January 22, 2026

Thordata + Gmail: executive review sentiment reports

Lisa Granqvist Partner Workflow Automation Expert

Get a free AI assessment → ⬇️ Use template

You finally sit down to understand what customers are saying, and it turns into five tabs, messy exports, duplicate reviews, and a “quick” analysis that somehow eats your afternoon. The worst part is the nagging doubt: did you miss the one complaint that explains your churn spike?

Product managers feel this when leadership wants an executive-ready update by tomorrow. A marketing lead trying to refine positioning feels it too. Same with customer success managers who need to spot risks before renewals. With review sentiment reports automation, you stop stitching together screenshots and start sending a clean summary that people actually read.

This n8n workflow pulls reviews from major platforms, runs a collective GPT-4.1 sentiment analysis, then emails a polished HTML report via Gmail. You’ll see what it does, what you need, and how it saves real hours every week.

How This Automation Works

The full n8n workflow, from trigger to final output:

n8n Workflow Template: Thordata + Gmail: executive review sentiment reports

Click to explore

flowchart LR

    subgraph sg0["Manual Flow"]
        direction LR
        n0["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Universal Safe Parser"]
        n1["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>URL Builder"]
        n2@{ icon: "mdi:play-circle", form: "rounded", label: "Manual Trigger", pos: "b", h: 48 }
        n3["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Prepare Review Sources"]
        n4@{ icon: "mdi:swap-vertical", form: "rounded", label: "Initialize Pagination", pos: "b", h: 48 }
        n5@{ icon: "mdi:swap-vertical", form: "rounded", label: "Loop Through Sources (Rate L..", pos: "b", h: 48 }
        n6["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/httprequest.dark.svg' width='40' height='40' /></div><br/>Scrape Page (Robust) via Tho.."]
        n7@{ icon: "mdi:swap-horizontal", form: "rounded", label: "Has More Pages?", pos: "b", h: 48 }
        n8@{ icon: "mdi:swap-vertical", form: "rounded", label: "Increment Page Counter", pos: "b", h: 48 }
        n9@{ icon: "mdi:cog", form: "rounded", label: "Rate Limit Delay (2s)", pos: "b", h: 48 }
        n10["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Deduplicate & Summarize All .."]
        n11["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Prepare Data for AI Analysis"]
        n12@{ icon: "mdi:robot", form: "rounded", label: "AI Review Sentiment Analysis", pos: "b", h: 48 }
        n13["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/webhook.dark.svg' width='40' height='40' /></div><br/>Return Final Results"]
        n14["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/webhook.dark.svg' width='40' height='40' /></div><br/>Webhook Trigger"]
        n15["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/form.svg' width='40' height='40' /></div><br/>Form Trigger – Submit Sources"]
        n16["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Render Elegant Email Template"]
        n17@{ icon: "mdi:message-outline", form: "rounded", label: "Send Executive Summary Email", pos: "b", h: 48 }
        n1 --> n6
        n2 --> n3
        n7 --> n8
        n7 --> n5
        n14 --> n3
        n4 --> n1
        n9 --> n1
        n0 --> n7
        n8 --> n9
        n3 --> n5
        n12 --> n16
        n11 --> n13
        n11 --> n12
        n16 --> n17
        n15 --> n3
        n10 --> n11
        n6 --> n0
        n5 --> n10
        n5 --> n4
    end

    %% Styling
    classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
    classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
    classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
    classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
    classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
    classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
    classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
    classDef disabled stroke-dasharray: 5 5,opacity: 0.5
    class n2,n15 trigger
    class n12 ai
    class n7 decision
    class n6,n13,n14 api
    class n0,n1,n3,n10,n11,n16 code
    classDef customIcon fill:none,stroke:none
    class n0,n1,n3,n6,n10,n11,n13,n14,n15,n16 customIcon

The Problem: Review data is scattered and easy to misread

Reviews are everywhere, but your time isn’t. Trustpilot says one thing, Capterra says another, and Product Hunt is a totally different vibe. When you try to piece it together manually, the process gets noisy fast: duplicates across platforms, inconsistent rating scales, and a bunch of “edge case” comments that are hard to categorize. You end up doing the same steps every month, then presenting a shaky summary because you couldn’t realistically read hundreds of reviews in one sitting.

It adds up fast. Here’s where it breaks down.

You spend about 10 minutes per platform just to find, filter, and export reviews, and that’s before any analysis starts.
Duplicates creep in from re-posted reviews, syndication, and scraping overlap, which quietly skews the themes you report.
Leadership wants “top risks and next steps,” but your notes look like a patchwork of quotes and hunches.
By the time you finish, the insights are already stale, so the team keeps shipping without closing the loop.

The Solution: One run pulls, deduplicates, analyzes, and emails the report

This workflow turns review chaos into a single executive summary you can forward as-is. It starts with a simple input: product URLs you submit via a form, a webhook, Telegram, or just the built-in defaults. From there, Thordata handles Cloudflare-safe scraping and pagination across five major review sources (Trustpilot, Capterra, Chrome Web Store, TrustRadius, and Product Hunt). Each page is parsed into a consistent format, duplicates are removed using deterministic IDs, and the cleaned dataset is bundled for analysis.

Then GPT-4.1 reads the reviews collectively, not as isolated one-offs, which means you get real patterns instead of a pile of mini-summaries. Finally, the workflow builds a responsive HTML email with sentiment badges, key stats, churn risk signals, critical issues, and actionable recommendations, and sends it straight through Gmail.

What You Get: Automation vs. Results

What This Workflow Automates

Results You’ll Get

Collects reviews from five major platforms in one run using Thordata.
Paginates safely, pauses for rate limiting, and keeps scraping stable.
Standardizes review formats with a universal parser, then removes duplicates.
Generates and sends a polished executive HTML report via Gmail automatically.

Get a “board-ready” summary in about 15 minutes instead of half a day of manual work.
Spot recurring praise and complaints across channels, not just the loudest platform.
Catch critical issues early (pricing confusion, onboarding friction, bugs) before they spread.
Turn review noise into clear recommendations your team can act on this week.
Share a consistent monthly or weekly report without rewriting the same narrative every time.

Example: What This Looks Like

Say you need a weekly review readout across five platforms. Manually, it’s easy to spend about 20 minutes per platform collecting and cleaning, plus another 2 hours skimming and summarizing, so roughly 4 hours total. With this workflow, you paste your product URLs once (maybe 5 minutes), let it run while it scrapes and paginates (often 10–20 minutes depending on volume), and the report lands in Gmail ready to forward. That’s about 3 hours back, every week, with fewer “did we miss something?” moments.

What You’ll Need

n8n instance (try n8n Cloud free)
Self-hosting option if you prefer (Hostinger works well)
Thordata for scraping reviews across platforms
Gmail to send the executive HTML report
OpenAI API key (get it from the OpenAI platform dashboard)

Skill level: Beginner. You’ll connect credentials, confirm the default sources, and run a test execution.

Don’t want to set this up yourself? Talk to an automation expert (free 15-minute consultation).

How It Works

A trigger kicks things off. You can run it manually, submit URLs through a webhook, or collect sources through a simple form intake. If you do nothing, it can even test using Thordata’s own review pages, which is handy for verifying your setup quickly.

Review sources are assembled and processed in batches. The workflow compiles platform URLs, iterates safely, and starts pagination. A built-in pause helps keep requests rate-limit safe, so your run doesn’t fall over halfway through.

Scraping and normalization happen automatically. Thordata pulls each page, then a universal parser converts every platform’s layout into one clean structure. After that, deduplication removes repeats so your analysis isn’t biased by copies of the same review.

AI produces an executive narrative, then email delivers it. GPT-4.1 analyzes the entire dataset together, and the workflow builds a responsive HTML report with themes, risks, and recommended next actions. Gmail sends it to your inbox (or a stakeholder list) so it becomes a repeatable reporting habit, not a one-off scramble.

You can easily modify the default review sources to cover competitors or a different product line based on your needs. See the full implementation guide below for customization options.

Step-by-Step Implementation Guide

Step 1: Configure the Trigger Type

This workflow can be triggered manually, via a webhook, or through a form submission. Configure each trigger based on how you want to launch the pipeline.

Open Manual Execution Start if you want to run tests directly in n8n with default values.
Configure Incoming Webhook Trigger with Path set to c35353c9-a0e5-4f0a-86a4-cc5466bd83fd and HTTP Method set to POST.
Set Response Mode to responseNode in Incoming Webhook Trigger so the response is handled by Return Webhook Output.
Open Form Intake Sources and confirm Path is sources, Form Title is Submit Product Review Sources, and the placeholders match your product inputs.

Tip: You can keep all three triggers enabled for flexibility—manual testing, webhook automation, and a shareable form.

Step 2: Connect the Review Source Assembly and Loop Controls

This stage assembles review source URLs, iterates through them safely, and initializes pagination for multi-page scraping.

Open Assemble Review Sources and confirm the default product values, especially name: "Thordata" and trustpilot_domain: "thordata.com".
Verify Assemble Review Sources returns one item per source in the sources array; this feeds Iterate Sources Safely.
In Iterate Sources Safely, keep the batching loop as-is; it controls pagination and source iteration.
In Set Pagination Start, ensure Include Other Fields is enabled and page_number is set to 1.

⚠️ Common Pitfall: If Set Pagination Start doesn’t include page_number, Dynamic URL Composer may throw a pagination error.

Step 3: Set Up the Scraping and Parsing Pipeline

These nodes build URLs, scrape HTML, parse reviews, and loop through pagination when needed.

In Dynamic URL Composer, keep the code that uses page_number and builds target_url from source_url.
Open Thordata Page Scraper and set URL to https://universalapi.thordata.com/request with Method set to POST.
Ensure the body parameters include url set to {{ $json.target_url }}, type set to html, js_render set to True, and clean_content set to css.
Credential Required: Connect your httpHeaderAuth credentials in Thordata Page Scraper.
Verify Universal Review Parser remains connected to parse HTML into standardized review objects.
In More Pages Check, keep the expression {{ ($json.reviews.length > 0 || $json.review_count > 0) && !['chromewebstore', 'producthunt'].includes($json.source_type) }} to trigger pagination.
In Increase Page Index, set page_number to {{ $json.page_number + 1 }} and keep Include Other Fields enabled.
In Rate Limit Pause, set Amount to 2 seconds to avoid throttling.

Tip: The pagination loop flows Universal Review Parser → More Pages Check → Increase Page Index → Rate Limit Pause → Dynamic URL Composer.

Step 4: Consolidate and Format the Review Data

These code nodes aggregate, deduplicate, and format reviews for AI analysis and responses.

Keep Deduplicate Review Summary as-is to read from global storage and remove duplicate reviews via unique_id.
In Deduplicate Review Summary, confirm summary.product is set to Thordata (update if your product name differs).
Open Format Data for AI and ensure it outputs product, total_reviews, avg_rating, all_reviews_text, and rating_distribution.

Tip: There are multiple code nodes in this workflow; only the critical path above (deduplication and formatting) needs frequent edits.

Step 5: Set Up the AI Analysis and Parallel Outputs

The formatted data is sent to the AI model and simultaneously returned to the webhook response.

In AI Sentiment Review, confirm the model is set to gpt-4.1-2025-04-14 and the prompt uses the provided JSON-only response template.
Credential Required: Connect your openAiApi credentials in AI Sentiment Review.
Format Data for AI outputs to both Return Webhook Output and AI Sentiment Review in parallel.
In Return Webhook Output, keep Respond With set to json and the response body mapping intact.

⚠️ Common Pitfall: If AI Sentiment Review returns non-JSON output, Build Email Report will fail JSON parsing.

Step 6: Configure the Email Report Output

The AI output is converted into an HTML email and sent through Gmail.

In Build Email Report, keep the HTML template generation logic intact to produce email_html.
Open Dispatch Summary Email and set Send To to your recipient, e.g., [YOUR_EMAIL].
Set Message to {{ $json.email_html }} and Subject to {{ $json.product_name }} Review Analysis – {{ new Date().toLocaleDateString() }}.
Credential Required: Connect your gmailOAuth2 credentials in Dispatch Summary Email.

Step 7: Test and Activate Your Workflow

Run a full test to verify pagination, parsing, AI output, and email delivery before activating the workflow.

Click Execute Workflow starting from Manual Execution Start to validate the default data path.
Optionally submit data through Form Intake Sources or send a POST request to Incoming Webhook Trigger for live inputs.
Confirm that Return Webhook Output returns a JSON response and Dispatch Summary Email sends the HTML report.
If everything looks correct, toggle the workflow to Active for production use.

🔒

Unlock Full Step-by-Step Guide

Get the complete implementation guide + downloadable template

Common Gotchas

Thordata credentials can expire or need specific permissions. If things break, check your n8n Credentials panel and confirm the HTTP Header Auth header is still present.
If you’re using Wait nodes or external rendering, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
Default prompts in AI nodes are generic. Add your brand voice early or you’ll be editing outputs forever.

Frequently Asked Questions

How long does it take to set up this review sentiment reports automation?

About 30 minutes if you already have your API keys.

Do I need coding skills to automate review sentiment reports?

No. You’ll mostly connect accounts and paste in the right URLs. The workflow logic is already built.

Is n8n free to use for this review sentiment reports workflow?

Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in OpenAI API usage (often a few cents per report) and any Thordata plan limits.

Where can I host n8n to run this automation?

Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.

Can I customize this review sentiment reports workflow for competitor tracking?

Yes, and it’s honestly straightforward. Update the default URLs in the “Assemble Review Sources” / “Prepare Review Sources” area so it points at competitor pages instead of your own product. If you want a different report style, tweak the prompt inside “AI Sentiment Review” and adjust the layout in “Build Email Report.” Common customizations include tagging reviews by platform, adding a “top quotes” section, and changing the final output from Gmail to Slack or Google Sheets.

Why is my Thordata connection failing in this workflow?

Most of the time it’s an expired or missing API key in your HTTP Request credentials. Double-check the header-based auth in n8n, then confirm your Thordata plan still has quota. If you’re scraping a lot in one run, you can also hit rate limits, so increasing the Wait duration usually stabilizes it.

How many reviews can this review sentiment reports automation handle?

Hundreds per run is typical, and you can push beyond that if your Thordata plan and OpenAI budget allow it.

Is this review sentiment reports automation better than using Zapier or Make?

For this use case, n8n is usually the better fit because the workflow needs pagination, deduplication, and conditional logic that gets awkward (and expensive) in simpler tools. You also have the option to self-host, which matters if you run big batches and don’t want per-task pricing surprises. Zapier or Make can still work if you’re only summarizing a small set of reviews from a single source. But once you care about “collective analysis” across platforms, n8n tends to be calmer to maintain. Talk to an automation expert if you want help choosing.

Set it up once, then let the workflow do the weekly grinding for you. You get a clean executive readout in your inbox, and you can spend your time actually acting on it.

Need Help Setting This Up?

Our automation experts can build and customize this workflow for your specific needs. Free 15-minute consultation—no commitment required.

Lisa Granqvist

Workflow Automation Expert

Expert in workflow automation and no-code tools.

{
"id": null,
"name": "Automated Review Intelligence Pipeline",
"versionId": null,
"meta": {
"instanceId": "0bf382315f62867076933ec547c149fb319cae48647d9828fc531d9d8c425604",
"templateId": null,
"templateCredsSetupCompleted": null
},
"nodes": [
{
"id": "flowpast-topbar-11076",
"name": "Flowpast Branding",
"type": "n8n-nodes-base.stickyNote",
"position": [
0,
20
],
"parameters": {
"color": 7,
"width": 1370,
"height": 80,
"content": "## Flowpast.com | Automation Workflow Library\n**\ud83d\udcd6 Full tutorial & setup guide:** flowpast.com"
},
"typeVersion": 1
},
{
"id": "6a3780d0-3d41-4572-8fbd-09aa88e15669",
"name": "Manual Execution Start",
"type": "n8n-nodes-base.manualTrigger",
"position": [
1115,
170
],
"parameters": [],
"typeVersion": 1
},
{
"id": "1bda3f48-85b5-4a18-b5e6-7bb4e1d866f0",
"name": "Incoming Webhook Trigger",
"type": "n8n-nodes-base.webhook",
"position": [
840,
185
],
"webhookId": "c35353c9-a0e5-4f0a-86a4-cc5466bd83fd",
"parameters": {
"path": "c35353c9-a0e5-4f0a-86a4-cc5466bd83fd",
"options": [],
"httpMethod": "POST",
"responseMode": "responseNode"
},
"typeVersion": 2.1
},
{
"id": "e0faf882-df2d-424f-956c-42bcf461b92c",
"name": "Form Intake Sources",
"type": "n8n-nodes-base.formTrigger",
"position": [
685,
130
],
"webhookId": "15f77aa6-9af1-40c5-8521-1d1ee74ac614",
"parameters": {
"path": "sources",
"options": [],
"formTitle": "Submit Product Review Sources",
"formFields": {
"values": [
{
"fieldLabel": "productName",
"placeholder": "Thordata",
"requiredField": true
},
{
"fieldLabel": "domain",
"placeholder": "thordata.com"
},
{
"fieldLabel": "capterraUrl",
"placeholder": "https://www.capterra.com/p/168264/ZoomInfo-sales/reviews"
},
{
"fieldLabel": "chromeStoreUrl",
"placeholder": "https://chromewebstore.google.com/detail/vpn-for-chrome-nordvpn-pr/[CONFIGURE_YOUR_API_KEY]"
},
{
"fieldLabel": "trustRadius",
"placeholder": "https://www.trustradius.com/products/leadgrabber-pro/reviews/all"
},
{
"fieldLabel": "producthunt",
"placeholder": "gemini-6"
}
]
},
"formDescription": "Provide review source URLs for a product. Placeholders shown for convenience, replace with your product's details as needed."
},
"typeVersion": 1
},
{
"id": "33b6d69b-4e91-4d18-af3f-71f1ddfaa4ae",
"name": "Assemble Review Sources",
"type": "n8n-nodes-base.code",
"position": [
390,
160
],
"parameters": {
"jsCode": "// === 1. Reset global storage for scraped reviews ===\nconst staticData = $getWorkflowStaticData('global');\nstaticData.allScrapedReviews = [];\n\n// === 2. Hard-coded defaults (your fallback) ===\nconst defaultProduct = {\n    name: \"Thordata\",\n    trustpilot_domain: \"thordata.com\",\n    capterra_url: \"\",\n    chrome_webstore_url: \"\",\n    trustradius_url: \"\",\n    producthunt_slug: \"\"\n};\n\n// === 3. Start with defaults ===\nlet product = { ...defaultProduct };\n\n// === 4. Override with Form input (highest priority) ===\nconst formData = $input.first()?.json;\n\nif (formData && (\n    formData.productName || \n    formData.domain || \n    formData.capterraUrl || \n    formData.chromeStoreUrl || \n    formData.trustRadius || \n    formData.producthunt\n)) {\n    product.name              = (formData.productName || defaultProduct.name)?.trim() || defaultProduct.name;\n    product.trustpilot_domain = (formData.domain || defaultProduct.trustpilot_domain)?.trim() || defaultProduct.trustpilot_domain;\n    product.capterra_url      = (formData.capterraUrl || defaultProduct.capterra_url)?.trim() || \"\";\n    product.chrome_webstore_url = (formData.chromeStoreUrl || defaultProduct.chrome_webstore_url)?.trim() || \"\";\n    product.trustradius_url   = (formData.trustRadius || defaultProduct.trustradius_url)?.trim() || \"\";\n    product.producthunt_slug  = (formData.producthunt || defaultProduct.producthunt_slug)?.trim() || \"\";\n}\n\n// === 5. Override with Webhook POST body (if present) ===\nelse if ($node[\"Incoming Webhook Trigger\"] && $input.first()?.json && Object.keys($input.first().json).length > 0) {\n    const body = $input.first().json;\n    \n    product.name              = (body.productName || defaultProduct.name)?.trim() || defaultProduct.name;\n    product.trustpilot_domain = (body.domain || defaultProduct.trustpilot_domain)?.trim() || defaultProduct.trustpilot_domain;\n    product.capterra_url      = (body.capterraUrl || \"\").toString().trim() || \"\";\n    product.chrome_webstore_url = (body.chromeStoreUrl || \"\").toString().trim() || \"\";\n    product.trustradius_url   = (body.trustRadius || \"\").toString().trim() || \"\";\n    product.producthunt_slug  = (body.producthunt || defaultProduct.producthunt_slug)?.trim() || \"\";\n}\n\n// (Manual \u201cExecute Workflow\u201d button \u2192 uses defaults only)\n\n// === 6. Build the standardized sources array ===\nconst sources = [];\n\nif (product.trustpilot_domain) {\n    sources.push({\n        product_name: product.name,\n        source_type: \"trustpilot\",\n        source_url: \"https://www.trustpilot.com/review/\" + product.trustpilot_domain\n    });\n}\n\nif (product.capterra_url) {\n    sources.push({\n        product_name: product.name,\n        source_type: \"capterra\",\n        source_url: product.capterra_url\n    });\n}\n\nif (product.chrome_webstore_url) {\n    sources.push({\n        product_name: product.name,\n        source_type: \"chromewebstore\",\n        source_url: product.chrome_webstore_url\n    });\n}\n\nif (product.trustradius_url) {\n    sources.push({\n        product_name: product.name,\n        source_type: \"trustradius\",\n        source_url: product.trustradius_url\n    });\n}\n\nif (product.producthunt_slug) {\n    sources.push({\n        product_name: product.name,\n        source_type: \"producthunt\",\n        source_url: \"https://www.producthunt.com/products/\" + product.producthunt_slug + \"/reviews?feed=single&filter=all&order=newest\"\n    });\n}\n\n// === 7. Return one item per source ===\nreturn sources.map(source => ({ json: source }));"
},
"typeVersion": 2
},
{
"id": "347afbbd-8f94-41aa-b0db-b159f70431a9",
"name": "Iterate Sources Safely",
"type": "n8n-nodes-base.splitInBatches",
"position": [
225,
120
],
"parameters": {
"options": []
},
"typeVersion": 3
},
{
"id": "202cbefc-0ac8-4f8f-b497-802f3598c0b1",
"name": "Set Pagination Start",
"type": "n8n-nodes-base.set",
"position": [
0,
395
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "d12b0a88-1a31-4cff-9abf-dc25101b6de4",
"name": "page_number",
"type": "number",
"value": 1
}
]
},
"includeOtherFields": true
},
"typeVersion": 3.4
},
{
"id": "a3ebf41d-973d-4426-800a-82454b59cd56",
"name": "Dynamic URL Composer",
"type": "n8n-nodes-base.code",
"position": [
180,
355
],
"parameters": {
"jsCode": "// === DYNAMIC URL BUILDER ===\nconst item = $input.first().json;\n\n// Get the current page number from the loop context, default to 1\nconst page = item.page_number || 1;\nlet url = item.source_url;\nconst type = item.source_type;\n\nif (!url) {\n  throw new Error(\"Pagination Error: 'source_url' was lost in the loop.\");\n}\n\n// 1. Chrome Web Store: Skip pagination (Infinite Scroll logic is too complex for this method)\nif (type === 'chromewebstore') {\n  return { ...item, target_url: url };\n}\n\n// 2. Pagination Logic\n// If we are on Page 1, we usually don't need the parameter (safer to avoid redirects)\n\nif (page > 1) {\n  // If ProductHunt already has params (like ?feed=single), this correctly uses '&'\n  const separator = url.includes('?') ? '&' : '?';\n  url = `${url}${separator}page=${page}`;\n}\n\nreturn { \n  ...item, \n  target_url: url \n};"
},
"typeVersion": 2
},
{
"id": "5708ec7e-2a50-49b1-bbd5-dc2b514e29ae",
"name": "Thordata Page Scraper",
"type": "n8n-nodes-base.httpRequest",
"onError": "continueRegularOutput",
"position": [
480,
380
],
"parameters": {
"url": "https://universalapi.thordata.com/request",
"method": "POST",
"options": {
"proxy": "http://td-customer-username:[YOUR_EMAIL]:9999",
"timeout": 120000
},
"sendBody": true,
"contentType": "form-urlencoded",
"authentication": "genericCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "url",
"value": "={{ $json.target_url }}"
},
{
"name": "type",
"value": "html"
},
{
"name": "js_render",
"value": "True"
},
{
"name": "clean_content",
"value": "css"
}
]
},
"genericAuthType": "httpHeaderAuth"
},
"credentials": {
"httpHeaderAuth": {
"id": "credential-id",
"name": ""
}
},
"retryOnFail": true,
"typeVersion": 4.3,
"alwaysOutputData": true
},
{
"id": "7a22b325-5cc4-4410-a3e0-4b6157bf2fd1",
"name": "Universal Review Parser",
"type": "n8n-nodes-base.code",
"position": [
655,
400
],
"parameters": {
"jsCode": "// === UNIVERSAL SAFE PARSER (STANDARDIZATION + UNIQUE ID) ===\nconst results = [];\n\nfor (const item of $input.all()) {\n  const json = item.json || {};\n  const html = json.html || json.data || json.content || \"\";\n  const sourceUrl =\n    $(\"Iterate Sources Safely\").first().json.source_url || \"\";\n  let sourceType = $(\"Iterate Sources Safely\").first().json\n    .source_type;\n  let productName = $(\"Iterate Sources Safely\").first().json\n    .product_name;\n\n  if (!sourceType || sourceType === \"unknown\") {\n    sourceType = detectSource(sourceUrl, html);\n  }\n\n  /** @type {any[]} */\n  let reviews = [];\n  let error = null;\n\n  if (!html || html.length < 200) {\n    error = \"No HTML content\";\n  } else {\n    try {\n      if (sourceType === \"trustpilot\") reviews = parseTrustpilot(html);\n      else if (sourceType === \"capterra\") reviews = parseCapterra(html);\n      else if (sourceType === \"chromewebstore\")\n        reviews = parseChromeWebStore(html);\n      else if (sourceType === \"producthunt\") reviews = parseProductHunt(html);\n      else if (sourceType === \"trustradius\") reviews = parseTrustRadius(html);\n      else error = `Unknown source: ${sourceType}`;\n    } catch (e) {\n      error = e instanceof Error ? e.message : String(e);\n      //error = e.message;\n    }\n  }\n\n  // === STANDARDIZE EVERY REVIEW + ADD UNIQUE ID ===\n  const standardizedReviews = (reviews || []).map((review) => {\n    const cleanText = (review.text || \"\").trim();\n    const cleanTitle = (review.title || \"\").trim();\n    const author = (review.author || \"Anonymous\").trim();\n    let date = null;\n    if (review.date) {\n      const d = new Date(review.date);\n      // Check if 'd' is a valid date object\n      if (!isNaN(d.getTime())) {\n        try {\n          date = d.toISOString().split(\"T\")[0];\n        } catch (e) {\n          date = null; // Fallback if ISO conversion fails\n        }\n      } else {\n        // Optional: If date is \"2 days ago\", it ends up here.\n        // You can leave it null or map it to today's date if strictly necessary.\n        date = null;\n      }\n    }\n\n    // === UNIQUE ID: deterministic hash based on source + author + date + first 100 chars of text ===\n    // This survives reformatting, extra spaces, etc. \u2014 99.99% duplicate-proof\n    const hashString = `${sourceType}|${author}|${date}|${cleanText.substring(0, 150)}`;\n    function simpleHash(str) {\n      let hash = 0;\n      for (let i = 0; i < str.length; i++) {\n        const char = str.charCodeAt(i);\n        hash = (hash << 5) - hash + char;\n        hash = hash & hash; // Convert to 32-bit integer\n      }\n      return Math.abs(hash).toString(36); // short, URL-safe string\n    }\n    const unique_id = `${sourceType}_${simpleHash(hashString)}`;\n\n    return {\n      unique_id, // \u2190 Your primary key / dedupe field\n      author: author,\n      rating: review.rating ? Number(review.rating) : null,\n      title: cleanTitle || \"[No title provided]\",\n      text: cleanText,\n      pros: review.pros || \"[Not specified]\",\n      cons: review.cons || \"[Not specified]\",\n      date: date,\n      job_title: review.job_title || \"[Not provided]\",\n      company_size: review.company_size || \"[Unknown]\",\n      industry: review.industry || \"[Unknown]\",\n      location: review.location || \"[Not specified]\",\n      verified: review.verified || false,\n\n      scraped_at: new Date().toISOString(),\n    };\n  });\n\n  const previousPage = $(\"Dynamic URL Composer\").first().json.page_number || 1;\n  results.push({\n    json: {\n      //...json,\n      // === CRITICAL: PASS METADATA FOR PAGINATION LOOP ===\n      product_name: productName,\n      page_number: previousPage,\n      review_count: standardizedReviews.length,\n      reviews: standardizedReviews,\n      source_url: sourceUrl,\n      source_type: sourceType,      \n      success: !error,\n      error_message: error,\n    },\n  });\n}\n\n  // === STEP 2: SAVE TO GLOBAL STORAGE ===\n  const staticData = $getWorkflowStaticData('global');\n  // Initialize if it doesn't exist (safety check)\n  if (!staticData.allScrapedReviews) staticData.allScrapedReviews = [];\n  \n  // Append the results from this specific loop iteration to the global list\n  // We map 'results' because it contains { json: ... } structure\n  const cleanItems = results.map(r => r.json);\n  staticData.allScrapedReviews = staticData.allScrapedReviews.concat(cleanItems);\n\nreturn results;\n\n// ==================== HELPER FUNCTIONS ====================\n\nfunction detectSource(url, html) {\n  const u = (url || \"\").toLowerCase();\n  const h = html || \"\";\n\n  // A. Check URL first\n  if (u.includes(\"trustpilot\")) return \"trustpilot\";\n  if (u.includes(\"capterra\")) return \"capterra\";\n  if (u.includes(\"chrome.google.com\") || u.includes(\"chromewebstore\"))\n    return \"chromewebstore\";\n  if (u.includes(\"producthunt\")) return \"producthunt\";\n  if (u.includes(\"trustradius\")) return \"trustradius\";\n\n  // B. Check HTML Signatures (Fallback)\n  if (\n    h.includes(\"data-service-review-card-paper\") ||\n    h.includes(\"trustpilot-widget\")\n  )\n    return \"trustpilot\";\n  if (h.includes(\"__next_f\") && h.includes(\"CapterraProduct\"))\n    return \"capterra\";\n  if (h.includes(\"T7rvce\") && h.includes(\"LfYwpe\")) return \"chromewebstore\";\n  if (\n    h.includes(\"DetailedReviewConnection\") ||\n    h.includes(\"DetailedReviewEdge\")\n  )\n    return \"producthunt\";\n  if (h.includes(\"Review_review__\") || h.includes(\"trScore\"))\n    return \"trustradius\";\n\n  return \"unknown\";\n}\n\nfunction decodeHtml(html) {\n  if (!html) return \"\";\n  return html\n    .replace(/&/g, \"&\")\n    .replace(/</g, \"<\")\n    .replace(/>/g, \">\")\n    .replace(/"/g, '\"')\n    .replace(/'/g, \"'\")\n    .replace(/<br\\s*\\/?>/gi, \"\\n\");\n}\n\n// --- 1. Trustpilot Parser ---\nfunction parseTrustpilot(html) {\n  if (\n    html.includes(\"cf-browser-verification\") ||\n    html.includes(\"Just a moment\")\n  ) {\n    throw new Error(\"Blocked by Cloudflare.\");\n  }\n  const cards = html.split('data-service-review-card-paper=\"true\"').slice(1);\n  const extracted = [];\n\n  for (const card of cards) {\n    const authorMatch = card.match(\n      /data-consumer-name-typography=\"true\">([^<]+)</,\n    );\n    const ratingMatch = card.match(/Rated (\\d) out of 5/);\n    const titleMatch = card.match(\n      /data-service-review-title-typography=\"true\">([^<]+)</,\n    );\n    const textMatch = card.match(\n      /data-service-review-text-typography=\"true\">([\\s\\S]*?)(?=<\\/p>|<div|$)/,\n    );\n    const dateMatch = card.match(/datetime=\"([^\"]+)\"/);\n    const locationMatch = card.match(\n      /data-consumer-country-typography=\"true\">([^<]+)</,\n    );\n\n    if (authorMatch && textMatch) {\n      extracted.push({\n        author: authorMatch[1].trim(),\n        rating: ratingMatch ? parseInt(ratingMatch[1]) : null,\n        title: titleMatch ? titleMatch[1].trim() : null,\n        text: decodeHtml(textMatch[1]).trim(),\n        date: dateMatch ? dateMatch[1] : null,\n        location: locationMatch ? locationMatch[1].trim() : null,\n        verified: card.includes(\"Verified\"),\n        source: \"trustpilot\",\n      });\n    }\n  }\n  return extracted;\n}\n\n// --- 2. Capterra Parser ---\nfunction parseCapterra(html) {\n  let fullDataString = \"\";\n  const regex = /<script>self\\.__next_f\\.push\\(\\[1,\"([\\s\\S]*?)\"\\]\\)<\\/script>/g;\n  let match;\n  while ((match = regex.exec(html)) !== null) {\n    fullDataString += match[1];\n  }\n\n  if (!fullDataString) {\n    throw new Error(\"No Next.js data script tags found.\");\n  }\n\n  // C. Build Data Map (Your Split Logic)\n  const dataMap = new Map();\n  // In the regex capture, newlines are stored as the literal string \"\\\\n\"\n  const chunks = fullDataString.split(\"\\\\n\");\n\n  chunks.forEach((chunk) => {\n    if (!chunk) return;\n    const firstColonIndex = chunk.indexOf(\":\");\n    if (firstColonIndex === -1) return;\n\n    const key = chunk.substring(0, firstColonIndex);\n    let value = chunk.substring(firstColonIndex + 1);\n    // Unescape quotes *after* splitting\n    value = value.replace(/\\\\\"/g, '\"');\n    dataMap.set(key, value);\n  });\n\n  if (dataMap.size === 0)\n    throw new Error(\"Data map empty. Regex succeeded but split failed.\");\n\n  // D. Recursive Resolver (Your Cache Logic)\n  const resolvedCache = new Map();\n  function resolve(key) {\n    if (resolvedCache.has(key)) return resolvedCache.get(key);\n\n    let valueStr = dataMap.get(key);\n    if (!valueStr) return null;\n\n    let result;\n    if (valueStr.startsWith(\"{\") || valueStr.startsWith(\"[\")) {\n      try {\n        result = JSON.parse(valueStr, (reviverKey, reviverValue) => {\n          if (\n            typeof reviverValue === \"string\" &&\n            reviverValue.startsWith(\"$\")\n          ) {\n            const refKey = reviverValue.substring(1);\n            return resolve(refKey);\n          }\n          return reviverValue;\n        });\n      } catch (e) {\n        result = null;\n      }\n    } else if (valueStr.startsWith(\"$\")) {\n      result = resolve(valueStr.substring(1));\n    } else if (valueStr.startsWith('\"') && valueStr.endsWith('\"')) {\n      result = valueStr.substring(1, valueStr.length - 1);\n    } else {\n      result = valueStr;\n    }\n    resolvedCache.set(key, result);\n    return result;\n  }\n\n  // E. Find Main Reviews Key (Your Key Finder)\n  let mainReviewsKey = null;\n  for (const [key, value] of dataMap.entries()) {\n    if (\n      value.includes('\"__typename\":\"CapterraProduct\"') &&\n      value.includes('\"textReviews\":')\n    ) {\n      const reviewKeyMatch = value.match(\n        /\"textReviews\":\"\\$([a-zA-Z0-9]{1,5})\"/,\n      );\n      if (reviewKeyMatch && reviewKeyMatch[1]) {\n        mainReviewsKey = reviewKeyMatch[1];\n        break;\n      }\n    }\n  }\n\n  // Fallback strategy if Product Object search fails\n  if (!mainReviewsKey) {\n    for (const [key, value] of dataMap.entries()) {\n      if (value.startsWith('[\"$') && value.endsWith('\"]')) {\n        // Basic heuristic: assume the first array of references might be it\n        mainReviewsKey = key;\n        break;\n      }\n    }\n  }\n\n  if (!mainReviewsKey) throw new Error(\"Could not find main review list key.\");\n\n  // F. Resolve List\n  const mainReviewsList = resolve(mainReviewsKey);\n  if (!Array.isArray(mainReviewsList))\n    throw new Error(\"Resolved review list is not an array.\");\n\n  // G. Map Data (Your Mapper)\n  const getRating = (rating) =>\n    rating !== null && rating !== undefined\n      ? parseFloat(rating).toString()\n      : null;\n\n  return mainReviewsList\n    .map((review) => {\n      if (!review || !review.reviewer) return null;\n\n      let formattedDate = review.writtenOn; // Keep original if parsing fails\n      try {\n        // Simple date formatting\n        formattedDate = new Date(review.writtenOn).toISOString().split(\"T\")[0];\n      } catch (e) {}\n\n      return {\n        author: review.reviewer.anonymityOn\n          ? \"Verified Reviewer\"\n          : review.reviewer.fullName,\n        job_title: review.reviewer.jobTitle,\n        company_size: review.reviewer.companySize,\n        industry: review.reviewer.industry,\n        rating: getRating(review.overallRating),\n        title: review.title,\n        text: review.generalComments,\n        pros: review.prosText,\n        cons: review.consText,\n        date: formattedDate,\n        verified: review.reviewer.isValidated ? \"Yes\" : \"No\",\n        source: \"capterra\",\n      };\n    })\n    .filter((r) => r !== null);\n}\n\n// --- 3. Chrome Web Store Parser ---\nfunction parseChromeWebStore(html) {\n  const chunks = html.split('class=\"T7rvce\"').slice(1);\n  const extracted = [];\n  for (const chunk of chunks) {\n    const authorMatch = chunk.match(/class=\"LfYwpe\">([^<]+)</);\n    const ratingMatch = chunk.match(/aria-label=\"(\\d) out of 5 stars\"/);\n    const dateMatch = chunk.match(/class=\"ydlbEf\">([^<]+)</);\n    const textMatch = chunk.match(/class=\"fzDEpf\"[^>]*>([\\s\\S]*?)<\\/p>/);\n\n    if (authorMatch) {\n      extracted.push({\n        author: authorMatch[1].trim(),\n        rating: ratingMatch ? parseInt(ratingMatch[1]) : null,\n        date: dateMatch ? dateMatch[1].trim() : null,\n        text: textMatch ? decodeHtml(textMatch[1]).trim() : null,\n        source: \"chromewebstore\",\n      });\n    }\n  }\n  return extracted;\n}\n\n// --- 4. Product Hunt Parser ---\nfunction parseProductHunt(html) {\n  const extracted = [];\n  const seenIds = new Set();\n  \n  const startMatch = html.match(/DetailedReviewConnection.*?edges/);\n  if (!startMatch) return []; // Return empty if not found\n  \n  const section = html.substring(startMatch.index);\n  const blocks = section.split(/DetailedReviewEdge.*?node.*?DetailedReview/);\n  blocks.shift();\n\n  const getVal = (block, key) => {\n    const r = new RegExp(`\"${key}\":\"([^\"]+)\"`);\n    const m = block.match(r);\n    return m ? m[1] : null;\n  };\n\n  for (const block of blocks) {\n    const reviewId = getVal(block, \"id\");\n    if (reviewId && seenIds.has(reviewId)) {\n      continue;\n    }\n    if (reviewId) seenIds.add(reviewId);\n    \n    const name = getVal(block, \"name\") || \"Anonymous\";\n    const body = getVal(block, \"body\") || getVal(block, \"overallExperience\");\n    const date = getVal(block, \"createdAt\");\n\n    if (body) {\n      extracted.push({\n        unique_id: `producthunt_${reviewId}`,\n        author: name,\n        text: decodeHtml(body.replace(/\\\\n/g, \"\\n\")),\n        date: date,\n        source: \"producthunt\",\n      });\n    }\n  }\n  return extracted;\n}\n\n// --- 5. TrustRadius Parser ---\nfunction parseTrustRadius(html) {\n  const blocks = html.split('<article class=\"Review_review');\n  blocks.shift();\n  const extracted = [];\n  for (const block of blocks) {\n    const titleMatch = block.match(\n      /Header_heading__[^>]+><a[^>]+>([^<]+)<\\/a>/,\n    );\n    const ratingMatch = block.match(/data-rating=\"([\\d\\.]+)\"/);\n    const nameMatch = block.match(/Byline_name__[^>]+>([^<]+)<\\/div>/);\n    const dateMatch = block.match(/datetime=\"([^\"]+)\"/);\n    const summaryMatch = block.match(\n      /Use Cases and Deployment Scope<\\/h3><p[^>]*>(.*?)<\\/p>/,\n    );\n\n    if (ratingMatch) {\n      extracted.push({\n        author: nameMatch ? decodeHtml(nameMatch[1]) : \"Anonymous\",\n        rating: parseFloat(ratingMatch[1]),\n        title: titleMatch ? decodeHtml(titleMatch[1]) : null,\n        text: summaryMatch ? decodeHtml(summaryMatch[1]) : null,\n        date: dateMatch ? dateMatch[1] : null,\n        source: \"trustradius\",\n      });\n    }\n  }\n  return extracted;\n}\n"
},
"typeVersion": 2,
"alwaysOutputData": true
},
{
"id": "fb1e5835-d23c-47fd-a4cc-54648f5b4817",
"name": "More Pages Check",
"type": "n8n-nodes-base.if",
"position": [
930,
350
],
"parameters": {
"options": [],
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "loose"
},
"combinator": "or",
"conditions": [
{
"id": "d0cee53f-bdb6-4e2a-a9ac-e18950366991",
"operator": {
"type": "boolean",
"operation": "equals"
},
"leftValue": "={{ ($json.reviews.length > 0 || $json.review_count > 0) && !['chromewebstore', 'producthunt'].includes($json.source_type) }}",
"rightValue": true
}
]
},
"looseTypeValidation": true
},
"typeVersion": 2.2
},
{
"id": "e2131ae6-aff2-47c8-8fbf-768f38696bc2",
"name": "Increase Page Index",
"type": "n8n-nodes-base.set",
"position": [
1085,
385
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "ac85baa8-149f-4c43-bb8e-58b44b39ecc0",
"name": "page_number",
"type": "number",
"value": "={{ $json.page_number + 1 }}"
}
]
},
"includeOtherFields": true
},
"typeVersion": 3.4
},
{
"id": "e457c5c7-51df-491b-addb-03e718d3ddfb",
"name": "Rate Limit Pause",
"type": "n8n-nodes-base.wait",
"position": [
1270,
345
],
"webhookId": "52fd859c-28b3-4995-83b3-ccf39dc90856",
"parameters": {
"amount": 2
},
"typeVersion": 1.1
},
{
"id": "c7af6943-9a5c-4d47-a1c6-4a18f76d8d40",
"name": "Deduplicate Review Summary",
"type": "n8n-nodes-base.code",
"position": [
1055,
630
],
"parameters": {
"jsCode": "// === STEP 3: READ GLOBAL STORAGE ===\nconst staticData = $getWorkflowStaticData('global');\nconst allItems = staticData.allScrapedReviews || [];\n\nconst allReviews = [];\nconst errors = [];\nconst seenIds = new Set();       // \u2190 Extra safety net\nconst dedupedReviews = [];\n\nallItems.forEach(item => {\n  if (!item) return [];\n  const json = item.json || item;\n  if (!json) return [];\n  \n  const { reviews = [], success, error_message, source_detected, source_url } = json;\n\n  if (!success) {\n    errors.push({ source: source_detected || source_url, error: error_message });\n  } else {\n  // Ensure reviews is actually an array before looping\n    if (Array.isArray(reviews)) {  \n    reviews.forEach(r => {\n      if (!seenIds.has(r.unique_id)) {\n        seenIds.add(r.unique_id);\n        dedupedReviews.push(r);\n      }\n    });\n  }\n  }\n});\n\nreturn [{\n  json: {\n    summary: {\n      product: \"Thordata\",\n      total_reviews_scraped: dedupedReviews.length,\n      unique_reviews: dedupedReviews.length,\n      sources_attempted: $input.all().length,\n      sources_failed: errors.length,\n      scraped_at: new Date().toISOString()\n    },\n    reviews: dedupedReviews,\n    errors: errors.length > 0 ? errors : null\n  }\n}];"
},
"typeVersion": 2
},
{
"id": "021e2bfb-a706-4932-888a-c54f43db63a9",
"name": "Format Data for AI",
"type": "n8n-nodes-base.code",
"position": [
890,
600
],
"parameters": {
"jsCode": "const { summary, reviews } = $input.first().json;\n\n// Format reviews into readable text\nconst reviewText = reviews.map((r, idx) => {\n  return `Review #${idx + 1} (${r.source} - Rating: ${r.rating}/5):\nAuthor: ${r.author || 'Anonymous'}\nDate: ${r.date}\n${r.title ? 'Title: ' + r.title : ''}\nText: ${r.text || '[No text provided]'}`;\n}).join('\\n\\n---\\n\\n');\n\n// Calculate rating distribution\nconst ratings = reviews.map(r => r.rating).filter(r => r);\nconst avgRating = ratings.length > 0 \n  ? (ratings.reduce((a, b) => a + b, 0) / ratings.length).toFixed(2) \n  : 0;\n\nconst ratingDist = {\n  '5': ratings.filter(r => r === 5).length,\n  '4': ratings.filter(r => r === 4).length,\n  '3': ratings.filter(r => r === 3).length,\n  '2': ratings.filter(r => r === 2).length,\n  '1': ratings.filter(r => r === 1).length\n};\n\n// Sources breakdown\nconst sources = [...new Set(reviews.map(r => r.source))];\n\nreturn [{\n  json: {\n    // Flattened for template\n    product: summary.product,\n    total_reviews: summary.unique_reviews,\n    avg_rating: avgRating,\n    \n    // For the prompt\n    all_reviews_text: reviewText,\n    \n    // Additional context\n    rating_distribution: ratingDist,\n    sources: sources,\n    date_range: {\n      earliest: reviews.map(r => r.date).sort()[0],\n      latest: reviews.map(r => r.date).sort().reverse()[0]\n    },\n    \n    // Keep original for reference\n    original_reviews: reviews,\n    original_summary: summary\n  }\n}];"
},
"typeVersion": 2
},
{
"id": "e0b090cb-ef8b-4b07-aa89-46df26731575",
"name": "Return Webhook Output",
"type": "n8n-nodes-base.respondToWebhook",
"position": [
625,
645
],
"parameters": {
"options": [],
"respondWith": "json",
"responseBody": "={{\n  {\n    \"product_name\": $json.product,\n    \"stats\": $json.original_summary,\n    \"data\": $json.original_reviews\n  }\n}}"
},
"typeVersion": 1.4
},
{
"id": "861af668-6cae-4e1e-9c72-1c29b959506e",
"name": "AI Sentiment Review",
"type": "@n8n/n8n-nodes-langchain.openAi",
"position": [
455,
595
],
"parameters": {
"modelId": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1-2025-04-14",
"cachedResultName": "GPT-4.1-2025-04-14"
},
"options": [],
"responses": {
"values": [
{
"content": "=You are a B2B product analyst. Analyze these {{$json.total_reviews}} product reviews collectively.\n\nPRODUCT: {{$json.product}}\nAVERAGE RATING: {{$json.avg_rating}}/5\nDATE RANGE: {{$json.date_range.earliest}} to {{$json.date_range.latest}}\nSOURCES: {{$json.sources}}\n\nRating Distribution:\n- 5 stars: {{$json.rating_distribution[\"5\"]}} reviews\n- 1 star: {{$json.rating_distribution[\"1\"]}} reviews\n\nREVIEWS:\n{{$json.all_reviews_text}}\n\nAnalyze all reviews collectively. Provide ONLY JSON, NO markdown:\n\n{\n  \"collective_sentiment_score\": <-1 to 1>,\n  \"collective_sentiment_label\": \"<positive or neutral or negative>\",\n  \"confidence\": <0-1>,\n  \"overall_emotion\": \"<satisfied or frustrated or delighted or disappointed>\",\n  \"urgency\": \"<critical or high or medium or low>\",\n  \n  \"batch_analysis\": {\n    \"common_praise\": [\"theme 1\", \"theme 2\", \"theme 3\"],\n    \"common_complaints\": [\"issue 1\", \"issue 2\", \"issue 3\"],\n    \"critical_issues\": [{\"issue\": \"description\", \"mentions\": 3, \"severity\": \"high\"}],\n    \"key_features_mentioned\": [\"feature 1\", \"feature 2\"],\n    \"pricing_sentiment\": \"<fair or overpriced>\",\n    \"support_quality\": \"<excellent or good or poor>\"\n  },\n  \n  \"trend\": {\n    \"direction\": \"<improving or declining or stable>\",\n    \"key_finding\": \"main insight\"\n  },\n  \n  \"recommended_actions\": {\n    \"product_team\": \"action\",\n    \"sales_team\": \"action\",\n    \"support_team\": \"action\"\n  },\n  \n  \"overall_health\": \"<healthy or at-risk>\",\n  \"churn_risk\": \"<low or medium or high>\"\n}"
}
]
},
"builtInTools": []
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 2
},
{
"id": "1d1fd03e-0fdf-47a1-8773-b542a7ed192a",
"name": "Build Email Report",
"type": "n8n-nodes-base.code",
"position": [
170,
620
],
"parameters": {
"jsCode": "// === 1. Get the AI raw JSON string ===\nconst openAiOutput = $input.first().json.output?.[0]?.content?.[0]?.text;\n\nif (!openAiOutput) {\n  throw new Error(\"OpenAI response missing\");\n}\n\nlet aiAnalysis;\ntry {\n  aiAnalysis = JSON.parse(openAiOutput);\n} catch (e) {\n  throw new Error(`Failed to parse AI response as JSON:\\n${openAiOutput}`);\n}\n\n// Now safe to use\nconst product = $('Format Data for AI').first().json.product || \"Unknown Product\";\nconst totalReviews = $('Format Data for AI').first().json.total_reviews || 0;\nconst avgRating = $('Format Data for AI').first().json.avg_rating ? Number($('Format Data for AI').first().json.avg_rating).toFixed(2) : \"N/A\";\n\n// === 3. Colors ===\nconst sentimentColor = {\n  positive: \"#10b981\",\n  neutral: \"#f59e0b\",\n  negative: \"#ef4444\"\n}[aiAnalysis.collective_sentiment_label] || \"#6b7280\";\n\nconst healthColor = aiAnalysis.overall_health === \"healthy\" ? \"#10b981\" : \"#ef4444\";\nconst churnColor = aiAnalysis.churn_risk === \"low\" ? \"#10b981\" : \n                   aiAnalysis.churn_risk === \"medium\" ? \"#f59e0b\" : \"#ef4444\";\n\n// === 4. Generate stunning responsive HTML email ===\nconst html = `<!DOCTYPE html>\n<html>\n<head>\n  <meta charset=\"utf-8\">\n  <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\">\n  <title>${product} Review Intelligence Report</title>\n  <link href=\"https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&display=swap\" rel=\"stylesheet\">\n  <style>\n    body { font-family: 'Inter', system-ui, sans-serif; background: #f9fafb; margin: 0; padding: 0; }\n    .container { max-width: 680px; margin: 40px auto; background: white; border-radius: 16px; overflow: hidden; box-shadow: 0 20px 40px rgba(0,0,0,0.08); }\n    .header { background: linear-gradient(135deg, #6366f1, #8b5cf6); padding: 48px 32px; text-align: center; color: white; }\n    .header h1 { margin: 0; font-size: 32px; font-weight: 700; }\n    .header p { margin: 12px 0 0; font-size: 17px; opacity: 0.95; }\n    .body { padding: 40px 36px; color: #1f2937; line-height: 1.7; }\n    .sentiment-badge { display: inline-block; padding: 12px 28px; border-radius: 12px; font-weight: 600; font-size: 20px; background: ${sentimentColor}15; color: ${sentimentColor}; }\n    .stats { display: grid; grid-template-columns: repeat(auto-fit, minmax(180px, 1fr)); gap: 20px; margin: 36px 0; }\n    .stat { background: #f8fafc; padding: 24px; border-radius: 14px; text-align: center; border: 1px solid #e2e8f0; }\n    .stat-value { font-size: 36px; font-weight: 700; margin: 0 0 8px; color: #0f172a; }\n    .stat-label { color: #64748b; font-size: 14px; text-transform: uppercase; letter-spacing: 0.8px; }\n    .section { margin-top: 44px; }\n    .section h2 { font-size: 22px; color: #111827; margin: 0 0 18px; padding-bottom: 10px; border-bottom: 3px solid #e5e7eb; position: relative; }\n    .section h2::after { content: ''; position: absolute; bottom: -3px; left: 0; width: 60px; height: 3px; background: #6366f1; }\n    ul { padding-left: 20px; margin: 16px 0; }\n    li { margin: 14px 0; position: relative; }\n    li::marker { color: #6366f1; }\n    .critical { background: #fef2f2; border-left: 5px solid #ef4444; padding: 16px; border-radius: 0 8px 8px 0; margin: 16px 0; }\n    .badge { padding: 6px 14px; border-radius: 999px; font-size: 13px; font-weight: 600; margin-left: 8px; }\n    .high { background: #fee2e2; color: #991b1b; }\n    .footer { background: #f8fafc; padding: 32px; text-align: center; color: #64748b; font-size: 14px; border-top: 1px solid #e2e8f0; }\n  </style>\n<link rel="preload" as="image" href="https://flowpast.s3.eu-north-1.amazonaws.com/featured_blog_images/107077.webp"><link rel="preload" as="image" href="https://flowpast.s3.eu-north-1.amazonaws.com/featured_blog_images/107077.webp"></head>
\n<body>\n  <div class=\"container\">\n    <div class=\"header\">\n      <h1>${product}</h1>\n      <p>AI-Powered Review Intelligence Report \u2022 ${new Date().toLocaleDateString('en-US', { weekday: 'long', year: 'numeric', month: 'long', day: 'numeric' })}</p>\n    </div>\n\n    <div class=\"body\">\n      <div style=\"text-align:center; margin-bottom:40px;\">\n        <div class=\"sentiment-badge\">\n          ${aiAnalysis.collective_sentiment_label.toUpperCase()} SENTIMENT\n        </div>\n      </div>\n\n      <div class=\"stats\">\n        <div class=\"stat\">\n          <div class=\"stat-value\">${totalReviews || 0}</div>\n          <div class=\"stat-label\">Reviews Analyzed</div>\n        </div>\n        <div class=\"stat\">\n          <div class=\"stat-value\">${avgRating || 'N/A'}</div>\n          <div class=\"stat-label\">Average Rating</div>\n        </div>\n        <div class=\"stat\">\n          <div class=\"stat-value\" style=\"color:${healthColor}\">${aiAnalysis.overall_health?.toUpperCase()}</div>\n          <div class=\"stat-label\">Product Health</div>\n        </div>\n        <div class=\"stat\">\n          <div class=\"stat-value\" style=\"color:${churnColor}\">${aiAnalysis.churn_risk?.toUpperCase()}</div>\n          <div class=\"stat-label\">Churn Risk</div>\n        </div>\n      </div>\n\n      <div class=\"section\">\n        <h2>Top Praises</h2>\n        <ul>${(aiAnalysis.batch_analysis?.common_praise || []).map(p => `<li>${p}</li>`).join('') || '<li>No significant praise detected</li>'}</ul>\n      </div>\n\n      <div class=\"section\">\n        <h2>Common Complaints</h2>\n        <ul>${(aiAnalysis.batch_analysis?.common_complaints || []).map(c => `<li>${c}</li>`).join('') || '<li>No major complaints found</li>'}</ul>\n      </div>\n\n      ${aiAnalysis.batch_analysis?.critical_issues?.length ? `\n      <div class=\"section\">\n        <h2>Critical Issues (Immediate Attention)</h2>\n        ${aiAnalysis.batch_analysis.critical_issues.map(issue => `\n          <div class=\"critical\">\n            <strong>${issue.issue}</strong>\n            <span class=\"badge high\">${issue.severity.toUpperCase()}</span>\n            <em>\u2014 ${issue.mentions} users mentioned</em>\n          </div>\n        `).join('')}\n      </div>` : ''}\n\n      <div class=\"section\">\n        <h2>Recommended Actions</h2>\n        <ul>\n          <li><strong>Product Team:</strong> ${aiAnalysis.recommended_actions?.product_team || 'N/A'}</li>\n          <li><strong>Sales Team:</strong> ${aiAnalysis.recommended_actions?.sales_team || 'N/A'}</li>\n          <li><strong>Support Team:</strong> ${aiAnalysis.recommended_actions?.support_team || 'N/A'}</li>\n        </ul>\n      </div>\n    </div>\n\n    <div class=\"footer\">\n      Generated automatically by <strong>Thordata Review Intelligence Engine</strong><br>\n      ${new Date().toLocaleString('en-US', { dateStyle: 'long', timeStyle: 'medium' })}\n    </div>\n  </div>\n</body>\n</html>`;\n\nreturn [{\n  json: {\n    email_html: html,\n    product_name: product,\n    total_reviews: totalReviews,\n    avg_rating: avgRating,\n    ai_analysis: aiAnalysis\n  }\n}];"
},
"typeVersion": 2
},
{
"id": "202ad2fa-de6d-491b-85a3-6fe7d044af9e",
"name": "Dispatch Summary Email",
"type": "n8n-nodes-base.gmail",
"position": [
25,
590
],
"webhookId": "c419639d-02dd-41a3-83da-d93c55e5fd18",
"parameters": {
"sendTo": "[YOUR_EMAIL]",
"message": "=={{ $json.email_html }}",
"options": {
"appendAttribution": false
},
"subject": "={{ $json.product_name }} Review Analysis \u2013 {{ new Date().toLocaleDateString() }}"
},
"credentials": {
"gmailOAuth2": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 2.1
}
],
"pinData": [],
"connections": {
"Dynamic URL Composer": {
"main": [
[
{
"node": "Thordata Page Scraper",
"type": "main",
"index": 0
}
]
]
},
"Manual Execution Start": {
"main": [
[
{
"node": "Assemble Review Sources",
"type": "main",
"index": 0
}
]
]
},
"More Pages Check": {
"main": [
[
{
"node": "Increase Page Index",
"type": "main",
"index": 0
}
],
[
{
"node": "Iterate Sources Safely",
"type": "main",
"index": 0
}
]
]
},
"Incoming Webhook Trigger": {
"main": [
[
{
"node": "Assemble Review Sources",
"type": "main",
"index": 0
}
]
]
},
"Set Pagination Start": {
"main": [
[
{
"node": "Dynamic URL Composer",
"type": "main",
"index": 0
}
]
]
},
"Rate Limit Pause": {
"main": [
[
{
"node": "Dynamic URL Composer",
"type": "main",
"index": 0
}
]
]
},
"Universal Review Parser": {
"main": [
[
{
"node": "More Pages Check",
"type": "main",
"index": 0
}
]
]
},
"Increase Page Index": {
"main": [
[
{
"node": "Rate Limit Pause",
"type": "main",
"index": 0
}
]
]
},
"Assemble Review Sources": {
"main": [
[
{
"node": "Iterate Sources Safely",
"type": "main",
"index": 0
}
]
]
},
"AI Sentiment Review": {
"main": [
[
{
"node": "Build Email Report",
"type": "main",
"index": 0
}
]
]
},
"Format Data for AI": {
"main": [
[
{
"node": "Return Webhook Output",
"type": "main",
"index": 0
},
{
"node": "AI Sentiment Review",
"type": "main",
"index": 0
}
]
]
},
"Dispatch Summary Email": {
"main": [
[]
]
},
"Build Email Report": {
"main": [
[
{
"node": "Dispatch Summary Email",
"type": "main",
"index": 0
}
]
]
},
"Form Intake Sources": {
"main": [
[
{
"node": "Assemble Review Sources",
"type": "main",
"index": 0
}
]
]
},
"Deduplicate Review Summary": {
"main": [
[
{
"node": "Format Data for AI",
"type": "main",
"index": 0
}
]
]
},
"Thordata Page Scraper": {
"main": [
[
{
"node": "Universal Review Parser",
"type": "main",
"index": 0
}
]
]
},
"Iterate Sources Safely": {
"main": [
[
{
"node": "Deduplicate Review Summary",
"type": "main",
"index": 0
}
],
[
{
"node": "Set Pagination Start",
"type": "main",
"index": 0
}
]
]
}
}
}