When companies engage call centre agents to sell products – insurance, financial services, pay-per-view entertainment – they want to communicate the highest value possible to customers at the least cost, while complying with regulatory requirements.

A bank could, for example, engage 50 agents a month who make thousands of calls, but only a sample of these calls are analysed for quality assurance. The analysis is often manual and, for long calls, could be tedious.

Voyc, a South African startup, says it has a solution that applies machine learning to analyze all calls, checking for key phrases and highlighting insights from agent-customer conversations.

Fairness for customers and companies

“We want to ensure that companies and their customers are treated fairly, by getting through all the calls and making sure nothing slips through the cracks,” Matthew Westaway, Voyc’s CEO, says to TechCabal.

Conceived in 2018, Westaway co-founded the startup with Lethabo Motsoaledi, Voyc’s CTO, while they were engineering students at the University of Cape Town. Inspiration came while both worked on an initial market research endeavour that required them to deliver insights to companies based on audio interviews with customers.

Insurance is their biggest client segment due to the regulatory pressure on companies to treat customers well, and the risk companies are exposed to from prospective/active customers who may sometimes not be truthful about pre-existing conditions or claims.

But Westaway believes his service applies beyond insurance as there are quality assurance needs wherever call centres are used for telemarketing.

Data-driven emotional intelligence

Audio transcription is not rocket science for many tech savvy people in 2020. There are a number of apps that are able to capture text from voice – though the range of functionalities remains possible for mostly European languages, and success varies. 

The broad family of voice-to-text platforms, which includes apps like Google Translate, operate based on Natural Language Processing (NLP) principles of artificial intelligence. The same architecture underlies Voyc’s promise to “achieve expert insights infinitely faster than the manual qualitative research process.” 

With 100% of calls analyzed, companies have the evidence-based resources to improve the quality of their call agents and telemarketers, Westaway says. 

Firms identify keywords and topics that should accompany certain calls and run quality checks based on the frequency of use by agents, as well as tone of voice. This is called sentiment analysis in NLP parlance and is an increasingly popular analytical tool among digital marketers. 

Voyc, which is a web-based platform, scores each agent’s calls and where a company chooses to integrate scripts, scores are partly based on conformity to the scripts. You could, for example, search a call’s transcript to find out how an agent responded when a customer asked to speak to the manager.

In essence, call centre and customer service personnel are trained through a data-driven process to become more emotionally intelligent, or to more firmly stick to the company line.

Voyc’s ‘feedback tech’ market

Westaway and Motsoaledi initially tried to pitch Voyc as a product for improving market research processes by providing data on audio interviews. But the low frequency of interviews even at market research agencies meant scale would be limited.

They pivoted to the call centre market after participating in a Techstars programme in Germany. After a tour of accelerator programmes last year including Google’s ‘London Immersion’ (where PiggyVest, 54Gene, and PayGo Energy also featured), Voyc has doubled down on the insurance market as the sector most likely to need its services at scale.

Since beginning full operations in 2018, all of Voyc’s current clients have been in South Africa where there are about 500 companies in the sector and the insurance penetration rate was 16.99% (as of 2017). 

One of such clients is a company that sells travel insurance to four South African banks, among them Standard Bank, Africa’s largest lender by assets.

But a 500-company addressable market in one of Africa’s top-three largest economies is hardly large. When you account for the fact that small companies may not find it operationally necessary or financially prudent to outsource quality assurance, the market does appear small.

Ahead of its time?

Hence, Voyc is looking beyond Africa for other clients and is currently exploring the UK market, after which they hope to enter the US. Their Africa strategy is to pitch their service to telcos and they have begun trials with at least one customer experience company, Westaway says.

But there is a labour cost reality to Westaway’s focus away from Africa: it is relatively cheaper to hire people to listen to and review calls for quality assurance in Africa than in the UK or US. It would be difficult to be price competitive in Africa, he says.

Two years in, Voyc says they have analyzed 1.4 million minutes of calls for clients. Though founded by Africans and piloted in Africa, it might just be a product a little ahead of its time. But there is value in there and if the encouraging post-COVID fortunes of Nigeria’s insurance market is a signal, Westaway and co may find a breakthrough here while prospecting for greener landscapes. 

Alexander Onukwue Author

Get the best African tech newsletters in your inbox