Build a Conversational Meta Ads Reporting Assistant with n8n

Input: chat messages; Processing: AI Agent calls Facebook Graph API to query ad accounts, campaigns, and insights; Output: conversational reports and account structure

Who

Marketers and advertisers who need quick access to Meta Ads data

You get

Query ad account lists, campaign details, and performance metrics (spend, impressions, conversions, etc.) via chat

You need

n8n

Steps

1. Configure AI Agent and chat trigger

   After importing the template into n8n, configure the Chat Trigger node (When chat message received) as the entry point. Set the AI Agent node's system message to: =You are a helpful Meta ads assistant. Current date is: {{ $today.format('yyyy-MM-dd') }} Connect the OpenAI Chat Model node, select model gpt-5-mini, and bind your openAiApi credential. Connect the Simple Memory node to maintain conversation context.

   Execution prompt · Source text。AI Agent node Options > System Message field

   Copy prompt

   =You are a helpful Meta ads assistant.
   
   Current date is: {{ $today.format('yyyy-MM-dd') }}

2. Register Meta Ads query tools

   Add three ToolWorkflow tool nodes to the AI Agent: list accounts, account details, and ad details. Each binds to the external workflow 'Meta Ads Agent' (workflowId: UXdblREvbkvy3WSs). The account details and ad details nodes use $fromAI so the Agent extracts parameters (account ID, date range) automatically. Also add a Calculator tool for numeric operations.

3. Configure sub-workflow inputs and routing

   The sub-workflow triggers via 'When Executed by Another Workflow', accepting command, id, since, and until. The 'since, until' node sets defaults: since defaults to start of current month, until to end of current month; override with provided values. The Switch node routes by command: list_accounts → list accounts; account → fetch campaigns/adsets/ads plus insights; list_ads → fetch ad-level insights. The 'fix missing act_' node prepends 'act_' to account IDs if missing.

4. Configure Facebook Graph API data queries

   Configure Facebook Graph API nodes per route, all using v23.0 with facebookGraphApi credential: - graph: adaccounts — node=me, edge=adaccounts, fields=name - account: campaigns — node={{ $json.id }}, edge=campaigns, fields=id,name,status, filter effective_status=["ACTIVE"] - account: adsets — same node, edge=adsets, fields=id,name,status,campaign - account: ads — same node, edge=ads, fields=id,name,status,campaign{id,name},adset{id,name} - account insights — edge=insights, includes spend/impressions/ctr/cpm/cpc/purchase_roas etc., time range from since/until - ads: insights — fetch ad-level insights data

5. Assemble and return results

   The Merge node combines campaigns, adsets, ads, and account_insights branches using combineByPosition. A Code node assembles the result into a JSON object returned to the main workflow. The ads: insights branch passes through Code1 to extract ads.data before returning.

You are setting up a conversational Meta Ads reporting assistant. First confirm you have: 1) n8n instance access; 2) OpenAI API key; 3) Facebook Graph API credentials with adaccounts permission. After importing the template, configure credentials, verify the AI Agent system message, and ensure the external sub-workflow 'Meta Ads Agent' is available. Test with a test account to verify the account list query returns data.