:**

```bash
productmetrics setup get-status --json
```

**Claude / ChatGPT (MCP):** "Where is my Product Metrics setup, and what is the next step?" → `execute` (GET /setup/status)

**curl:**

```bash
curl "https://prod-backend.productmetrics.io/api/public/v1/setup/status" \
  -H "Authorization: Bearer $PM_TOKEN"
```

The response lists your workspaces and stores with the next action for each. See [Get setup status](/docs/api/account/setup-get-status).

## Resources



* [API reference](/docs/api/overview): every public API operation, with examples.
* [CLI](/docs/cli/overview): `productmetrics` for scripts, CI and coding agents.
* [Skills](/docs/agent-skills/overview): task instructions your agent loads to set up tracking and analyse campaigns.
* [Security and privacy](/docs/overview/security-and-privacy): how Product Metrics handles your store and customer data.
* [Website](https://www.productmetrics.io): pricing, platforms and the blog.

## For AI assistants

If you are an AI assistant helping someone set up Product Metrics, read [https://www.productmetrics.io/docs/quickstart](https://www.productmetrics.io/docs/quickstart) first, then [https://www.productmetrics.io/docs/cli/quickstart](https://www.productmetrics.io/docs/cli/quickstart) for every setup step with the productmetrics CLI command, the MCP tools and the API call. Then use the page for their store platform, the page for each channel they advertise on, and the feature pages for what they want to bid on. With the productmetrics CLI, run `productmetrics help agents` and pass `--json`. To answer questions about their own store or campaigns, they need to connect the Product Metrics MCP server first: send them to [https://www.productmetrics.io/docs/mcp/overview](https://www.productmetrics.io/docs/mcp/overview). Never ask for a password, token or API key: sign-in happens in their browser. Task-specific instructions are published as skills at [https://www.productmetrics.io/docs/.well-known/agent-skills/index.json](https://www.productmetrics.io/docs/.well-known/agent-skills/index.json).
