Thanks to recent improvements in Internet connectivity and the emergence of numerous global big-hitters, South Africa is making a name for itself in the technological space.
Founded in 2014, Bloemfontein-based performance marketing company Xineoh is making waves internationally thanks to its vectorisation algorithm, which has already generated in excess of $30-million in revenue for clients across the globe, and attracted $2-million in investor backing so far in 2017.
A pioneer in the AI and machine learning space, Xineoh’s distinctive algorithm bears similarities to those used by the likes of and Netflix, tapping into user behaviour to predict potential purchasing behaviour and serve users bespoke information likely to incite conversion.
Unlike these tried and tested models, Xineoh’s algorithm relies on available rather than inputted information, creating associations based on the way other users have interacted with the same product or service, thus enabling it to be used across a wide variety of industries and generate uncannily accurate recommendations.
The results speak for themselves: Xineoh’s prediction rate currently sits at about 14,5% – international competitors typically sit at between 3% and 8% – and a previous version of its algorithm has already been effectively used by one of the world’s top real estate portals to reduce its cost-per-lead by 60% to 80%.
“What makes our solution so special is the fact that it’s entirely scalable and easy to deploy across a wide range of sectors,” explains CEO Vian Chinner. “Ultimately, our aim is to help businesses move customers as efficiently as possible from first contact through to conversion, using artificial intelligence to tap into the real causative factors proven to drive purchasing behaviour.”
This theory will be put to the test with the upcoming launch of VideoLlama, a personal shopping assistant for video streaming.
Set to be rolled out initially in the US, the service taps into the likes of Netflix, Amazon and Hulu to find users the best possible streaming deals, and uses the Xineoh algorithm to offer calculated recommendations based on future, rather than past, behaviour.
“The major problem with streaming services today is that they tend to offer up recommendations based on how much you will like a specific movie or show,” explains Chinner. “And, while they aren’t inherently wrong, they don’t necessarily offer up real value for viewers.
“With Video Llama, we’ve tapped into our algorithm to serve up realistic recommendations based on the behaviour of others, serving the types of movies or shows you’re likely to stream based on the time of day, your company and various other latent variables.”
VideoLlama enables users to swipe left or right based on their preferences, and specify various viewing parameters – for instance, whether they’re watching with a partner or a child. The algorithm then retrieves the relevant titles, and quickly begins to further refine its offerings based on an individual’s expressed preferences.
Over the coming months, Xineoh plans to tap into a series of other verticals, with a specific focus on the property sector, which Chinner believes is rife with opportunity.
“Property is a high-investment purchase and, given the economic environment, real estate portals now need to entice users within the first 10 seconds if they have any realistic hope of sustaining interest,” explains Chinner. “Using our algorithm, these sites will be better able to serve users homes based on what they’re actually looking for, tapping into known associations to produce a refined, concise offering far more likely to result in conversion.”
Xineoh has offices in Bloemfontein, Cape Town, Oregon and Vancouver.