Google Ads MCP: Connect Claude and ChatGPT

Ad network MCP guide

Connect Google Ads to Claude or ChatGPT with Lemon AI’s unified MCP and API. Distinguish allocated asset value from post-install ROAS.

To analyze Google Ads in Claude or ChatGPT, connect a reporting tool through MCP, then ask it to retrieve a specific account, date range, and set of metrics. Google's own MCP server gives agents direct access to Google Ads reporting. Lemon AI provides another route: connect Google Ads alongside your other networks, then use one MCP to query creative performance and the attributes attached to each asset.

Copying a report into chat still works for a one-off investigation. The difficulty comes on the next question: a new date range, a different country, or the creative behind a row. A connected assistant can fetch that evidence instead of asking you for another export.

Choose direct Google reporting or a unified creative view

Google's MCP documentation describes account discovery, resource metadata, and GAQL reporting. Its documented release is read-only. Use that route when your question needs a Google-specific resource; do not assume Lemon exposes every Google Ads API field.

For the direct route, configure the Google Cloud project and credentials required by Google's current guide, connect the server to your MCP client, and first list accessible customers. A locally running server and a remotely hosted server have different setup paths. ChatGPT's web connection needs a supported remote connection, not a local command pasted into its Server URL field.

Connect Google Ads to Lemon

1. Sign in to Lemon and open Connections.
2. Choose Google Ads, complete Google's authorization, and select the accounts you want to analyze. Use the Google identity that can access the intended account.
3. Allow the initial sync to finish. Open Creative Analytics and confirm that the campaign and dates you intend to query contain data.
4. Open API & MCP and use the server URL below to connect your assistant. Network authorization and assistant authorization are separate steps.

Lemon organizes the synced data into creative-level reporting and makes assigned creative attributes available to the agent. That lets a follow-up move from “which video?” to “which hook or format?” without assembling another spreadsheet. See how attribute analysis works.

Add Lemon to Claude or ChatGPT

Use the same remote MCP server for all the networks you connect in Lemon:

https://api.lemon-ai.com/mcp

In Claude, open Customize → Connectors → + Add → Add custom connector. Enter a name and the URL, continue, and sign in to Lemon. If Claude asks how to register the OAuth client, choose Register automatically. Enable the connector in the conversation. Organization-managed accounts may need an owner to add it first. See the Claude setup instructions.

In ChatGPT, open Plugins, choose Add custom MCP server, and enter the URL. Select OAuth authentication, authorize Lemon, and install the resulting personal plugin. Mention it in a Work chat to use its tools. Availability depends on your workspace settings; follow the current ChatGPT MCP instructions if the interface differs.

Start by asking the assistant to call list_ad_accounts and let you select the intended account. Then use list_filter_values to discover dates and filters. Your ad-network sign-in connects data to Lemon; this second authorization lets the AI app read it. Do not paste network credentials into chat.

Ask for an asset comparison with a defined scope

Start with delivery and add outcomes only after discovering the available windows:

Use Lemon AI to list my Google Ads accounts and available date ranges. Let me choose the account and App campaign. For two complete weeks, return each video asset's impressions, clicks, cost, and CTR. Show the exact dates and filters. Flag truncated results and missing fields before interpreting the ranking.

The relevant tool sequence is list_ad_accounts, list_filter_values, then query_creative_performance. The account ID comes from discovery; it is not the Google customer ID copied out of Ads Manager. The worked MCP creative-performance workflow shows the request structure.

For a cross-network question, keep the account scopes separate initially. “Google versus Meta” can otherwise hide different countries, operating systems, creative formats, or attribution bases under one summary.

Know what an allocated asset outcome means

A Google ad can contain several assets. Lemon allocates Google conversion-lag outcomes from matched ad totals to assets. An allocated value is useful for inspection, but it is not a separate conversion event observed on that asset.

Synthetic illustration, not customer data or an API response: suppose one matched ad has $400 in conversion value for a particular lag window. Its eligible assets receive the following allocation:

The sum preserves the matched ad's $400 in this synthetic example. It does not establish that removing the end card would reduce revenue by $100. The two elements may have appeared together, and allocation is not a randomized test of each element's contribution.

Lemon's Google outcome suffixes describe conversion lag, such as lag_lt_8d. Do not rename that window “D7 post-install ROAS.” Discover the suffix and inspect maturity before comparing it with another provider. The attribution-window guide explains why the clocks matter.

Turn the comparison into a creative question

Two useful follow-ups keep the evidence attached to the answer:

For the same Google account, dates, and video population, list the available creative attributes. Use attribute_insights to investigate CTR by hook. Show coverage, qualified creative counts, uncertainty, and creative examples. Describe an association, not a proven cause.

Recheck the apparent winning hook within the same campaign and country. Does the difference persist, or did a change in delivery mix explain it? Propose one controlled creative test and state what result would contradict the recommendation.

An attribute must exist and have sufficient coverage before it can answer the question. The model should not infer a hook from a filename and present that guess as a stored label.

Use the same reporting layer through the API

For a script or another AI app, create a Lemon API key in API & MCP. Retrieve your account IDs with GET /v1/ad-accounts, then make a scoped read. This example uses a placeholder account and dates; substitute values returned for your account.

curl --get 'https://api.lemon-ai.com/v1/creatives/performance' \
  --header "Authorization: Bearer $LEMON_API_KEY" \
  --data-urlencode 'ad_network_account_id=YOUR_ACCOUNT_ID' \
  --data-urlencode 'date_from=2026-10-01' \
  --data-urlencode 'date_to=2026-10-07' \
  --data-urlencode 'metric=impressions' \
  --data-urlencode 'metric=clicks' \
  --data-urlencode 'metric=cost' \
  --data-urlencode 'sort_by=cost' \
  --data-urlencode 'sort_order=descending' \
  --data-urlencode 'limit=25'

MCP and REST expose Lemon's reporting definitions. Inspect missing fields, result limits, and any outcome metadata before interpreting a result. A limited top-creatives response is not an account total. Use the API documentation for supported fields and pagination.

When the result does not match Google Ads

First align account, dates, timezone, currency, campaign, and reporting grain. Compare the same outcome definition rather than a platform-wide “Conversions” total with an allocated asset window. A recently completed sync can still contain an immature conversion window.

If the account is absent, check the Google identity used for authorization and account selection in Lemon. If the account exists but the chosen campaign is absent, discover available filters before changing the query. If the response is truncated, narrow it before declaring an overall winner.

For a repeatable starting point, connect your ad accounts, then use the reporting API and MCP overview to choose the interface your team needs.