{ "edges": [ { "to": "f19469a6-7667-4595-93ae-750695e0b3d6", "from": "7ff62ebf-8eca-4085-ab2d-aa3ad1656f22", "toPort": "0", "fromPort": "0" }, { "to": "e7a6fbb9-ce74-49a1-a21b-e883e008d388", "from": "184c61e2-e4b9-496d-90fc-1d2fcfce9a15", "toPort": "0", "fromPort": "0" }, { "to": "8cb84507-2f08-45eb-940e-bee02b6f06a3", "from": "17afd3f1-7697-46bf-a303-7a91ba14b4c8", "toPort": "0", "fromPort": "ai_tool" }, { "to": "8cb84507-2f08-45eb-940e-bee02b6f06a3", "from": "e7a6fbb9-ce74-49a1-a21b-e883e008d388", "toPort": "0", "fromPort": "0" }, { "to": "7ff62ebf-8eca-4085-ab2d-aa3ad1656f22", "from": "e6de4884-bf7c-45f1-a5c8-3481b9ed5f5b", "toPort": "0", "fromPort": "0" }, { "to": "7843456d-dba3-484c-8982-f76640ac7ce6", "from": "3a24b5cd-a138-4b86-845a-055af932339a", "toPort": "0", "fromPort": "ai_tool" }, { "to": "8cb84507-2f08-45eb-940e-bee02b6f06a3", "from": "6644413f-f864-4e8a-b85f-de5c3794cfbf", "toPort": "0", "fromPort": "ai_languageModel" }, { "to": "7843456d-dba3-484c-8982-f76640ac7ce6", "from": "26492b87-50d3-4104-9b3b-9aa3a4ca10f3", "toPort": "0", "fromPort": "ai_languageModel" }, { "to": "e6de4884-bf7c-45f1-a5c8-3481b9ed5f5b", "from": "ec486f27-d925-477d-b2e3-bccbdcd08b4a", "toPort": "0", "fromPort": "ai_languageModel" }, { "to": "e6de4884-bf7c-45f1-a5c8-3481b9ed5f5b", "from": "7843456d-dba3-484c-8982-f76640ac7ce6", "toPort": "0", "fromPort": "0" }, { "to": "7843456d-dba3-484c-8982-f76640ac7ce6", "from": "8cb84507-2f08-45eb-940e-bee02b6f06a3", "toPort": "0", "fromPort": "0" } ], "nodes": [ { "id": "184c61e2-e4b9-496d-90fc-1d2fcfce9a15", "name": "Every Day 8 AM", "type": "trigger", "params": { "kind": "n8n-nodes-base.scheduleTrigger", "rule": { "interval": [ { "field": "cronExpression", "expression": "0 8 * * *" } ] } }, "position": [ 80, 704 ] }, { "id": "e7a6fbb9-ce74-49a1-a21b-e883e008d388", "name": "Get Current Date", "type": "passthrough", "params": { "mode": "dateTime", "options": {} }, "position": [ 272, 704 ] }, { "id": "8cb84507-2f08-45eb-940e-bee02b6f06a3", "name": "Website Analysis Agent", "type": "llm", "params": { "text": "=Today's Date: {{ $json.formattedDate }}\n\nFetch website performance data from Databox for the last 7 days compared to the previous 7 days. Calculate the exact date ranges based on today's date.\n\nFocus on: traffic volume, engagement quality (bounce rate, pages per session, session duration), and conversion metrics (goal completions, conversion rate).", "prompt": "=Today's Date: {{ $json.formattedDate }}\n\nFetch website performance data from Databox for the last 7 days compared to the previous 7 days. Calculate the exact date ranges based on today's date.\n\nFocus on: traffic volume, engagement quality (bounce rate, pages per session, session duration), and conversion metrics (goal completions, conversion rate).", "options": { "systemMessage": "You are a website analytics specialist. Fetch website performance data from Databox via MCP and produce a structured plain-text analysis.\n\nSTEP-BY-STEP:\n1. Call list_accounts to get the account ID.\n2. Call list_data_sources to find connected website analytics sources (Google Analytics, GA4, Adobe Analytics, etc.).\n3. For each website analytics source found, fetch these metrics for both the last 7 days and the previous 7 days:\n - Sessions or Users\n - Bounce Rate\n - Pages per Session\n - Average Session Duration\n - Goal Completions or Conversions\n - Conversion Rate\n4. Calculate week-over-week (WoW) percentage change for each metric.\n\nOUTPUT FORMAT:\nProduce a structured plain-text summary (NOT HTML) with:\n- Data source name\n- Sessions/Users: current value, WoW change %\n- Bounce Rate: current value, WoW change %\n- Pages per Session: current value, WoW change %\n- Avg Session Duration: current value, WoW change %\n- Goal Completions: current value, WoW change %\n- Conversion Rate: current value, WoW change %\n- 2-3 short observations about website behavior patterns\n\nIf no website analytics source is connected, output: \"No website analytics data available.\"\nDo NOT produce HTML - output plain text only." }, "provider": "openai", "promptType": "define", "approximated": true, "systemMessage": "You are a website analytics specialist. Fetch website performance data from Databox via MCP and produce a structured plain-text analysis.\n\nSTEP-BY-STEP:\n1. Call list_accounts to get the account ID.\n2. Call list_data_sources to find connected website analytics sources (Google Analytics, GA4, Adobe Analytics, etc.).\n3. For each website analytics source found, fetch these metrics for both the last 7 days and the previous 7 days:\n - Sessions or Users\n - Bounce Rate\n - Pages per Session\n - Average Session Duration\n - Goal Completions or Conversions\n - Conversion Rate\n4. Calculate week-over-week (WoW) percentage change for each metric.\n\nOUTPUT FORMAT:\nProduce a structured plain-text summary (NOT HTML) with:\n- Data source name\n- Sessions/Users: current value, WoW change %\n- Bounce Rate: current value, WoW change %\n- Pages per Session: current value, WoW change %\n- Avg Session Duration: current value, WoW change %\n- Goal Completions: current value, WoW change %\n- Conversion Rate: current value, WoW change %\n- 2-3 short observations about website behavior patterns\n\nIf no website analytics source is connected, output: \"No website analytics data available.\"\nDo NOT produce HTML - output plain text only." }, "position": [ 608, 704 ] }, { "id": "17afd3f1-7697-46bf-a303-7a91ba14b4c8", "name": "Databox MCP Tool", "type": "integration", "params": { "options": {}, "service": "mcp", "operation": "read", "endpointUrl": "https://mcp.databox.com/mcp", "authentication": "mcpOAuth2Api" }, "position": [ 752, 896 ] }, { "id": "7843456d-dba3-484c-8982-f76640ac7ce6", "name": "Paid Acquisition Agent", "type": "llm", "params": { "text": "=Today's Date: {{ $('Get Current Date').item.json.formattedDate }}\n\nWebsite analysis from Agent 1:\n{{ $json.output }}\n\nNow fetch paid acquisition performance data from Databox for the same period (last 7 days vs previous 7 days). Focus on all connected paid advertising platforms - fetch spend efficiency (CPC, CTR, ROAS), volume metrics (impressions, clicks), and conversion data per platform.", "prompt": "=Today's Date: {{ $('Get Current Date').item.json.formattedDate }}\n\nWebsite analysis from Agent 1:\n{{ $json.output }}\n\nNow fetch paid acquisition performance data from Databox for the same period (last 7 days vs previous 7 days). Focus on all connected paid advertising platforms - fetch spend efficiency (CPC, CTR, ROAS), volume metrics (impressions, clicks), and conversion data per platform.", "options": { "systemMessage": "You are a paid acquisition analyst. Fetch paid channel performance data from Databox via MCP and produce a structured plain-text analysis.\n\nSTEP-BY-STEP:\n1. Call list_accounts to get the account ID.\n2. Call list_data_sources to find all connected paid advertising platforms (Google Ads, Facebook Ads, LinkedIn Ads, TikTok Ads, Microsoft Ads, Reddit Ads, Snapchat Ads, Pinterest Ads, X/Twitter Ads, YouTube Ads).\n3. For EACH connected paid platform, fetch these metrics for both the last 7 days and the previous 7 days:\n - Spend or Cost\n - Clicks\n - CPC (Cost Per Click)\n - CTR (Click-Through Rate)\n - Impressions\n - ROAS or Conversions\n4. Calculate WoW percentage change for each metric on each platform.\n5. Calculate aggregated totals across all platforms: total spend, total clicks, total impressions, total conversions, average CPC, average CTR.\n\nOUTPUT FORMAT:\nProduce a structured plain-text summary (NOT HTML) with:\n- For EACH connected platform: name, spend, clicks, CPC, CTR, impressions, ROAS/conversions with WoW changes\n- Aggregated totals across all platforms with WoW changes\n- 2-3 short observations about paid performance patterns\n\nIf no paid ads platforms are connected, output: \"No paid advertising data available.\"\nDo NOT produce HTML - output plain text only." }, "provider": "openai", "promptType": "define", "approximated": true, "systemMessage": "You are a paid acquisition analyst. Fetch paid channel performance data from Databox via MCP and produce a structured plain-text analysis.\n\nSTEP-BY-STEP:\n1. Call list_accounts to get the account ID.\n2. Call list_data_sources to find all connected paid advertising platforms (Google Ads, Facebook Ads, LinkedIn Ads, TikTok Ads, Microsoft Ads, Reddit Ads, Snapchat Ads, Pinterest Ads, X/Twitter Ads, YouTube Ads).\n3. For EACH connected paid platform, fetch these metrics for both the last 7 days and the previous 7 days:\n - Spend or Cost\n - Clicks\n - CPC (Cost Per Click)\n - CTR (Click-Through Rate)\n - Impressions\n - ROAS or Conversions\n4. Calculate WoW percentage change for each metric on each platform.\n5. Calculate aggregated totals across all platforms: total spend, total clicks, total impressions, total conversions, average CPC, average CTR.\n\nOUTPUT FORMAT:\nProduce a structured plain-text summary (NOT HTML) with:\n- For EACH connected platform: name, spend, clicks, CPC, CTR, impressions, ROAS/conversions with WoW changes\n- Aggregated totals across all platforms with WoW changes\n- 2-3 short observations about paid performance patterns\n\nIf no paid ads platforms are connected, output: \"No paid advertising data available.\"\nDo NOT produce HTML - output plain text only." }, "position": [ 960, 704 ] }, { "id": "e6de4884-bf7c-45f1-a5c8-3481b9ed5f5b", "name": "Correlation Agent", "type": "llm", "params": { "text": "=Today's Date: {{ $('Get Current Date').item.json.formattedDate }}\n\nWEBSITE ANALYSIS (from Agent 1):\n{{ $('Website Analysis Agent').item.json.output }}\n\nPAID ACQUISITION ANALYSIS (from Agent 2):\n{{ $json.output }}\n\nUsing ONLY the data above (do not call any external tools), produce a complete daily intelligence report as an HTML email body. Structure it as follows:\n\n1. Executive Summary - 2-3 sentences on the single most important cross-channel finding\n2. Website Performance - table with key metrics and WoW changes\n3. Paid Acquisition - per-platform tables with spend, clicks, CPC, CTR, ROAS and WoW changes\n4. Correlation Analysis - the unique insight layer:\n - Does this week's paid spend correlate with website traffic volume?\n - Which paid channels drive the highest-quality visitors (lowest bounce rate, highest conversion rate)?\n - Estimated cost per website conversion by channel (total paid spend divided by goal completions)\n - Efficiency ranking of paid channels (best to worst)\n5. Recommendations - exactly 3 specific, actionable recommendations based on the data\n6. Footer\n\nOutput ONLY the HTML body content starting with a

tag. Do not include , , or tags.", "prompt": "=Today's Date: {{ $('Get Current Date').item.json.formattedDate }}\n\nWEBSITE ANALYSIS (from Agent 1):\n{{ $('Website Analysis Agent').item.json.output }}\n\nPAID ACQUISITION ANALYSIS (from Agent 2):\n{{ $json.output }}\n\nUsing ONLY the data above (do not call any external tools), produce a complete daily intelligence report as an HTML email body. Structure it as follows:\n\n1. Executive Summary - 2-3 sentences on the single most important cross-channel finding\n2. Website Performance - table with key metrics and WoW changes\n3. Paid Acquisition - per-platform tables with spend, clicks, CPC, CTR, ROAS and WoW changes\n4. Correlation Analysis - the unique insight layer:\n - Does this week's paid spend correlate with website traffic volume?\n - Which paid channels drive the highest-quality visitors (lowest bounce rate, highest conversion rate)?\n - Estimated cost per website conversion by channel (total paid spend divided by goal completions)\n - Efficiency ranking of paid channels (best to worst)\n5. Recommendations - exactly 3 specific, actionable recommendations based on the data\n6. Footer\n\nOutput ONLY the HTML body content starting with a

tag. Do not include , , or tags.", "options": { "systemMessage": "You are a marketing intelligence analyst specializing in cross-channel correlation analysis. You do NOT call any tools - you analyze only the data provided in the prompt.\n\nYour job is to synthesize website analytics and paid acquisition data into a single HTML report that surfaces insights neither dataset could reveal on its own.\n\nHTML FORMATTING RULES:\n- Use Databox brand color #3164FA for all headings and table headers\n- Color WoW changes: green (#059669) for improvements, red (#dc2626) for declines\n- Font: 'Helvetica Neue', Arial, sans-serif throughout\n- Table headers: background #3164FA, white text, padding 12px\n- Alternating table rows: #ffffff and #f9fafb\n- Section headings:

with color #3164FA, font-weight 600\n- Executive summary box: background #F7F9FC, left border 4px solid #3164FA, padding 15px, border-radius 4px\n- Correlation insights box: background #FFF8E7, left border 4px solid #F59E0B, padding 15px, border-radius 4px\n- Recommendations box: background #F0FDF4, left border 4px solid #059669, padding 15px, border-radius 4px\n- Footer: small gray text, border-top, font-size 11px\n- No em-dashes - use hyphens instead\n- Format numbers with thousand separators (commas)\n- Currency as $X,XXX.XX, percentages as X.XX%\n- If data for a section is missing, note it gracefully rather than showing an error" }, "provider": "openai", "promptType": "define", "approximated": true, "systemMessage": "You are a marketing intelligence analyst specializing in cross-channel correlation analysis. You do NOT call any tools - you analyze only the data provided in the prompt.\n\nYour job is to synthesize website analytics and paid acquisition data into a single HTML report that surfaces insights neither dataset could reveal on its own.\n\nHTML FORMATTING RULES:\n- Use Databox brand color #3164FA for all headings and table headers\n- Color WoW changes: green (#059669) for improvements, red (#dc2626) for declines\n- Font: 'Helvetica Neue', Arial, sans-serif throughout\n- Table headers: background #3164FA, white text, padding 12px\n- Alternating table rows: #ffffff and #f9fafb\n- Section headings:

with color #3164FA, font-weight 600\n- Executive summary box: background #F7F9FC, left border 4px solid #3164FA, padding 15px, border-radius 4px\n- Correlation insights box: background #FFF8E7, left border 4px solid #F59E0B, padding 15px, border-radius 4px\n- Recommendations box: background #F0FDF4, left border 4px solid #059669, padding 15px, border-radius 4px\n- Footer: small gray text, border-top, font-size 11px\n- No em-dashes - use hyphens instead\n- Format numbers with thousand separators (commas)\n- Currency as $X,XXX.XX, percentages as X.XX%\n- If data for a section is missing, note it gracefully rather than showing an error" }, "position": [ 1328, 704 ] }, { "id": "7ff62ebf-8eca-4085-ab2d-aa3ad1656f22", "name": "Prepare Email", "type": "code", "params": { "jsCode": "const agentOutput = $input.first().json.output || '';\n\nconst today = new Date();\nconst dateStr = today.toLocaleDateString('en-US', {\n month: 'long',\n day: 'numeric',\n year: 'numeric'\n});\n\nreturn [{\n json: {\n subject: `Daily Paid Acquisition Report - ${dateStr}`,\n htmlBody: agentOutput\n }\n}];" }, "position": [ 1808, 704 ] }, { "id": "f19469a6-7667-4595-93ae-750695e0b3d6", "name": "Send Email", "type": "integration", "params": { "sendTo": "user@example.com", "message": "={{ $json.htmlBody }}", "options": {}, "service": "gmail", "subject": "={{ $json.subject }}", "operation": "read" }, "position": [ 2000, 704 ] }, { "id": "3a24b5cd-a138-4b86-845a-055af932339a", "name": "Databox MCP Tool 2", "type": "integration", "params": { "options": {}, "service": "mcp", "operation": "read", "endpointUrl": "https://mcp.databox.com/mcp", "authentication": "mcpOAuth2Api" }, "position": [ 1136, 896 ] }, { "id": "ec486f27-d925-477d-b2e3-bccbdcd08b4a", "name": "OpenAI Chat Model 3", "type": "llm", "params": { "model": "gpt-4o", "options": { "maxTokens": 4096 }, "provider": "openai", "maxTokens": 4096, "builtInTools": {} }, "position": [ 1328, 896 ] }, { "id": "26492b87-50d3-4104-9b3b-9aa3a4ca10f3", "name": "OpenAI Chat Model 2", "type": "llm", "params": { "model": "gpt-4o", "options": { "maxTokens": 2048 }, "provider": "openai", "maxTokens": 2048, "builtInTools": {} }, "position": [ 928, 896 ] }, { "id": "6644413f-f864-4e8a-b85f-de5c3794cfbf", "name": "OpenAI Chat Model 1", "type": "llm", "params": { "model": "gpt-4o", "options": { "maxTokens": 2048 }, "provider": "openai", "maxTokens": 2048, "builtInTools": {} }, "position": [ 560, 896 ] } ] }

Send a daily paid acquisition and website intelligence report with Databox, GPT-4o and Gmail