Restaurant Lead Intelligence Platform

End-to-end lead generation and enrichment pipeline for the restaurant vertical. Built at Shipday to transform raw restaurant data into qualified, actionable leads with verified contact information.

Problem

I was handed lead lists of 10,000+ restaurants with no qualification, no prioritization, and 60%+ missing contact information. Manual research took 4-5 minutes per lead. At that rate, properly qualifying the list would take 100+ hours of pure data entry. Meanwhile, I was expected to be on the phone selling.

The lists also had no ICP filtering. They included every type of restaurant from fine dining to food trucks, with no way to identify which ones were actually good prospects for a delivery logistics platform.

Solution

A multi-workflow system that handles discovery, enrichment, scoring, and delivery:

Workflow 1: ToastTab Lead Generator (toasttab-lead-gen.json)

• Uses Serper API to search Google for site:toasttab.com/local across target states
• ToastTab presence is itself a buying signal: it means the restaurant has a modern POS system and is likely technology-forward
• Extracts and deduplicates restaurant URLs from search results
• Scrapes each ToastTab page for business name, address, phone, cuisine type, website
• Searches Google for restaurant email addresses using business name + city + "contact email"
• Uses Claude API to intelligently extract email addresses from search result snippets
• Writes enriched leads to Google Sheets with status tracking
• Covers 6+ states: CT, MA, MD, MN, WI, IN

Workflow 2: Email Enrichment Pipeline (email-enrichment.json)

• Reads restaurants from the master Google Sheet that need email enrichment
• Runs multi-source search: Google results, Facebook pages, official website scraping
• Claude extracts and validates email addresses from unstructured web content
• Confidence scoring based on email domain match and prefix type
• Prioritizes business emails (info@, contact@, orders@) over generic ones
• Updates the master sheet with results, confidence scores, and enrichment timestamps

Workflow 3: Google Maps Discovery (google-maps-scraper.json)

• Alternative discovery method using Google Maps/Places API
• Searches for restaurants by cuisine type and location
• Extracts business details, ratings, review counts, operating hours
• Feeds into the same enrichment pipeline

Scoring Algorithm

The 150-point scoring system evaluates 12+ operational signals:

Signal	Weight	What It Indicates
Cuisine type (Pizza/Italian)	High	Delivery-heavy cuisine = higher platform need
Google rating (below 4.0)	Medium	Potential operational challenges = pain point
Review count (high volume)	Medium	Established business with customer base
Third-party delivery presence	High	Already doing delivery = easier adoption
Delivery volume indicators	High	High volume = more revenue potential
POS system type	Medium	Tech-forward = more likely to adopt
Driver status	High	"Hiring drivers" = active delivery operation
Location count	Medium	Multi-location = larger deal potential
Website quality	Low	Professional online presence = business maturity
Pain point indicators	High	Mentions of delivery challenges in reviews
Years in business	Low	Established = more stable revenue
Operating hours	Low	Late-night/extended hours = delivery demand

Email Extraction Methodology

The Claude-powered email extraction handles cases that regex cannot:

• Obfuscated emails: "info [at] restaurant [dot] com"
• Context-dependent extraction: Distinguishing business emails from platform emails
• Confidence assessment: Domain match to business name increases confidence
• Exclusion logic: Filters out noreply@, support@wix.com, @facebook.com, test@, example.com
• Fallback: If Claude extraction fails, raw regex serves as backup

Discovery rate: 65-80% of restaurants processed yield at least one verified email address.

Results

• Processed 1,000+ restaurants across CT and MA markets
• 600+ verified contacts extracted for targeted outreach
• $6.49 average cost per state run
• 15-20 hours/week saved on manual research
• 85%+ data confidence on enriched records
• System scales to any state by adding search queries (no code changes needed)

Files

File	Description
workflows/toasttab-lead-gen.json	ToastTab discovery via Google Search
workflows/email-enrichment.json	Production email enrichment pipeline
workflows/google-maps-scraper.json	Google Maps based discovery
database/schema.sql	PostgreSQL table definitions
docs/scoring-algorithm.md	Full 150-point scoring breakdown
docs/email-extraction.md	AI extraction methodology and examples

Transferability

This architecture is vertical-agnostic. To adapt it to a different market:

1. Change the discovery source -- Replace ToastTab with G2, Crunchbase, or any directory
2. Adjust ICP criteria -- Replace cuisine type with tech stack, replace review count with employee count
3. Update enrichment sources -- Different verticals have different data sources
4. Modify scoring weights -- Each market has different signals that matter

The pipeline structure (Discover, Enrich, Score, Deliver) stays exactly the same.

Setup

1. Import workflow JSON files into your n8n instance
2. Configure credentials: Serper API, Anthropic, Google Sheets, ScraperAPI
3. Create a Google Sheet with the expected column structure (see schema)
4. Update target states and search queries in the "Generate Search Queries" node
5. Run manually for first batch, then enable scheduling