What is WorkMagic MCP?

WorkMagic MCP lets you connect your own AI tool — as Claude or ChatGPT — directly to your WorkMagic data. Instead of logging into the WorkMagic dashboard, you can ask questions about your attribution and sales data, build dashboards, and adjust settings in plain language, right inside the AI tool you already use.

It works through the Model Context Protocol (MCP), an open standard that lets AI tools securely connect to outside data and services. You bring your own AI tool; WorkMagic provides the secure connection to your data. The MCP is free with your WorkMagic subscription.

What you can do

Ask your data — attribution and sales performance across channels, platforms, campaigns, and tactics, in any attribution model, including WorkMagic's incrementality-adjusted model.
Build dashboards — create and update dashboards and charts by describing what you want in natural languages, so that the dashboards speak to your business needs.
Pull recurring reports — for example, a weekly performance summary across channels you can re-run on demand.
Set things up — define custom metrics, group campaigns into tactics, and extract dimensions from your ad names to build naming convention rules.
Optimize budget (MMM) — run cross-channel budget-optimization scenarios and see how reallocating spend is forecast to change results.
Understand incrementality — design geo lift tests and read their results to see true incremental impact.

Before you start

You will need:

A WorkMagic account you can log into. New to WorkMagic? Book a demo to get started.
The right access in WorkMagic. The MCP works with your existing WorkMagic permissions — it can only see and do what your own login can.
Your data connected. Make sure the sales and ad platforms you want to ask about are integrated in WorkMagic and your attribution results are available — the MCP reflects what is already in your account.
An MCP-compatible AI tool. For example Claude (Desktop, Code, or web), ChatGPT, or another tool that supports custom MCP connectors.

How to connect

The WorkMagic MCP server URL is below, when you add it, you will be sent to WorkMagic to sign in and authorize the connection.

`https://mcp.workmagic.io/mcp`

In Claude

Open Settings, then Connectors (in Claude, "Customize" then "Connectors").
Click "Add custom connector".
Paste https://mcp.workmagic.io/mcp and click "Add".
Sign in to WorkMagic when prompted to authorize.

In ChatGPT

Open Settings, then Connectors, and add a custom MCP connector.
Paste https://mcp.workmagic.io/mcp.
Disable OIDC for the connection (required for ChatGPT), then authorize via WorkMagic.

Tools available

The WorkMagic MCP exposes the following tools. Your AI tool decides which to use based on what you ask — you do not need to name tools. Tools marked "makes a change" will ask you to confirm before writing anything.

Ask your data

Tool	What it does	Try asking
`knowledge-base-ask`	Grounds the AI in your WorkMagic data model (tables, metrics, definitions) before it queries. Runs automatically.	Used automatically — no need to prompt it directly.
`database-query-sql`	Answers questions about your attribution and sales data.	What was my incrementality-adjusted ROAS and CAC by channel for the last 7 days? Compare ad spend, orders, and Shopify sales across Meta, Google, and TikTok for the last 30 days. Which campaigns are below my target CAC this month, using the incrementality-adjusted model? Show daily ad spend and attributed sales for the last 14 days. Break down ROAS by tactic for Google over the last 30 days.

Dashboards

Tool	What it does	Try asking
`dashboard-list`	Lists your dashboards.	List my WorkMagic dashboards. Which dashboard tracks Meta performance?
`dashboard-get`	Opens a dashboard's charts and settings.	What charts are on my "Meta Weekly Performance" dashboard?
`dashboard-metrics-list`	Shows which metrics and dimensions are available to chart.	What metrics can I chart for ad-level attribution dashboard?
`dashboard-section-create` (makes a change)	Adds a chart to a custom dashboard you build on WorkMagic.	Add a line chart of daily ad spend versus attributed sales to my Meta dashboard.
`dashboard-create` (makes a change)	Builds a new custom dashboard with charts on WorkMagic.	Build a dashboard tracking blended ROAS by channel over the last 30 days. Create a dashboard comparing Meta, Google, and TikTok performance.

Custom metrics

Tool	What it does	Try asking
`custom-metric-list`	Lists your custom metrics.	List my custom metrics.
`custom-metric-create` (makes a change)	Creates a formula-based custom metric.	Create a custom metric for contribution margin equal to net sales minus ad spend.
`custom-metric-update-or-delete` (makes a change)	Edits or removes a custom metric.	Rename my "POAS" custom metric to "Profit on Ad Spend".

Ad tactics

Tool	What it does	Try asking
`tactic-list`	Lists your tactics and their rules.	List my ad tactics. What rule defines my "Non-Brand Shopping" tactic?
`tactic-create` (makes a change)	Groups campaigns into a named tactic by name-matching rules.	Create a tactic that groups all Meta campaigns with "Prospecting" in the name. Make a "Branded Search" tactic for Google campaigns containing "search" and "brd".
`tactic-update-or-delete` (makes a change)	Edits or removes a tactic.	Add "smart+" to the match rules for my "TT Web" tactic under TikTok.

Naming conventions

Tool	What it does	Try asking
`naming-convention-list`	Lists your naming-convention rules.	List my naming-convention rules for Meta at the ad level.
`naming-convention-create` (makes a change)	Extracts a custom dimension from ad names.	Extract the state names from my campaign names where it is the second segment split by underscore and make a "region" naming convention across my ad platforms.
`naming-convention-separators`	Gets or sets the characters used to split ad names.	What separators are set for my naming conventions?
`naming-convention-update-or-delete` (makes a change)	Edits or removes a rule.	Delete my "Audience" naming-convention rule for TikTok.

Media Budget Optimizer

Tool	What it does	Try asking
`budget-optimizer-list`	Lists your saved budget-optimization scenarios.	List my budget optimization scenarios.
`budget-optimizer-create` (makes a change)	Runs a new MMM-based budget-optimization scenario and forecasts the outcome.	Optimize my budget to maximize incremental revenue at a 4x ROAS target for next month. What if I shift 20% of Meta budget to Google — how does forecast ROAS change?
`budget-optimizer-compare`	Compares results across scenarios.	Compare my "aggressive" and "conservative" budget scenarios.

Lift tests

Tool	What it does	Try asking
`lift-test-list`	Lists your lift tests and their status.	List my lift tests. Which lift tests are currently running?
`lift-test-scan`	Scans your account to recommend a test design (geo level, holdout, method).	Can I run a Meta lift test? What design would you recommend?
`lift-test-create` (makes a change)	Creates a lift test from a confirmed design.	Set up the recommended Meta geo lift test as a draft.
`lift-test-readout`	Reads the results of a completed test (incremental lift, iROAS, confidence).	What was the incremental lift and iROAS from my Snapchat test?

Best practices

1. Be specific. WorkMagic data is rich, so the more you specify, the better the answer. Include the date range, the attribution model (for example, incrementality-adjusted, last paid click, any click), the metrics you want (spend, orders, sales, ROAS, CAC, new-customer ROAS), and the scope (all stores, Shopify only, or a specific platform, channel, or tactic). If you do not say which model, always ask which one the answer used.

2. Give your AI tool context up front. Pasting a short WorkMagic context note into your AI tool's project or instructions helps it use the right metrics and models from the start, especially for recurring reports.

3. Start broad, then drill in. Begin with a high-level question ("how is my CAC trending this month?"), then narrow ("now break that down by tactic for Meta").

4. Confirm before it writes anything. Some tools build dashboards or change settings. Your AI tool will ask you to confirm — review the change before approving.

5. Tie answers back to your dashboard. Name the metric and model you use in WorkMagic so the numbers reconcile with what you see in the platform.

FAQ

Does it work when my AI app is closed? Is this an API?

No. The MCP runs live inside your AI tool while the app is open — it answers your questions on demand. It is not a background or scheduled data feed.

What does it cost?

The WorkMagic MCP is free with your existing WorkMagic subscription. You use your own AI tool, and you pay your AI provider (for example, Anthropic or OpenAI) for your usage of that AI tool.

What can it access, and can it change my settings?

It works within your existing WorkMagic permissions — it can only see and do what your own login can. It can read your attribution and sales data, and read or build dashboards, custom metrics, tactics, and naming conventions. Anything that makes a change asks you to confirm first.

Which AI tools are supported?

Any tool that supports custom MCP connectors — for example, Claude (Desktop, Code, web), ChatGPT.

A number looks different from my dashboard.

Make sure you are asking for the same attribution model and metric as your dashboard uses, and that the relevant platform is connected to WorkMagic. Ask the AI which model and date range it used.

How is my data kept secure?

You authorize through WorkMagic's sign-in (OAuth), and access is scoped to your account on WorkMagic's side.

How do I disconnect?

Remove the WorkMagic MCP connector in your AI tool's connector settings.

What is the difference between using WorkMagic through the platform and through the MCP?

Both give you the same trusted WorkMagic data and measurement — they are simply different ways to work with it, and each has its strengths. The platform is the full visual experience: guided setup, polished dashboards, and every view we have designed for you. The WorkMagic MCP brings that same data into the AI tool you already use, which is especially valuable when you want to:

Set things up in bulk, with AI planning, drafting, and proposing first. Instead of configuring naming convention rules one by one, ask the AI to propose a full set of naming conventions or tactics at once, then review and approve — faster than by hand.
Centralize your reporting. If you already bring other data sources into your AI tool, the MCP lets WorkMagic become part of that single daily report, alongside everything else.
Build deeply customized views. For a cut of the data the platform does not surface directly, you can shape exactly the monitoring or report you want, in plain language, and have AI create a customized artifact to review data comfortably based on your business needs.

Is my data private? Is it used to train AI models?

Your WorkMagic data stays yours. The MCP only makes it available to the AI tool you connect, under your own login and permissions. WorkMagic does not use your data to train AI models. How your prompts and results are handled on the AI side is governed by your AI provider's terms, so review your provider's data settings.

Can my whole team — or my agency managing several brands — use it?

Yes. Each person connects with their own WorkMagic login, so everyone works within their own permissions. If you manage more than one WorkMagic account — for example, an agency with several brands — you can clarify to your AI tool which account to use.

How fresh is the data I get back?

The MCP reads your live WorkMagic data, so answers reflect what is currently in your account — the same data behind your dashboards. Keep in mind that the most recent days may still be settling as attribution matures.

Questions?

Reach out to your WorkMagic team — we are happy to help you get the most out of WorkMagic MCP. During the beta, please share feedback or report issues to your CSM. Not a customer yet? Book a demo.

WorkMagic MCP: Connect WorkMagic to your AI tool