Homepage / Case Studies / Building African Language Voicebots to Enable Better B2C Interactions

Building African Language Voicebots to Enable Better B2C Interactions

Disclaimer: Aya Data holds to the confidentiality agreements made with our clients. We shall not disclose any confidential information regarding clients and/or projects, pursuant to the individual agreements made. Consequently, client and project names and any identifiable information may be kept anonymous in our Case Studies.

Overview

AI chatbots and voicebots have become some of the most efficient and cost-effective tools for increasing customer satisfaction. These tools can provide quick and accurate responses to customer inquiries, lowering the workload of customer support teams, reducing response times, and enabling 24-hour customer support. Consequently, many businesses have rapidly started implementing these tools.

Data Science and AI

Image segmentation

Industry

Fintech

Headquarters

London, UK

Company Size

250+

Challenge

The Client, a company that provides financial services in 7 African countries and has over 6 million customers, needed to lower the cost of running 24/7 customer support call centers in multiple host markets. Additionally, due to the large customer base, delivering efficient and timely support was challenging.

Even though the Client had previously implemented chatbots and interactive voice recognition solutions, the CS still struggled to keep up with the workload. The end result was prolonged wait times and dissatisfied customers.

The address these issues, the Client decided to develop and deploy AI voicebots in their priority market. The voicebots would enable better, more fluent, and more efficient interactions with customers and lower the CS team’s active workload.

Solution

Aya Data collaborated with industry partners to develop and deploy voicebots solutions in five local African dialects. The development and deployment involved creating speech-to-text and text-to-speech models that were capable of processing regional dialects and generating appropriate responses.

Thousands of hours of voice data in local dialects were collected and annotated. This data was used to train models to recognize a limited number of possible intents from voice commands. The intents were further linked to appropriate responses within a custom voicebot architecture.

Results

After the voicebots were deployed, the Client noted a significant improvement in customer service across all target markets. The voicebots could handle approximately 50% of customer inquiries, effectively lowering the CS team’s workload by the same amount. Response times were shortened and customers reported higher satisfaction rates.

By deploying AI voicebots in local African dialects, the Client managed to alleviate the strain on their call centers in multiple markets, reduced operational costs, improved the efficiency of customer interactions, and ultimately improved the customer experience.