# Clay Workflow - PHT Outreach Bot ## Overview Single Clay table that finds → enriches → personalizes → exports 10 prospects/day to Instantly. --- ## Clay Table Structure ``` ┌──────────────────────────────────────────────────────────────────┐ │ CLAY TABLE: PHT Prospects │ └──────────────────────────────────────────────────────────────────┘ COLUMNS (left to right): 1. Company Name [Input from scraper] 2. Location [Auto-filled] 3. Website [Auto-filled] 4. Cold Room Count [Enrichment] 5. Fruit Type [Enrichment] 6. ✓ Qualified? [Formula: rooms ≥10 + fruit match] --- 7. Contact Name [Apollo People Search] 8. Job Title [Apollo] 9. Email [Apollo → Hunter fallback] 10. LinkedIn URL [Apollo] 11. Email Valid? [ZeroBounce check] --- 12. Website Copy [Scrape homepage] 13. Pain Points [GPT-4 extract from copy] 14. Similar Customer [Lookup: Zespri/Stemilt/Costa based on location+fruit] --- 15. Email Subject [GPT-4 generate] 16. Email Body [GPT-4 generate personalized] 17. LinkedIn Message [GPT-4 generate] --- 18. Sent Date [Timestamp when exported] 19. Status [Queued/Sent/Replied/Bounced] ``` --- ## Workflow Steps (Clay Enrichments) ### STEP 1: Find Companies (Input) **Option A: Google Maps Scraper** (Clay native integration) - Search: "cold storage apples [location]" - Search: "packhouse citrus [location]" - Runs daily, adds 50 new rows **Option B: Upload CSV** (from Apify/Apollo) - Drag-drop your existing MASTER-LIST-FINAL.csv - Maps columns automatically **Option C: Apollo Organization Search** (Clay integration) - Industry: "Food & Beverage" - Keywords: "cold storage" OR "packhouse" OR "cool room" - Employee count: 20-500 - Runs as scheduled enrichment --- ### STEP 2: Enrich Company Data ``` Column: Cold Room Count ┌─────────────────────────────────────┐ │ GPT-4 Web Scraper │ │ Prompt: "Visit {website} and find │ │ mentions of cold storage rooms, │ │ CA rooms, or facility size. │ │ Return number of rooms if found." │ └─────────────────────────────────────┘ ↓ "50 CA rooms" → 50 "No data" → Skip row ``` ``` Column: Fruit Type ┌─────────────────────────────────────┐ │ GPT-4 Web Scraper │ │ Prompt: "What fruits does │ │ {company_name} store or pack? │ │ Return: apples, citrus, pears..." │ └─────────────────────────────────────┘ ↓ Filter: Keep only apples/citrus/pears ``` ``` Column: ✓ Qualified? ┌─────────────────────────────────────┐ │ Formula │ │ IF(rooms ≥ 10 AND fruit IN │ │ ["apples","citrus","pears"], TRUE) │ └─────────────────────────────────────┘ ↓ Filters table to qualified only ``` --- ### STEP 3: Find People ``` Column: Contact Name, Job Title, Email, LinkedIn ┌──────────────────────────────────────────────────┐ │ Apollo People Search │ │ Input: {company_name} │ │ Filters: │ │ - Title contains: "Quality" OR "Operations" │ │ OR "General Manager" OR "Cold Storage" │ │ - Seniority: Manager, Director, VP │ │ Return: Top 3 contacts │ └──────────────────────────────────────────────────┘ ↓ Pick #1 (highest seniority) ``` **Fallback enrichments:** ``` If Apollo email = null: ↓ Hunter.io email finder ↓ If still null: ↓ Prospeo (Clay integration) ↓ If still null → Skip row ``` --- ### STEP 4: Validate Email ``` Column: Email Valid? ┌─────────────────────────────────────┐ │ ZeroBounce Email Validation │ │ (or MillionVerifier) │ │ Input: {email} │ │ Return: Valid/Invalid/Risky │ └─────────────────────────────────────┘ ↓ Filter: Keep only "Valid" ``` --- ### STEP 5: Scrape Website for Context ``` Column: Website Copy ┌─────────────────────────────────────┐ │ Clay Web Scraper │ │ Scrape: {website}/about │ │ Extract: First 500 words │ └─────────────────────────────────────┘ ``` ``` Column: Pain Points ┌─────────────────────────────────────┐ │ GPT-4 Analysis │ │ Prompt: "Read this website copy: │ │ {website_copy} │ │ │ │ Identify 1-2 pain points related │ │ to fruit storage quality control, │ │ export compliance, or post-harvest │ │ management. Be specific." │ └─────────────────────────────────────┘ ↓ "Exports to EU = strict ethylene limits" "Multi-variety storage = ripening sync issues" ``` ``` Column: Similar Customer ┌─────────────────────────────────────┐ │ Formula / Lookup │ │ IF location = NZ → "Zespri" │ │ IF location = USA + citrus → │ │ "Wonderful Citrus" │ │ IF location = USA + apples → │ │ "Stemilt" │ │ IF location = AU → "Costa Group" │ └─────────────────────────────────────┘ ``` --- ### STEP 6: Generate Personalized Emails ``` Column: Email Subject ┌─────────────────────────────────────────────────┐ │ GPT-4 Writer │ │ Prompt: "Write 1 email subject line for │ │ {contact_name} at {company_name}. │ │ │ │ Context: │ │ - They store {fruit_type} │ │ - Pain point: {pain_points} │ │ - We help companies like {similar_customer} │ │ │ │ Style: Curious, helpful, NOT salesy. │ │ Length: 4-7 words max. │ │ NO "Quick question" or "Touching base" │ │ │ │ Examples: │ │ - "Export rejections at {company}?" │ │ - "How {similar_customer} cut waste 40%" │ │ - "Ethylene question for {contact_name}" │ │ │ │ Return ONLY the subject line." │ └─────────────────────────────────────────────────┘ ``` ``` Column: Email Body ┌─────────────────────────────────────────────────┐ │ GPT-4 Writer │ │ Prompt: "Write a personalized cold email │ │ from Jonny Shannon to {contact_name}, │ │ {job_title} at {company_name}. │ │ │ │ Context: │ │ - They store {fruit_type} in {location} │ │ - Likely pain: {pain_points} │ │ - We help {similar_customer} monitor ethylene │ │ to prevent premature ripening/waste │ │ │ │ Structure: │ │ 1. Personal observation (1 sentence) │ │ 2. Relevant pain point (1 sentence) │ │ 3. How we help similar company (1 sentence) │ │ 4. Soft CTA: offer video walkthrough │ │ │ │ Tone: Helpful, genuine, brief (60 words max) │ │ NO hype, NO "revolutionize", NO "game-changer"│ │ │ │ End with: │ │ 'I recorded a quick video showing how │ │ [similar_customer] uses it. Want me to send │ │ it over?' │ │ │ │ Sign: Jonny (no last name, keep casual) │ │ │ │ Return ONLY the email body." │ └─────────────────────────────────────────────────┘ ``` **Example Output:** ``` Subject: How Zespri cut ethylene spikes by 60% Hi Sarah, Saw FreshPack exports kiwis to 40+ countries — keeping consistent ripeness across that many shipments must be challenging. We help Zespri monitor ethylene in real-time so they catch hot spots before fruit goes out. Cut their rejections by 60% last season. I recorded a quick video showing how they use it. Want me to send it over? Jonny ``` --- ### STEP 7: LinkedIn Message (Parallel Track) ``` Column: LinkedIn Message ┌─────────────────────────────────────────────────┐ │ GPT-4 Writer │ │ Prompt: "Write a LinkedIn connection message │ │ to {contact_name}. │ │ │ │ Context: {company_name} stores {fruit_type} │ │ We help similar companies monitor ethylene. │ │ │ │ Length: 250 characters max (LinkedIn limit) │ │ Tone: Casual, no pitch, just connection │ │ │ │ Example: │ │ 'Hey Sarah — saw you're managing quality at │ │ FreshPack. We work with a few kiwi exporters, │ │ would be great to connect!' │ │ │ │ Return ONLY the message." │ └─────────────────────────────────────────────────┘ ``` --- ## Export & Automation ### Daily Schedule (Clay Automation) ``` 6:00 AM NZDT - Clay Runs ├─ Add 50 new companies (Google Maps scraper) ├─ Enrich all new rows (Steps 2-6) ├─ Filter: Qualified = TRUE, Email Valid = TRUE ├─ Sort by: Company size (largest first) ├─ Select: Top 10 rows └─ Export to Instantly webhook ``` ### Instantly Integration ``` Clay Export Webhook → Instantly API ┌─────────────────────────────────────┐ │ Instantly Campaign: "PHT Daily 10" │ │ │ │ Sends: │ │ - 2 emails from Account 1 (8am) │ │ - 2 emails from Account 2 (10am) │ │ - 2 emails from Account 3 (12pm) │ │ - 2 emails from Account 4 (2pm) │ │ - 2 emails from Account 5 (4pm) │ │ │ │ Follow-ups: │ │ - Day 3: Value add (case study) │ │ - Day 8: Final bump (breakup) │ └─────────────────────────────────────┘ ``` ### PhantomBuster Integration (LinkedIn) ``` Clay Export CSV → PhantomBuster ┌─────────────────────────────────────┐ │ Phantom: LinkedIn Auto-Connect │ │ │ │ Inputs from Clay: │ │ - {linkedin_url} │ │ - {linkedin_message} │ │ │ │ Daily limit: 20 connections │ │ Runs: 9am NZDT daily │ └─────────────────────────────────────┘ ``` --- ## Monitoring Dashboard (Clay Views) ### View 1: Pipeline ``` ┌──────────┬───────┬──────────┬─────────┬─────────┐ │ Prospect │ Sent │ Opened │ Replied │ Booked │ ├──────────┼───────┼──────────┼─────────┼─────────┤ │ Found │ 347 │ — │ — │ — │ │ Enriched │ 289 │ — │ — │ — │ │ Queued │ 30 │ — │ — │ — │ │ Sent │ — │ 178 │ — │ — │ │ Opened │ — │ — │ 47 │ — │ │ Replied │ — │ — │ — │ 12 │ └──────────┴───────┴──────────┴─────────┴─────────┘ ``` ### View 2: Daily Output (Today) ``` ✅ 10 emails queued in Instantly ✅ 20 LinkedIn connections sent via Phantom ✅ 3 replies from yesterday (flagged in Telegram) ⚠️ 2 bounces (re-verify emails) ``` --- ## Cost Breakdown **Clay:** $149/mo (Starter plan) - Unlimited tables - 2,000 enrichment credits/mo - GPT-4 included **Instantly:** $37/mo (Growth) - 5 email accounts - Unlimited sends - Warmup included **PhantomBuster:** $56/mo (Starter) - 20 LinkedIn connects/day - Auto-messaging **ZeroBounce:** $16/mo (pay as you go) - Email validation **Total: $258/mo** --- ## Build Timeline **Day 1:** Set up Clay table, import existing prospect list **Day 2:** Configure enrichment waterfall (Apollo → Hunter → Prospeo) **Day 3:** Build GPT-4 personalization prompts, test outputs **Day 4:** Connect Instantly webhook, test 1 email send **Day 5:** Add PhantomBuster, set daily schedules **Day 6:** Monitor + refine prompts based on open rates --- ## Success Metrics (Week 1) - **50 emails sent** (10/day × 5 days) - **100 LinkedIn connections** (20/day × 5 days) - **Target: 10% reply rate** = 5 interested prospects - **Target: 2 booked calls** If we hit that → scale to 20/day Week 2. --- Want me to start building this in Clay? I'll need: 1. Clay account login (or I can sign up with maxjasonweber@gmail.com) 2. Apollo API key 3. Instantly webhook URL