This Holiday Season, Global technology brand HONOR, is celebrating “Unsung Heroes” with a moving holiday movie and an exciting social giveaway. These individuals, often…
Machine learning is a complicated subject. First experimented with by computer scientist Arthur Samuel back in 1956, the American eventually managed to build a program that could beat him at a game of checkers. As badass as that was, it doesn’t really compare to the advancements we’ve witnessed over the past few years. Today, tech like Google’s self-driving cars and Microsoft’s near real-time voice translation are frequently making the headlines on tech sites.
A rebranded form of artificial intelligence, these intelligent pattern recognition algorithms could revolutionise the way we work in practically every industry.
Yet there are still very few formal education institutions that cater for this ground-breaking field of science, especially in South Africa. One startup looking to change that is DataProphet, which makes profit out of a very niche and overlooked industry: call centres.
The Cape Town-based machine learning company offers call centres globally with increased sales by around 20% to 30%, among other innovative services. The profitable startup is also setting up another office in San Fransisco, where it expects to expand its fast-growing client base.
Back in 2013, when co-founder Richard Craib (pictured right) came back from doing a course in this subject at Stanford University, fellow business partners Daniel Schwartzkopff (middle) and Frans Cronje (left) quickly noticed that there was a definite gap in the local market, so they made their move.
“We decided it would be a good idea to start a machine learning company in South Africa because you couldn’t even study it here,” says Schwartzkopff, who’s also successfully exited BetVIP — the world’s first licensed bitcoin-only betting site. “Machine learning skills are rare in South Africa; we thought we’d take advantage of an otherwise untapped market”.
Consultancy before product
Before honing in on the relatively untapped call centre industry, the team acted as consultants related to the field of machine learning. This gave them exposure to multiple industries and became very important to help mould the business model that exists today — a strategy which has paid off really well.
For instance, the team found that clients are far more likely to offer a percentage of increased revenue brought about by their algorithms, as opposed to sharing a percentage of cost reduction. “We thus began to focus more exclusively on the outbound sales environment as our products can show a measurable increase in value practically overnight,” shares Schwartzkopff.
The company’s flagship product, the Agent Lead Matching Algorithm (ALMA), uses data to “intelligently match call agents with leads” where call centres are charged based on the results. If ALMA achieves between 0% and 10% uplift in sales, it’s free. An increase in sales between 10% and 20% is worth R500 per seat while anywhere between 20% and 30% is R1 000.
“We found the call centre industry to be an incredible niche where it hasn’t been approached with this kind of technology much. Most of the [traditional] processes are quite rudimentary. We found it was really easy to generate that size return for them.”
Cronje adds that the company’s also working on voice analytics that can assess whether the agent is good or bad based on the way they speak and refining interview criteria.
There’s an estimated total of 300 000 call agent seats in South Africa. DataProphet has already secured about ten clients locally and internationally, but hopes to expand its client base and raise venture capital for series A in the US.
The startup is backed by “several million rand” from a local VC called XCO Capital — a deal which is said to have been strategic in opening a lot of doors to its initial clients.
Beyond call centres
The beauty of fields like machine learning is that it can be applied to a variety of different fields, which means companies like DataProphet don’t have to limit themselves to call centres.
“Machine learning is essentially a rebrand of the more classical term artificial intelligence,” explains Schwartzkopff. “It is the study of algorithms that change the computer’s traditional role as a calculator — the algorithms make the computer capable of learning from and making predictions on data.”
In the past, the team has developed programs for lawyers to help find fraudulent emails using artificial neural networks, for example, or building algorithms to assist hedge funds.
Especially in emerging markets, there’s a lot of room for this kind of tech.
Cronje notes that there are some courses beginning to emerge in South Africa on machine learning and artificial intelligence. Though higher education is going to have to pick up the pace if it’s really to take advantage of this innovative field of science.