phone-verification-waterfall

Maintain a continuously refreshed inventory of callable leads with verified phone numbers. Automates the waterfall from HubSpot lead identification → Apollo phone verification → Clay enrichment (for Apollo misses) → HubSpot sync. Ensures Tim ALWAYS has 50+ verified dial prospects queued and ready, eliminating "nobody to call" downtime and maximizing daily dial velocity.

Built on: Lead lifecycle → phone verification → queue prioritization by ICP + intent signals.

<quick_start> Standalone run (manual trigger): "verify phones" or "phone waterfall" or "callable leads" or "who can I call" → pulls all HubSpot leads missing phones → enriches via Apollo + Clay → writes back to HubSpot → outputs sortable callable queue

Scheduled run: Every Monday & Wednesday 6:15 AM (between prospect-enrich 6:00 AM and prospect-refresh 6:30 AM) — keeps phone inventory fresh twice-weekly and allows newly refreshed leads to be phone-verified before daily calls

Trigger phrases:

• "verify phones"
• "phone waterfall"
• "callable leads"
• "who can I call"
• "dial list"
• "always have someone to call"
• "phone check"
• "enrich phones" </quick_start>

<success_criteria>

• HubSpot lead pull returns 100+ contacts missing phones from all ICP verticals
• Apollo phone lookup succeeds for 40-50% of contacts (baseline)
• Clay waterfall fills 60-70% of remaining Apollo misses
• HubSpot phone field updated for 70%+ of missing numbers
• Final callable queue contains 50+ contacts sorted by ICP score + intent
• Zero contacts belong to customers (lifecycle_stage = 'customer') — Golden Rules block them
• Zero contacts owned by Account Executives (Lex, Phil, Ron, Anthony) — no AE lead theft
• Zero channel partners or existing device owners included
• Execution time under 8 minutes (HubSpot → Apollo → Clay → sync) </success_criteria>

Pipeline Stages

STAGE 1: PULL LEADS          STAGE 2: APOLLO PHONE      STAGE 3: CLAY WATERFALL    STAGE 4: HUBSPOT SYNC    STAGE 5: CALLABLE QUEUE
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
HubSpot Query       →     Apollo People Match    →    Clay Enrichment       →   Update Phone Field   →    Prioritized CLI
phone = null/empty  →     (Name + Email + Org)   →    (Multiple Providers)  →   (API bulk update)    →    ICP Score Sort
lifecycle_stage     →     Return: Verified #     →    Return: Verified #    →   Mark: enriched_at    →    Intent Signals
!= 'customer'       →     Success Rate: 40-50%   →    Success Rate: 30-40%  →   (Contact & Company) →    Output: CSV + Chat
No Excluded Owner   →     (No cost)              →    (Find + Enrich)        →   Log completion       →    Time-to-dial: <5s
Golden Rules Pass   →                            →                          →                        →

Stage 1: Pull Leads Needing Phones

HubSpot Query:

Criteria (ALL must match):
- phone IS NULL OR phone = '' (empty string)
- lifecyclestage != 'customer'
- AE-owned (82625923, 423155215, 190030668): EXCLUDE if <90d activity. KEEP as STALE AE LEAD if >90d inactive (Tim reviews ATL/BTL + value). Ron=all, Lex/Phil=NA only.

Golden Rules Hard Gate (SKIP if ANY match):
- first_conversion contains ['Pearl', 'setup', 'Connect', 'signup']
  (Existing product engagers — not cold prospects)
- company.device_count >= 1
  (Already owns equipment — likely existing relationship)
- is_channel = true
  (Channel partners excluded)
- engagement_overview contains ['Pearl', 'trial', 'demo', 'customer']
  (Existing touchpoints)

ATL/BTL Classification Gate (applied after Golden Rules)

Classify each contact by job title using approved ATL/BTL definitions (see CLAUDE.md § ATL/BTL Classification v1.0):

ATL Title Keywords (budget authority — PRIORITY DIAL): Chief (CIO, CTO, CFO, COO) • VP (AVP, SVP, EVP) • President • Provost • Vice Provost • Superintendent • Director (of IT, Technology, Facilities, Academic Technology, Procurement, Materials Mgmt, Medical Education, Court Administration) • Dean • Court Administrator • Clerk of Court • City Manager • County Manager • Senior Pastor • Executive Pastor

BTL Title Keywords (no budget authority — LOWER PRIORITY): Technician • Specialist • Coordinator • Support • Administrator (Systems/Network/Database) • Engineer (AV/Network/Systems) • Operator • Instructor/Professor/Faculty • Designer • Assistant • Clerk (non-Court Admin) • Volunteer • Intern • Student • Resident • Help Desk

NEVER ATL — Auto-classify BTL regardless of other signals: Warehouse Manager • Network Manager • Systems Administrator • AV Technician • Graphic Design Instructor • Program Administrator • Web Designer • Classroom Support • Lab Coordinator • Maintenance • Building Engineer • Multimedia Services Manager • Video Production Specialist • Streaming Crew

Gray Zone — Flag for manual review: Manager (AV/Facilities/IT) — ATL only if reports to Director+ AND has delegated budget >$25K Department Chair • Program Director • Director of Educational Technology

Tag each contact: atl_btl_tier = 'ATL' | 'BTL' | 'GRAY'

Output: Unsorted list of 100+ contacts with:

• contact_id (HubSpot)
• first_name, last_name
• email (primary business email)
• company_name
• hs_job_title
• hubspot_owner_id, hubspot_owner_name
• company.industry, company.num_employees, company.hs_lead_status
• first_conversion (for Golden Rules check)

Tool: hubspot_search_contacts with filters applied

</Stage 1>

Stage S — Suppression Gate

Before phone enrichment spend, exclude suppressed contacts:

• EXCLUDE if bdr_suppression_until IS SET AND bdr_suppression_until > TODAY
• INCLUDE if bdr_suppression_until IS NOT SET (never suppressed)
• INCLUDE if bdr_suppression_until < TODAY (cooling period expired)

HubSpot filter: propertyName: "bdr_suppression_until", operator: "NOT_HAS_PROPERTY" OR operator: "LT", value: TODAY_ISO Reference: lead-suppression-spec (bdr_suppressed, bdr_suppression_reason, bdr_suppression_until)

---

Stage 2: Apollo Phone Lookup

For each contact from Stage 1, run in parallel batches:

Input:  contact_id, first_name, last_name, email, company_name
        ↓
apollo_people_match(
  first_name: "{first_name}",
  last_name: "{last_name}",
  email: "{email}",
  organization_name: "{company_name}",
  domain: "{company_domain}",
  reveal_personal_emails: false
)
        ↓
Output: phone (mobile_phone or corporate_phone, verified)
        ↓
Match rate: 40-50% success (Apollo baseline)
        ↓
NO COST — Apollo is included free

Logic:

• Apollo returns mobile_phone or corporate_phone (both acceptable for dial)
• If phone returned: tag contact apollo_phone_source = true
• If no phone: move to Stage 3 (Clay waterfall)
• Keep all results for final merge

Tool: apollo_people_match (batch parallel)

</Stage 2>

Stage 3: Clay Waterfall Enrichment

For contacts where Apollo returned NO phone, use Clay:

Option A: Company-level enrichment (recommended for speed)

find-and-enrich-contacts-at-company(
  companyIdentifier: "{company_domain}",
  contactFilters: {
    job_title_keywords: extracted from hs_job_title,
    locations: optional if available
  },
  dataPoints: {
    contactDataPoints: [
      { type: "Email" },
      { type: "Summarize Work History" }
    ]
  }
)

Then for contacts found, run add-contact-data-points to enrich phones:

add-contact-data-points(
  taskId: "{from find-and-enrich response}",
  dataPoints: [
    { type: "Custom", customDataPoint: "phone number from multiple providers" }
  ],
  entityIds: [extracted contact IDs]
)

Option B: Direct contact enrichment (if company approach yields few)

find-and-enrich-list-of-contacts(
  contactIdentifiers: [
    {
      contactName: "{first_name} {last_name}",
      companyIdentifier: "{company_domain}"
    }
  ],
  dataPoints: {
    contactDataPoints: [
      { type: "Custom", customDataPoint: "verified phone number" }
    ]
  }
)

Expected yield:

• 30-40% of Apollo misses will get phones via Clay
• Combined success rate: 40-50% (Apollo) + 30-40% of remainder (Clay) ≈ 65-70% total

Cost: Clay credits consumed (monitor spend against monthly allocation)

Tool: find-and-enrich-contacts-at-company or find-and-enrich-list-of-contacts

</Stage 3>

Stage 4: HubSpot Sync

For all contacts with verified phones (Apollo + Clay combined):

Batch update HubSpot contacts:
For each contact:
  UPDATE contact {contact_id}
  SET:
    phone = "{verified_phone}"
    phone_source = "apollo" OR "clay" (track origin)
    phone_verified_at = NOW()
    phone_waterfall_pass = true

Log event to company:
  ADD ACTIVITY: "Phone verified via Apollo/Clay waterfall"
  (Creates audit trail for Sales)

Sync strategy:

• Use HubSpot API in batches (500 contacts per batch)
• Tag with phone_source (apollo, clay, manual)
• Tag with phone_verified_at timestamp
• Log completion to deal/company object if linked

Tool: HubSpot REST API (contacts batch update) via hubspot_update_contact (per-contact; no batch MCP tool) or direct HTTP

</Stage 4>

Stage 5: Callable Queue & Prioritization

Build final output sorted by:

1. ATL/BTL Tier (primary): ATL contacts first, then GRAY, then BTL

2. ICP Vertical Score (secondary):

   Higher Ed ..................... 90
   Courts/Legal ................... 85
   Government ..................... 80
   Corporate AV ................... 80
   Healthcare ..................... 75
   Houses of Worship .............. 70
   K-12 Schools ................... 65
   

3. Intent Signals (tertiary):

   • Hiring active AV/IT roles: +15 points
   • Recent facility expansion news: +10 points
   • Tech stack shows obsolete AV (Extron, Matrox replacement signal): +12 points
   • Published RFP or procurement event: +10 points
   • Website mentions studio/control room/broadcast: +8 points
   • Recent funding/acquisition: +5 points

4. Engagement Recency (quaternary):

   • Touched within 30 days: top of queue
   • Touched 31-60 days: middle
   • No recent touch: bottom (but still callable)

Final Callable Queue Format:

See reference/callable-queue-format.md for the full ASCII queue output template (summary block, ready-to-dial table, dialer quick links, next steps).

Output formats:

• Chat: Rich formatted table (template in reference/callable-queue-format.md) with top 10-20 by ICP score
• CSV download: Full queue with all data for import to dialpad/Aircall
• By Vertical: Secondary view grouped by industry for vertical hunting
• Integration: Link to import queue to Dialpad, Aircall, or sales engagement tool

Tool: Assemble from Stage 4 output, sort, format, and present

</Stage 5>

Reply-Scan Integration

Automated queuing from reply scans (10 AM, 1 PM, 3 PM):

When reply scans discover new leads with missing phone numbers:

• These leads are automatically queued for the next scheduled phone waterfall run (Monday or Wednesday 6:15 AM)
• Standard enrichment workflow applies (Apollo → Clay waterfall → HubSpot sync)

Urgent high-ICP lead handling:

• For leads scoring 80+ ICP and deemed high-priority prospects, trigger an immediate single-contact Apollo lookup
• If Apollo succeeds, append phone to contact record and add to current day's dial queue
• If Apollo fails, hold for next scheduled waterfall run (standard workflow)

Logic:

IF reply_scan_found_lead AND lead.phone IS NULL
  IF lead.icp_score >= 80 AND lead.priority = 'high'
    → TRIGGER immediate apollo_people_match(lead)
    → IF success: append phone, mark urgent_dial = true
    → ELSE: queue for next waterfall run
  ELSE
    → Queue for next scheduled waterfall (Mon or Wed 6:15 AM)

This ensures urgent prospects get callable status same-day while maintaining efficiency of batch enrichment for standard leads.

Scheduled Automation

Cron: Every Monday & Wednesday 6:15 AM (UTC, or Tim's local timezone)

MONDAY MORNING SEQUENCE:
6:00 AM  → bdr-v3-prospect-enrich (refresh prospect company data)
6:15 AM  → phone-verification-waterfall-skill (verify phones on refreshed data)
6:30 AM  → bdr-v3-prospect-refresh (refresh deal velocity metrics)

WEDNESDAY MORNING SEQUENCE (mid-week refresh):
6:00 AM  → bdr-v3-prospect-enrich (refresh prospect company data)
6:15 AM  → bdr-v3-phone-waterfall-midweek (verify phones on mid-week refreshed data)
6:30 AM  → bdr-v3-prospect-refresh (refresh deal velocity metrics)

Why 6:15 AM Monday & Wednesday?

• Monday: Sunday evening prospect research populates HubSpot with new prospects; 6:00 AM enrich ensures firmographic data is current; 6:15 AM phone verification catches those prospects while fresh; 6:30 AM velocity refresh gives Tim complete picture before first calls
• Wednesday: Mid-week lead refresh captures newly added prospects from Tue-Wed research; ensures stable of callable leads remains 50+ through week; separate task (bdr-v3-phone-waterfall-midweek) allows independent scheduling and monitoring

Operational notes:

• If phone waterfall takes >8 min, send Slack alert (indicates API slowness or large backlog)
• Log success rate to dashboard (aim for 65%+ total verified)
• If Apple/Apollo API rate limits hit, queue and retry with exponential backoff
• On error: skip Clay enrichment and return Apollo-only results (better to have some phones than none)

## Required MCP Tools

Epiphan CRM (primary data source)

• hubspot_search_contacts — Pull leads with phone = null, apply Golden Rules filters
• hubspot_get_contact — Retrieve full contact details if needed
• crm_search_customers — Check company against Epiphan customer database (device_count, channel flag)

Apollo (phone verification stage 2)

• apollo_people_match — Look up person by name + email + company, return verified phone
• (No cost — included in Apollo plan)

Clay (phone waterfall stage 3)

• find-and-enrich-contacts-at-company — Batch enrich by company + job title filters
• add-contact-data-points — Add custom phone data point to enrich results
• find-and-enrich-list-of-contacts — Enrich specific contacts by name + company
• get-task — Poll for async enrichment results
• (Costs Clay credits — monitor spend)

HubSpot API (sync stage 4)

• hubspot_update_contact (per-contact loop; batch only via direct REST API) — Write verified phones back to HubSpot
• Or native HubSpot bulk import if preferred

Sibling Skills & Related Tasks Referenced

• prospect-research-to-cadence-skill — Lead quality gates, Golden Rules filter, ICP scoring
• sales-revenue-skill — ICP vertical definitions, intent signal weighting
• hubspot-revops-skill — HubSpot bulk operations, contact filtering patterns, owner ID mapping
• CLAUDE.md § ATL/BTL Classification v1.0 — Approved title keyword patterns for ATL/BTL/Gray classification (2026-03-17)

Upstream & Downstream Scheduled Tasks

• bdr-v3-prospect-enrich — Monday 6:00 AM firmographic refresh (runs before phone waterfall)
• bdr-v3-phone-waterfall-midweek — Wednesday 6:15 AM mid-week phone verification run (NEW)
• bdr-v3-prospect-refresh — Monday 6:30 AM velocity refresh (runs after phone waterfall)

External Dependencies

• HubSpot API: Contact search + bulk update
• Apollo API: People match (no additional cost)
• Clay API: Waterfall enrichment (requires active subscription + available credits)
• Dialpad/Aircall API: Optional integration for direct dial queue sync

Estimated Monthly Cost

• Apollo: $0 (included)
• Clay: $150-300 depending on contacts enriched (monitor waterfall stage)
• HubSpot: Already in use
• Total variable: ~$200-300/month (Clay credits)

<scheduled_automation>

Task IDs:

• bdr-v3-phone-waterfall (Monday 6:15 AM run)
• bdr-v3-phone-waterfall-midweek (Wednesday 6:15 AM run)

Trigger: Recurring, every Monday & Wednesday at 6:15 AM (user's local timezone)

Slack Notification on Completion:

✅ Phone Waterfall Complete (Mon 6:15 AM)
   Pulled:       {N} leads
   Verified:     {X+Y} phones ({%})
   HubSpot Sync: ✓
   Queue:        Ready to dial
   Leads/hour:   {rate}

On Error:

⚠️ Phone Waterfall Partial (Mon 6:15 AM)
   Apollo completed ({X%})
   Clay failing — check credits
   Manual action: Continue with Apollo-only queue

</scheduled_automation>

---

Reference: Golden Rules Filter

These are hard gates — if ANY condition matches, skip the contact entirely:

Condition	Value	Action
lifecyclestage	'customer'	SKIP
first_conversion	contains 'Pearl' OR 'setup' OR 'Connect'	SKIP
Company device_count	>= 1	SKIP
Company is_channel	true	SKIP
hubspot_owner_id	82625923, 423155215, 190030668	90-day stale exception: KEEP if no activity >90d (Tim reviews ATL/BTL + value). SKIP if <90d. Ron=all, Lex/Phil=NA only.

</Golden Rules>

---

Example: Phone Waterfall in Action

See reference/waterfall-example.md for a full worked run (247 leads pulled -> 159 verified callable, 64.4% success, 7 min 30 sec).

Emit Outcome Sidecar

Write ~/.claude/skill-analytics/last-outcome-phone-verification-waterfall.json: {"ts":"[UTC ISO8601]","skill":"phone-verification-waterfall","version":"1.0.0","variant":"default","status":"[success|partial|error]","runtime_ms":[ms],"metrics":{"contacts_processed":[n],"apollo_verified":[n],"clay_verified":[n],"total_callable":[n],"verification_rate_pct":[n]},"error":null,"session_id":"[YYYY-MM-DD]"}