🔍 Search Intent Mismatch Detector

📋 What This Workflow Does:

The Problem It Solves: Imagine someone searches “how to bake chocolate cake” (they want to learn), but Google shows them your cake shop’s pricing page (you want them to buy). That’s a search intent mismatch—the searcher won’t find what they need, they’ll bounce, and you’ll lose potential conversions.

This workflow automatically:

Pulls your top-performing keywords from Google Search Console
Reads the actual content of each page
Uses AI to determine if the page content matches what searchers expect
Sends you a prioritized report of pages that need fixing

🔄 Workflow Explained:

Stage 1: Setup & Data Collection

Node	What It Does
Manual Trigger	Start button—click this to run the workflow manually
⚙️ Setup Configuration	Your settings: website URL, thresholds, email address, business description. Think of it as the workflow’s control panel.
📊 GSC: Fetch Top Keywords	Connects to Google Search Console and pulls keyword data (which queries bring traffic, how many clicks, what pages rank). Uses a regex filter to focus only on main pages (homepage + first-level pages like `/services/` or `/about/`).

Stage 2: Data Processing

Node	What It Does
📋 Process & Filter GSC Data	Cleans up the GSC data: removes low-traffic keywords (below your thresholds), sorts by clicks, and keeps only the top 100 keywords. Adds helpful info like “traffic potential” scores.
Split Keywords for Processing	Takes the list of 100 keywords and breaks them into individual items, so each keyword can be processed one by one.
Batch Keywords (10 at a time)	Groups keywords into batches of 10. This prevents overwhelming the AI and web scraping services with too many requests at once.

Stage 3: Content Analysis Loop

Node	What It Does
Deduplicate Pages	Multiple keywords often point to the same page (e.g., “car insurance” and “auto insurance” both go to your insurance page). This node groups them together so we only scrape each page once—saving time and API costs.
Filter Valid URLs	Safety check: only processes URLs that start with “http”. Filters out any bad data before scraping.
Read URL Content (Jina AI)	Visits each page and extracts the text content in a clean format. If a page fails to load, it retries up to 2 times. Has a 60-second timeout for slow pages.
Prepare AI Analysis Input	Packages everything the AI needs: the page content, the keywords pointing to it, and performance metrics. Trims content to 8,000 characters to keep costs reasonable.
🧠 Claude: Intent Mismatch Analysis	The brain of the operation! Sends the data to Claude AI, which analyzes: (1) What is the searcher looking for? (2) What does this page actually offer? (3) Do they match? Returns structured analysis with severity scores and recommendations.
Parse Claude Analysis	Extracts the AI’s response and converts it to structured data the workflow can use. Handles errors gracefully if the AI response isn’t perfectly formatted.
Loop Back for Next Batch	After processing one batch, this sends the workflow back to process the next batch of 10 keywords. Continues until all keywords are analyzed.

Stage 4: Report Generation

Node	What It Does
Aggregate All Results	Collects all the individual analyses into one big dataset.
📊 Generate Final Report	Crunches the numbers: How many mismatches? What’s the severity breakdown? Which pages need immediate attention? Creates prioritized lists and calculates traffic at risk.
📧 Format HTML Email Report	Creates a beautiful, visual email report with color-coded severity badges, stats cards, and actionable recommendations.
📧 Send Email Report	Delivers the HTML report to your inbox via Gmail.

Stage 5: Data Storage

Node	What It Does
Prepare Sheets Data	Formats the keyword-level data for storage.
Create a Database Page (Notion)	Saves each keyword analysis to your Notion database for tracking over time. Stores: keyword, page URL, clicks, impressions, position, intent classification, page type, and whether a mismatch was detected.

🎯 Understanding the Intent Classifications

Keyword Intent Types

Intent	What They Want	Signal Words
INFORMATIONAL 📚	Learn something	“how to”, “what is”, “why”, “guide”, “tutorial”
TRANSACTIONAL 🛒	Buy/convert now	“buy”, “price”, “order”, “subscribe”, “near me”
COMMERCIAL_INVESTIGATION 🔍	Research before buying	“best”, “review”, “vs”, “compare”, “alternative”
NAVIGATIONAL 🧭	Find a specific page	Brand name, “login”, “official”

Page Types

Page Type	Description
PRODUCT_PAGE	Selling a product or service
PRICING_PAGE	Pricing, plans, packages
BLOG_EDUCATIONAL	How-to guides, tutorials, learning content
COMPARISON_PAGE	Reviews, comparisons, “best of” lists
LANDING_PAGE	Lead capture, signup forms
HOMEPAGE	Main site overview
CATEGORY_PAGE	Product/service category listing
SUPPORT_PAGE	Help docs, FAQs

Mismatch Examples

If Keyword Intent Is…	But Page Type Is…	Problem
INFORMATIONAL	PRODUCT_PAGE	Person wants to learn, but you’re selling → They’ll bounce
TRANSACTIONAL	BLOG_EDUCATIONAL	Person wants to buy, but you’re teaching → Lost sale
COMMERCIAL_INVESTIGATION	HOMEPAGE	Person wants comparisons, but sees brand overview → Not helpful

⚙️ Setup Instructions

Prerequisites

n8n instance (self-hosted or cloud)
Google Search Console property verified
Anthropic API key (for Claude)
Gmail account with OAuth2
Notion account with database created
Jina AI account (free tier works)

Step 1: Import the Workflow

Open n8n → Settings → Import
Upload the JSON file
Save the workflow

Step 2: Configure Credentials

Create these credentials in n8n (Settings → Credentials):

Credential	Used For
Google Search Console OAuth2	Fetching keyword data
Jina AI API	Reading page content
Anthropic API	Claude AI analysis
Gmail OAuth2	Sending email reports
Notion API	Storing results

Step 3: Set Up Notion Database

Create a Notion database with these properties:

Property	Type
Name (Title)	Title (for keyword)
Page URL	URL
Clicks	Number
Impressions	Number
Position	Number
Keyword Intent	Rich Text
Page Type	Rich Text
Mismatch Detected	Rich Text

Step 4: Configure the Workflow

Edit the ⚙️ Setup Configuration node:

{
  "site_url": "<https://yoursite.com/>",
  "date_range_days": 30,
  "min_clicks_threshold": 1,
  "min_impressions_threshold": 10,
  "top_keywords_limit": 100,
  "report_email": "your@email.com",
  "business_context": "Brief description of your business"
}

Update the GSC node regex filter if needed (currently filters for root + first-level pages only).

Update the Notion node with your database ID.

Step 5: Test & Activate

Click “Test Workflow” to run manually
Check for errors
Toggle to “Active” for scheduled runs

💰 Cost Estimates

Service	Cost	Notes
Claude Sonnet 4	~$0.90/run	100 keywords ≈ 300K tokens
Jina AI	Free	1,000 requests/month included
n8n	Varies	Self-hosted free, cloud varies
Total	~$4/month	Weekly runs

🔧 Customization Options

Change the Schedule

Add a Schedule Trigger node connected to Setup Configuration:

Cron expression: 0 9 * * 1 = Every Monday at 9 AM

Adjust Thresholds

In Setup Configuration, change:

min_clicks_threshold: Minimum clicks to include a keyword
min_impressions_threshold: Minimum impressions
top_keywords_limit: How many keywords to analyze (max 100 recommended)

Change the GSC Filter

The current regex ^https:\\/\\/yoursite\\.com\\/$|^https:\\/\\/yoursite\\.com\\/[^\\/]+\\/?$ only includes:

Homepage
First-level pages (e.g., /about/, /services/)

To include deeper pages, modify or remove this filter in the GSC node.

🐛 Troubleshooting

Issue	Solution
GSC returns no data	Check OAuth2 credentials, verify site property exists
Jina timeout errors	Normal for slow pages—the retry will handle most cases
Claude parse errors	Check AI response in execution logs—may need prompt adjustment
Notion not saving	Verify database ID and property names match exactly
Email not sending	Check Gmail OAuth2 credentials, verify “less secure apps” settings

📊 Understanding Your Report

Severity Levels

Level	Score	Meaning	Action
HIGH 🔴	76-100	Critical mismatch, major CVR loss	Fix immediately
MEDIUM 🟠	51-75	Significant opportunity	Plan to address
LOW 🔵	21-50	Minor optimization needed	Nice-to-have
ALIGNED 🟢	0-20	Content matches intent	No action needed

Key Metrics

Mismatch Rate: % of keywords with intent mismatches
Traffic at Risk: Total clicks going to mismatched pages
CVR Opportunity: Estimated conversion rate improvement if fixed

📈 Best Practices

Run weekly to catch new mismatches as rankings change
Focus on HIGH severity first - biggest impact
Track improvements in Notion database over time
Create new content for keywords that don’t have a matching page
Optimize existing pages by adding sections that match searcher intent

📝 Changelog

v1.0.0 - Initial Release

GSC data collection with page-level regex filter
Jina AI native node with retry logic
Claude Sonnet 4 analysis
Notion database storage
HTML email reports with severity badges
Intent classification: 4 types
Page type classification: 8 types
Severity scoring: 0-100 scale

FULL JSON CODE:

	Search_Intent_Mismatch_Detector_Sanitized.json41.99 KB • JSON File
Download