Callido and Dolpheen WhatsApp Ads with Chatbot
Callindo & Dolpheen Success Story

Transforming Customer Interaction with AI

By leveraging Callindo’s innovative AI-driven solutions, integrating Woztell, and utilizing OpenAI’s ChatGPT, Dolpheen optimized their lead generation and pre-sales processes, focusing on high-value customers and setting new benchmarks for efficiency and customer service excellence

Callido and Dolpheen WhatsApp Ads with Chatbot

Partner: Callindo
Company: Dolpheen
Industry: Fin-tech
Solution: Artificial Intelligence for WhatsApp
Platform: WhatsApp

When Dolpheen, a leader in debt mediation in Indonesia, sought to enhance their lead generation, qualification, and pre-sales processes, they turned to Callindo. The collaboration leveraged advanced AI and automation to transform Dolpheen’s operations. The result was a remarkable 90:10 efficiency ratio of bot vs. human interactions, far exceeding all expectations.

Dolpheen faced several operational challenges, including optimizing lead generation, supporting scalable growth, and managing customer interactions more effectively. Callindo was approached to provide a comprehensive solution to streamline these processes and drive future growth.

Client background and challenges

Dolpheen, well-known for its innovative debt mediation services, needed to overcome specific hurdles:

Operational Optimization: Streamlining lead generation and qualification processes.
Scalable Growth: Implementing a system capable of supporting rapid scaling.
Enhanced Customer Interaction: Managing pre-sales interactions through effective automation.

“At the beginning, we knew we needed a straightforward approach to solve Dolpheen’s main pain points,” Martin Sebela, Founder of Callindo, said. “Our goal was to set up a working framework to support their future growth.”

Challenges Identified

During the initial A/B testing phase, several critical insights emerged:

  • Behavioral Differences: Easy access to start communication via WhatsApp motivated more people to try it. This led to the realization that the entire targeting campaign needed tighter alignment with the eligible audience. We also tested an adjusted flow connected to the original process as a quick fix. This allowed us to compare human versus chatbot interactions directly.
  • Manual Testing of the Bot: Fine-tuning or A/B testing different scenarios wasn’t possible for the initial setup. We mapped and simulated typical user flows, which was challenging as it wasn’t easy to play a potential
    customer accurately. This testing phase took longer than expected, but we wanted to ensure it would work perfectly from the first run.
  • Language Challenge: Training the chatbot in Bahasa Indonesia was essential. The detailed instructions and comprehensive knowledge base led to outstanding results, with the vast majority of end clients unable to distinguish that they were interacting with a chatbot.

Rapid changes in the AI world

During the implementation, OpenAI introduced a new version of ChatGPT, version 4.0. This upgrade provided better performance at a lower cost and introduced new functionalities. One of the most impactful features was the chatbot’s ability to analyze attached pictures. Dolpheen’s flow used this capability to recognize details about clients’ debts, offering a robust and straightforward solution.

I was amazed by how advanced the functionalities had become. Training our agents to utilize these new features was surprisingly quick, but we had to do it one-on-one. It was an unexpected yet welcome challenge, said the Operational Manager from Dolpheen.

Callindo’s Innovative Solutions

Callindo provided a suite of solutions, integrating Woztell for automation and leveraging OpenAI’s ChatGPT as the assistant:

  • Automated Lead Generation and Qualification: An AI-driven chatbot powered by OpenAI and integrated with WOZTELL handled initial customer interactions, gathered information, and analyzed client needs.
  • Pre-Sales Automation: The chatbot guided potential customers through informative phases and needs analysis, reducing the workload on human agents.
  • Focus on High-Value Prospects: The process was streamlined to ensure that human agents focused only on the top 10% of pre-selected, high-value customers.

“The onboarding and training of new agents is often very painful and
expensive. ChatGPT bots offer incredible scalability—you train one, and then you can deploy as many as you need without additional costs,” said Michal Wasserbauer, co-founder of Callindo.

“We had considered adding WhatsApp to our campaigns earlier, but issues with response times, office hours, and agent availability made it tricky. With the AI chatbot, we no longer have these concerns. The chatbot can respond 24/7 with consistent quality and enthusiasm,” said Randi Hidayat, Head of Digital Marketing at Dolpheen Indonesia.

Project Scope Details

The collaboration between Callindo and Dolpheen was marked by rapid discovery and development, employing a test-and-fail approach to identify and address challenges swiftly. Close cooperation with the client was essential, ensuring that every iteration brought us closer to a solution that met Dolpheen’s unique needs.

Initial Campaign Flow

In the initial campaign flow, we aimed to compare the current webform registration
process with an immediate WhatsApp communication approach. The chatbot flow
performed well, but significant gaps were identified in the Facebook campaigns
themselves. Due to CTR optimization, the target audience changed, leading to a
higher quantity of leads but with poor quality and response rates. To genuinely test
the quality of the ChatGPT chatbot, we decided to adjust our plans.

Callindo Dolpheen ads Planned PoC Flow

Adjusted Campaign

Heureka! Within two days, we completely overhauled the flow. The adjusted campaign flow retained the initial web form registrations but replaced human contact with the ChatGPT chatbot. Despite numerous unplanned disadvantages, we successfully achieved the goal of maintaining the same sales volume by focusing on just 10% of high-quality, pre-heated leads. There were even instances where the bot completely sold the service. The chatbot managed initial inquiries and lead qualification, effectively reducing the workload on human agents and only escalating complex or high-value cases.

Summary of Results

The implementation of the adjusted campaign flow yielded impressive results. The AI-driven chatbot managed 90% of initial customer interactions, significantly freeing up human agents to focus on the top 10% of pre-qualified, high-value customers. This automation improved operational efficiency and enhanced customer satisfaction through quicker and more consistent responses. The integration of image recognition capabilities further streamlined processes by allowing the chatbot to analyze and extract information from attachments, providing a comprehensive solution for client debt management.

Facebook Campaign

The campaign delivered substantial engagement metrics:

Facebook Impressions


Facebook Clicks




WhatsApp Valid Numbers


The click-through rate (CTR) of 1.64% is average, but the high conversion rate of 26.80% indicates that many registrants lacked complete knowledge of the service. A key finding is that 90% of active WhatsApp users in Indonesia were successfully reached, underscoring the significant potential for ChatGPT WhatsApp bots in this market.



Conversion Rate


Coverage of WA


Sales funnel

There remains a 50% gap where leads do not respond, which can be reduced through instant messaging and personalized introductions. Despite this, the funnel performance is impressive: 12% of chats reach the selling stage, with over 6% pre-negotiated by the bot and 1.71% of services sold directly by the bot. This aligns perfectly with our 90% to 10% ratio goal. Human agents receive qualified, pre-heated leads ready to accept the service, maximizing efficiency and effectiveness.

Facebook Campaign Ads funnel performance

Implementation Process

The implementation was meticulously planned and executed in phases:

  1. Assessment and Customization: A detailed analysis of Dolpheen’s processes and requirements was conducted.
  2. Chatbot Development: A sophisticated chatbot using OpenAI’s ChatGPT was developed and integrated.
  3. Training and Integration: Comprehensive training was provided for the remaining human agents on handling high-value customer interactions.
  4. Continuous Optimization: Regular monitoring and iterative improvements were made to ensure peak performance.

“The Woztell platform was incredibly easy to use,” said Abdul F. Tirtayasa, Data
Scientist at Dolpheen. “It allowed us to integrate advanced AI solutions seamlessly,
enhancing our operational capabilities.”

Results and Benefits

The collaboration yielded outstanding results:

  • Increased Efficiency: Automated lead generation and qualification handled 90% of initial interactions, freeing human agents to focus on the top 10% of high-value customers.
  • Enhanced Customer Experience: The chatbot provided consistent, accurate, and quick responses, improving overall customer satisfaction.
  • Operational Cost Reduction: A significant reduction in the need for human agents led to cost savings while maintaining high-quality customer interactions.
  • Language Mastery: The chatbot’s proficiency in Bahasa Indonesia was top-notch, with most users not realizing they were interacting with an AI.
  • WhatsApp Dominance: Confirmed the dominance of WhatsApp in Indonesia, with more than 90% of users registering their phone numbers with the app, making it a crucial channel for customer interactions.


The partnership between Dolpheen and Callindo transformed operational challenges into triumphs. By leveraging Callindo’s innovative AI-driven solutions, integrating WOZTELL, and utilizing OpenAI’s ChatGPT, Dolpheen optimized its lead generation and pre-sales processes, focusing on high-value customers and setting new benchmarks for efficiency and customer service excellence. Dolpheen and Callindo are set to continue this journey of success, driving growth and setting new industry standards. And just in case, the chatbot, which currently communicates fluently in Bahasa Indonesia, can easily switch to Vietnamese, Chinese, or Japanese by changing a few words in the definition—proving that the language barrier is no longer an issue.

“At the beginning, we focused on a straightforward approach to solve the main pains and set up a framework for future growth,” Martin Sebela reiterated. “This project has shown how effective strategic implementation can be.”