Best Copilot Data Connectors for Marketing Data in 2026
Microsoft 365 Copilot can already read your team's emails, files and Teams chats. Your Google Ads, Meta Ads, GA4 and CRM data sit outside Microsoft 365, so Copilot needs a connector to answer questions about spend, conversions or pipeline.
Copilot data connectors come in two groups. Microsoft builds and approves the first group, which you manage in the Microsoft 365 admin center or add in Copilot Studio. Third-party platforms provide the second group and cover the ad and analytics sources that Microsoft doesn't. This guide explains what Microsoft's options can and can't do with marketing data, then compares seven third-party tools built for it.
What are Copilot connectors?
Copilot connectors bring external data into Microsoft 365 Copilot and Microsoft Search. They were previously called Microsoft Graph connectors, so older docs also say copilot graph connectors or "graph connectors in Microsoft 365 Copilot".
The Copilot connectors gallery in the Microsoft 365 admin center lists more than 100 prebuilt connectors. These Microsoft 365 Copilot connectors cover Google services, Azure services and dozens of SaaS apps, and developers can build custom connectors for other sources.
For marketing, the copilot connectors list is thin: there's no Google Ads, GA4, Meta Ads, LinkedIn Ads, TikTok Ads or Shopify connector. Microsoft copilot connectors were built for documents, tickets and CSV files, and synced content goes through semantic indexing. Semantic indexing can find a campaign brief but can't calculate "spend by campaign last month," so marketing teams usually add third-party Copilot data connectors.
Microsoft 365 Copilot connectors: synced vs federated
Microsoft 365 Copilot connectors come in two models. They differ in where the data lives, which licenses you need, and which marketing sources they cover.
Synced connectors
Synced connectors copy content from the source into Microsoft Graph and index it. An admin sets a crawl schedule, from every 15 minutes to every 12 hours, with periodic full crawls to pick up permission changes. Copilot can then cite the indexed items in answers, and the same content shows up in Microsoft Search.
Only the title and content properties go through semantic indexing, so a well-structured source returns better answers. Indexing is included with Microsoft 365 licenses. Microsoft Search works with synced content on any Microsoft 365 plan, but grounding Copilot answers in it requires the Microsoft 365 Copilot add-on or Microsoft 365 E7.
Microsoft-built synced connectors that marketing teams use include:
- Salesforce CRM, for contacts, leads, accounts and opportunities
- Google Drive and Dropbox, for campaign briefs, creative files and reports
- CSV, for ad platform exports you save on a schedule
- Jira and GitLab, for marketing ops and website tickets
- MediaWiki, for brand guidelines and internal wikis
The gallery also covers ServiceNow and Azure SQL, which are usually IT or data team sources.
Federated connectors
Federated copilot connectors don't copy data. They use the Model Context Protocol (MCP) to request data from the source when a user asks a question, and each user signs in with their own credentials, usually through OAuth 2.0. The source system enforces its own permissions.
Microsoft made federated connectors generally available in spring 2026. For marketers, the most relevant ones are HubSpot, Canva, Notion and Intercom, and the analytics category also lists Mixpanel, Contentsquare and Polar Analytics. They work in Copilot Chat, agents, the Researcher agent, and Copilot in Word, Excel, PowerPoint and Outlook.
This type requires the Microsoft 365 Copilot add-on or E7 and doesn't work with Copilot Studio or pay-as-you-go licenses. Write actions, such as updating a record, started rolling out in October 2026, and they ask the user for approval by default.
| Synced | Federated | |
|---|---|---|
| Where data lives | Copied into Microsoft Graph | Stays in the source |
| How Copilot gets it | Semantic indexing and search | MCP request at query time |
| Authentication | Microsoft Entra ID app or service account | User's own sign-in (OAuth or service-specific) |
| Marketing examples | Salesforce CRM, Google Drive, CSV exports | HubSpot, Canva, Mixpanel |
| License for Copilot use | Copilot add-on or E7 | Copilot add-on or E7 |
Copilot Studio connectors and MCP
Copilot Studio is Microsoft's tool for building custom agents, and it's where most marketing data reaches Copilot. Its connectors work differently from the Microsoft copilot connectors you manage in the admin center.
Power Platform connectors. Copilot Studio connectors are the same ones that Power Automate and Power Apps use. Microsoft lists over 1,500 Power Platform connectors, split into standard and premium tiers. They call APIs on request, so an agent can read a record, create a lead or trigger a flow. The full copilot studio connectors list is in Microsoft's connector reference. Some premium connectors depend on your Copilot Studio plan, and Power Platform data policies can block them.
Copilot Studio custom connectors. If an ad network or martech tool has no prebuilt connector, you can wrap its API in a custom connector with an OpenAPI definition. This usually takes a developer, or at least someone comfortable with API authentication.
Copilot Studio MCP connectors. Copilot Studio can add any remote MCP server as a tool, under Tools > Add a tool > New tool > Model Context Protocol. Most third-party Copilot MCP connectors for marketing data work this way. You paste the vendor's server URL, choose an authentication method, and the agent can query your ad, analytics and CRM data through that vendor. You can then publish the agent to Microsoft 365 Copilot and Teams, if your admin allows that channel.
Agents and plugins. A declarative agent is a Copilot agent built from instructions and knowledge sources, with no custom code, such as a connector agent that only answers from your brand guidelines. API plugins let an agent take actions in other apps.
Best Copilot data connectors for marketing teams
The seven Copilot data connectors below document a connection for marketing sources such as ad platforms, GA4 and CRMs. Prices were checked on vendor sites in October 2026. Most of these tools connect through Copilot Studio, so whoever sets them up needs maker access there, which an admin usually grants.
Coupler.io
Coupler.io is a data integration platform and AI analytics solution with 400+ sources, including Google Ads, Meta Ads, GA4, HubSpot and Salesforce. Describe the report you need, such as campaign spend and conversions across Google Ads and Meta Ads, and Coupler AI helps set up a data flow that combines those sources. You then add the Coupler.io MCP server as a tool in Copilot Studio.
Your agent answers from that one refreshed dataset, so cross-channel questions like ROAS by channel or cost per lead get a consistent answer. There's a free plan, and paid plans start at $24/month billed annually.
Best for: marketing teams that want Copilot answers based on several ad, analytics and CRM sources combined into one prepared dataset.
Supermetrics
Supermetrics is a marketing data platform with 170+ sources. Its Marketing Analyst agent runs inside Microsoft 365 Copilot and Teams, and users add it from All agents in Copilot. Admins can deploy it to selected users or the whole organization through the Teams admin center. Microsoft Copilot is one of the core destinations you can pick on the Starter plan, at $44/month billed yearly for 3 sources. There's a 14-day trial.
Best for: marketing teams that work in Teams and want a ready-made agent with no Copilot Studio setup.
Windsor.ai
Windsor.ai is a self-serve marketing data tool with 350+ sources. You add Windsor MCP to a Copilot Studio agent with your API key. Windsor.ai is also listed as a Power Platform connector, currently in preview. All plans include MCP access. The free plan covers 1 source and 1 account, and Basic costs $19/month billed annually.
Best for: small teams and agencies that want to test Copilot on their ad data at low cost.
Improvado
Improvado is an enterprise marketing analytics platform. Its MCP documentation lists 1,400+ sources. It connects to both Copilot Studio, as an MCP tool, and Microsoft 365 Copilot. There's a free limited plan and an MCP Only plan at $100/month. Advanced and Enterprise plans are priced by quote.
Best for: large marketing teams and agencies that already use Improvado for reporting.
Funnel
Funnel is a marketing data hub that collects and standardizes data from 600+ channels. Funnel MCP connects to a Copilot Studio agent through OAuth credentials, which you create as a system user in Funnel. Its tools are read-only and run with the signed-in user's permissions. Funnel MCP is available to Funnel customers, and the Business plan starts at $600/month billed annually.
Best for: teams that already keep standardized marketing data in Funnel.
Dataslayer
Dataslayer connects 50+ marketing sources, such as Google Ads, Meta Ads and GA4, to spreadsheets, BI tools and AI assistants. It works as a remote MCP server in Copilot Studio, and users sign in with their Dataslayer account. Pricing is based on users and connectors. Paid plans start at $35/month billed annually, and there's a free plan and a 15-day trial.
Best for: small teams that report on a few core ad and analytics sources.
CData Connect AI
CData Connect AI gives Copilot Studio one remote MCP endpoint for hundreds of business systems. For marketing teams, that covers CRM and marketing automation sources such as Salesforce and Salesforce Marketing Cloud, plus ops sources like Jira and ServiceNow. Data stays in the source and is queried on request. Paid plans start at $79/month billed annually, and there's a free trial.
Best for: teams that need Copilot to query CRM and marketing automation data alongside sales or finance systems.
Copilot marketing data connector comparison
| Tool | Sources | How it connects | Free option | Entry price |
|---|---|---|---|---|
| Coupler.io | 400+ | MCP in Copilot Studio | Free plan; 7-day trial | $24/mo billed annually |
| Supermetrics | 170+ | Microsoft 365 Copilot agent | 14-day trial | $44/mo billed yearly |
| Windsor.ai | 350+ | MCP in Copilot Studio; Power Platform connector (preview) | Free plan, 1 source | $19/mo billed annually |
| Improvado | 1,400+ | MCP in Copilot Studio or Microsoft 365 Copilot | Free limited plan | $100/mo (MCP Only) |
| Funnel | 600+ | MCP in Copilot Studio | None listed | $600/mo billed annually |
| Dataslayer | 50+ | MCP in Copilot Studio | Free plan; 15-day trial | $35/mo billed annually |
| CData Connect AI | Hundreds | MCP in Copilot Studio | Free trial | $79/mo billed annually |
How to add Copilot connectors
The steps depend on whether you use a prebuilt connector from Microsoft or a third-party MCP server for your ad and analytics data.
From the Microsoft 365 admin center
You need the AI administrator role for these steps.
- Open the admin center and go to Copilot > Connectors.
- On the Gallery tab, pick a data source, for example Salesforce CRM or Google Drive.
- Name the connection the way your team will refer to it, add the source URL, and choose an authentication method.
- Roll it out to a test group first, then select Create. Indexing starts with default settings.
- Add a description that says what the content is and when people use it. Copilot uses this description to decide when to search the source.
Federated connectors such as HubSpot appear in the same place. Admins can enable them, limit them to Entra ID groups, or turn them off.
In Copilot Studio, with an MCP server
Here's the process with Coupler.io. Other Copilot Studio data connectors that use MCP follow a similar pattern.
- In Coupler.io, describe the data you need, for example: "Combine Google Ads and LinkedIn Ads spend by campaign for the last 90 days." Coupler AI helps set up the data flow. Choose Microsoft Copilot Studio as the destination.
- In Copilot Studio, open your agent and check that generative orchestration is on. It's the default for new agents.
- Go to Tools > Add a tool > New tool > Model Context Protocol, and paste the MCP server URL from Coupler.io.
- Set authentication to OAuth 2.0 with dynamic discovery, then create the connection.
- Run the data flow, set a refresh schedule, and ask the agent a question. The first time, each user approves the connection in the chat.
Which connector to start with
| If your team... | Start with |
|---|---|
| Wants to ask Copilot about spend, conversions and pipeline in one chat | Coupler.io |
| Works in Teams and wants a ready-made agent without Copilot Studio | Supermetrics |
| Wants to test Copilot on a few ad accounts with a small budget | Windsor.ai or Dataslayer |
| Already reports in Funnel or Improvado | That platform's MCP server |
| Needs CRM and marketing automation data next to sales or finance systems | CData Connect AI |
| Uses HubSpot as its CRM and has Copilot licenses | Microsoft's HubSpot federated connector |
| Keeps briefs, creative files and brand guidelines in Google Drive or Dropbox | A synced connector from Microsoft's gallery |
Most third-party Copilot data connectors offer a free plan or trial. Connect one real ad account, ask the agent the questions you normally answer in a spreadsheet, and check the numbers against the source before you roll the agent out to your team.
FAQ
What's the difference between Microsoft 365 Copilot and Copilot Studio?
Microsoft 365 Copilot is the assistant your team uses in Copilot Chat, Teams and Office apps. Copilot Studio is where you build custom agents and give them tools, such as an MCP server for your marketing data. An agent you build in Copilot Studio can be published to Microsoft 365 Copilot and Teams, so your team uses it in the same place.
Do third-party connectors need a Microsoft 365 Copilot license?
Not always. M365 Copilot connectors from the gallery need the Copilot add-on or E7 for Copilot answers. A Copilot Studio agent follows Copilot Studio licensing, so access depends on how your organization licenses and publishes agents. Check with your Microsoft admin before you pick a tool.
How do I check that Copilot's marketing numbers are right?
Ask the agent a question you already know the answer to, such as last month's spend in one ad account, and compare it with the platform's own report. Check the date range, currency and attribution window too, since ad platforms and connectors can use different defaults.