Outbound sales have always been a game of numbers: calls made, connections reached, deals closed.
However, in 2025, the game has fundamentally changed. The rise of AI voice agents means that scaling from a few hundred calls to tens of thousands per week is no longer a human resource problem. Instead, it’s become a data, technology, and process challenge.
We decided to put Retell AI to the test.
Our goal wasn’t just to prove that AI could make cold calls. We wanted to see how far we could scale, identify what bottlenecks would emerge, and discover how pairing calls with WhatsApp automation could multiply our results.
The experiment unfolded in four distinct phases. Here’s what we discovered along the way.
Phase 1: Laying the Groundwork with 500 Calls per Week
We started conservatively with 500 AI-driven cold calls a week.
This served as our “sandbox” phase, designed to validate the tech stack, refine our scripts, and test our lead data before committing to serious volume. We knew that getting the fundamentals right at this stage would be crucial for successful scaling later.
Our initial setup included:
- Retell AI voice agents with conversational AI specifically tuned for B2B cold outreach
- Simple scripts featuring clear qualification questions
- Generic business phone numbers to avoid overcomplication early on
- Basic CRM integration for call logs and MQL tagging
The results were encouraging
- Approximately 1% of calls converted to Marketing Qualified Leads (MQLs)
- Average call duration was 42 seconds before either qualification or hang-up
- AI successfully handled basic objections like “send me an email” or “I’m not interested” without sounding robotic
What we learned in Phase 1
Voice quality proved to be critical. AI that sounds natural dramatically reduces hang-up rates. We also discovered that keeping scripts short was essential. Longer introductions led to early disconnects, and the sweet spot was getting to the qualifying question within 10 seconds.
Additionally, voicemail capability became a must-have feature. About 30% of calls went to voicemail, and our AI successfully left short, friendly messages with a callback number.
By the end of Phase 1, we had validated that Retell AI could work effectively at a small scale. However, the biggest bottleneck was already becoming obvious: lead data quality.
Phase 2: Confronting the Data Dilemma
As we quickly learned, cold calling lives or dies by the quality of the contact list.
Our first lists came from Wappalyzer and similar data providers. While these were inexpensive, they presented two significant challenges. Many numbers turned out to be administrative lines rather than direct dials. Furthermore, while the mobile numbers we did reach converted better, they were much harder to obtain.
To understand this better, we ran two side-by-side tests:
List A: High-volume, low-cost business numbers
List B: Mobile-verified decision-maker numbers, approximately 3x cost per lead
Our findings were revealing:
- List A delivered a lower connection rate, requiring more calls per MQL
- List B achieved higher conversion but offered smaller reach without significant budget increases
The key lesson became clear: there’s a critical trade-off between cost-per-lead and cost-per-MQL. Sometimes cheaper data isn’t actually cheaper once you factor in wasted dials.
Operational adjustments we made in Phase 2:
- Started tagging each lead source in our CRM for better performance tracking
- Created call disposition codes for more detailed post-call analysis
- Introduced call attempt logic where numbers received a maximum of 3 AI attempts before being retired
Phase 2 crystallized one important truth: scaling AI cold calls isn’t just about adding more agents. It’s about feeding them the right data so every minute of AI talk time delivers maximum value.
Phase 3: Scaling to 10,000-15,000 Calls per Week
With our process validated and a deeper understanding of our data sources, we were ready to accelerate.
We scaled our operation to include:
- 20 active AI agents
- 1,500 calls per day
- 16 new UK phone numbers for outbound calling to improve pickup rates and avoid number fatigue
Why multiple numbers mattered: Carriers and spam filters flag numbers when they’re overused. By rotating across 15 to 20 numbers per market, we kept our caller ID reputation clean and maintained higher connection rates.
Our cost structure at scale:
- $100 to $120 per 1,000 calls
- AI agent capacity of approximately 80 calls per hour, fully automated
- Infrastructure costs remained flat while output multiplied significantly
Scaling brought new challenges:
Data freshness became critical. At high volume, lists get burned faster, requiring a steady pipeline of fresh numbers. Time zone management also became important, as AI agents needed scheduling to call during appropriate local hours.
We also refined our call routing logic so that if AI detected a human answering mid-ring, it could adapt its introduction to sound more natural.
By the end of Phase 3, we had built an outbound engine capable of running at sustained high volume. However, conversion rates began to plateau, and it became clear that calls alone weren’t enough to maximize ROI, a challenge familiar to any cold call service aiming to scale outreach effectively.
Phase 4: The WhatsApp Follow-up Breakthrough
This phase delivered the breakthrough we needed.
We integrated TimelinesAI with RetellAI, enabling automatic WhatsApp follow-ups after specific call outcomes:
- Missed call/no answer: Send a short message with our value proposition plus call-to-action link
- Voicemail left: Follow with a WhatsApp note referencing the voicemail
- Mild interest expressed: Send personalized information plus booking link immediately
Why WhatsApp proved so effective: Many people ignore calls from unknown numbers but read WhatsApp messages within minutes. Additionally, messaging allows for richer follow-up content including links, PDFs, and videos without overwhelming the prospect during the initial phone conversation.
The impact was substantial:
- Engagement rates increased by 35 to 40% compared to calls alone
- Prospects started replying to WhatsApp messages even if they had ignored the initial call
- Our sales team could continue conversations asynchronously without scheduling friction
At this point, our process evolved from a simple “AI calling campaign” to a comprehensive multi-channel outbound engine where voice opened the door and WhatsApp built the relationship.
WhatsApp as the Conversion Multiplier
The real value wasn’t in replacing calls but rather in layering messaging on top of them.
We shifted our thinking from viewing RetellAI as a standalone tool to making it the first touchpoint in a strategic chain:
- Call: AI reaches out with a natural conversation opener
- WhatsApp: TimelinesAI sends follow-up content instantly if no live conversation occurs
- CRM sync: All touchpoints log automatically for sales team context
- Nurture: If no immediate conversion occurs, drip messages continue on WhatsApp over 1 to 2 weeks
The result was more touchpoints, more responses, and higher pipeline yield from the same lead list.
Why TimelinesAI Became Our Conversion Multiplier
Several factors made TimelinesAI particularly effective:
Seamless Transition From Voice to Text
After an AI call, TimelinesAI instantly triggers a personalized WhatsApp message. This keeps the conversation alive while the prospect still remembers the interaction.
Dramatically Higher Open and Reply Rates
WhatsApp messages get opened far more than emails or follow-up calls, often within minutes. This alone improved our engagement rate by over 3x compared to calls alone.
Shared Inbox for Faster Team Response
All inbound replies from WhatsApp land in a central shared inbox. Whether handled by a sales rep or another AI workflow, every lead receives timely attention without relying on individual phone numbers.
CRM Integration That Actually Works
By connecting TimelinesAI directly with our CRM, follow-up sequences are automated and logged without manual input. This ensures that every AI-initiated conversation turns into a structured sales opportunity.
In essence, RetellAI opened the door, and TimelinesAI made sure the prospect walked through it. The combination allowed us to move from “hit-and-miss” cold calls to a predictable outreach system that works even when humans aren’t directly involved.
If you’re using RetellAI for cold calling and relying solely on voice to convert leads, you’re leaving money on the table. By combining AI-driven calls with instant WhatsApp follow-ups powered by TimelinesAI, we transformed a good outreach process into an exceptional one.
This isn’t about replacing your sales team. It’s about supercharging them with the right sequence and the right tools. RetellAI gets you in the door. TimelinesAI keeps you there until the deal closes.
Ready to transform your AI calls into real sales conversations? Sign up for TimelinesAI today and seamlessly connect it to your RetellAI and CRM setup. Because every cold call deserves a warm follow-up.
