Vapi AI will call the account's support line, navigate the IVR tree, and return structured intelligence: platform, AI maturity, quality score, menu depth, wait time.
Scan Results
Rock 04 · Pipeline
Scored. Mystery Shopped. Ready to work.
50 accounts heat-ranked by signal intensity. 40 CX leaders mapped with ICP scores. The Mystery Shopper runs before any human picks up the phone — so Touch 1 opens with what we actually found on their call.
The differentiator
Mystery Shopper · Powered by Vapi
We call their support line before your SDR does.
A Vapi AI agent dials each account's customer support number, navigates the full IVR tree, and returns structured intelligence: platform fingerprint, AI maturity score, IVR depth, wait times, callback availability.
Touch 1 becomes: "We called your support line on [date]. Menu depth 4. No AI at the front door. Here's what we found — and what United Airlines did about it."
How it works — 4 steps, fully automated
1
Phone Discovery
Web crawler finds the 1-800 support number from the account's contact page. 55 numbers pre-loaded. Others auto-discovered.
2
Vapi Outbound Call
AI agent dials, navigates every IVR menu level, presses through to general support. If a human answers: "Sorry, wrong number." Hangs up.
3
Transcript Analysis
LLM analyzes the call transcript. Extracts platform fingerprint, AI detection, menu depth, quality score 1–10.
4
Signal + Sequence Update
Results feed the heat score. Outreach sequence updated with specific call findings as the Touch 1 hook.
Sample output — Ally Financial (scored)
platform
Genesys
quality_score
3 / 10
menu_depth
4 levels
ai_detected
No
wait_time
6 min
callback
Offered
Pipeline at a glance
50
ICP accounts scored and heat-ranked
40
CX leaders mapped with ICP scores
10
Outreach sequences live — 3-touch
8
Signal sources running nightly
Top accounts — heat ranked · click any row for full signal detail
10-K attrition + Genesys job posting + new VP CX hire = auto-escalate to critical. Logic in signals.ts.
Live
$0 · in-house
Sourcing methodology — how every number was produced
01
Account identification — 50 accounts from 180+ candidates
ICP-first. Every account traces back to a confirmed signal, not a list purchase.
TheirStack
Queried 174M job postings for "Genesys", "Avaya", "NICE inContact", "Five9" in IT/Operations/Contact Center roles posted within 12 months. Company headcount filter: 5,000+. Returned ~180 companies across all verticals.
SEC EDGAR
Pulled most recent 10-K for every public company on the list. Scanned Risk Factors and MD&A for: "agent attrition", "customer service efficiency", "contact center transformation", and budget callouts with dollar amounts. 2+ hits = escalated priority.
Grok
Real-time X/Twitter and Reddit scan per company. Looking for: VP CX hire announcements (new buyer = evaluation window), IVR complaint spikes ("transferred 3 times", "can't reach a human"), exec interviews mentioning AI or contact center transformation in the last 90 days.
Curation
Filtered to four ICP verticals: Telecom, FSI, Airlines, Healthcare. Removed existing Cresta customers (United, Optimum — kept as case study refs). Merged subsidiaries. Deduped. Final list: 50 accounts.
02
Heat score — computed, not assigned
Every score traces back to a signal. No gut feel in the model.
One human checkpoint: approving the account list. Everything after that is automated.
LinkedIn
Title targeting in priority order: CCO / Chief Customer Officer (economic buyer) → VP Customer Experience (champion) → SVP Customer Service (operator) → CTO/VP Eng (if stack decision is technical). LinkedIn search per account, extract profile URL.
Vayne.io
Profile URL → structured data: current title, tenure, past companies, education, recent post activity. Feeds the sequence personalization layer and warm intro path mapping.
FullEnrich
LinkedIn URL → work email (verified) + mobile where available. Hit rate ~70% on VP+ titles at public companies. Falls back to People Data Labs for misses.
PDL
Tenure data — months in current role. Contacts <12 months in role are flagged as higher priority: new leaders evaluate vendors. Also used for headcount and title history.
Swarm
Maps work history, education, and professional networks against Cresta's team. Output: warm intro paths ranked by connection strength. "Cresta AE worked at Comcast 2019–2021 → direct LinkedIn connection to the SVP CX." Contacts sorted into warm / lukewarm / cold.
Outreach sequences — Ping Wu voice
Rick Germano
SVP Customer Experience · Comcast
Decision Maker
Day 0 · Mystery Shopper Hook
Hi Rick, I'm Ping Wu, CEO of Cresta. We're helping contact centers like Comcast transform customer experience with AI. United Airlines cut AHT by 15% and wait times by 15% using our platform, with 97% agent satisfaction. I'd love to discuss how we can support Comcast's goals, especially with your Genesys setup.
Day 5 · ROI Math
Hi Rick, just following up. Cresta's AI agents replace 40–60% of inbound volume, dropping costs from $8–15 per human interaction to $0.08–0.30 per AI interaction. Can we set up a quick call to explore this for Comcast?
Day 12 · Social Proof + Exit
Hi Rick, I'll stop reaching out after this. If you're curious how Optimum and others leverage Cresta — or our Forrester Leader status — let me know.
Allison Ausband
EVP Customer Experience · Delta Air Lines
Decision Maker
Day 0 · Peer Reference
Hi Allison, I'm Ping Wu, CEO of Cresta. United Airlines deployed Cresta and saw 15% lower handle time, 15% lower wait times — and 97% of agents said they'd be disappointed if it was removed. I'd love to share what they did and explore a conversation with Delta.
Day 5 · Agent Experience Angle
Hi Allison, the number I keep coming back to is 97%. That's the share of United's agents who said they'd miss Cresta if it went away. AI that agents love performing better — the sequence matters.
Day 12 · Exit
Last note — happy to share what a Cresta deployment looks like at an airline specifically, including the AHT math at United's scale. Let me know if there's any interest.