African retailers listen up: the ‘spray and pray’ days of advertising are over. Get on the bus

The general mood at the conference was that Africa is underestimated and poorly understood, even by firms that may want to set up on the continent

TAMPA, FLORIDA: Africa’s consumer-facing industries are expected to grow ?by more than $400 billion by 2020, but many companies, particularly those new to the continent, have little idea how to translate the opportunity into action.

Still, most of the raw material to make sense of the African market already exists, and is being created at a faster rate than ever before – data.

One of the year’s biggest digital marketing and e-commerce conferences just ended in Florida, where marketing and sales professionals were wowed by the latest in ultra-personalised customer engagement.

The vast majority of attendees at the IBM Amplify conference were from North America and Europe, possibly because these are more developed markets for digital marketing, sales and e-commerce, but the general mood at the conference was that Africa is underestimated and poorly understood, even by companies that may want to set up on the continent.

There are savvier solutions for commerce being showcased in Tampa, such as dynamic sequencing of an items list for an online retailer – the order in which the customer sees the items for sale on the website depends not only on their preferences but on the store’s inventory.

When an item is running low it falls lower in the queue, such that a customer rarely, or never, orders something that is out of stock – one of the most annoying experiences in online shopping.

The fun stuff

There are also the more fun, more frivolous features of cognitive computing on display, such as the Cognitive Cocktail Bar powered by IBM Watson. Watson – IBM’s much vaunted cognitive computing system – asks questions and analyses your answers to infer characteristics of your personality; then predicts your preferred drink. While you sip it, read your full personality profile, delivered to your inbox.

“We now know customers better than we ever have,” said Harriet Green, IBM’s general manager for the Internet of Things, commerce and education.

“Technology allows us to move far beyond the traditional ways of knowing your customer, that was focus groups and surveys. We can now record every move a customer makes, have heat maps that follow customers in stores to see exactly how they move through shop, follow unsolicited mentions online; and from that, craft the most seamless and relevant customer experience possible.”

The potential is certainly there – IBM’s cognitive computing system called “Watson” is a self-coding algorithm, which allows it to process “dark, messy, unstructured data” such as banter on social media, learn and adapt to particular contexts.

Lisa Claes, executive director of one of Australia’s largest banks, ING, and an IBM customer, put it succinctly: “The ‘spray and pray’ day of marketing is over. We work to target the right customer, with the right message, at the right time, and in the right way.”

Unlike traditional programmes where a programmer tells the software what to do, here, the machine is self-coding, and so in theory can learn and adapt to any situation presented to it.

“Cognitive computing is so fundamentally different from what we have had until now,” said Kareem Yusef, vice president of marketing and development at IBM Commerce.

“You don’t encode behaviour into the machine, you create a system that can learn behaviour, understand concepts, and – importantly – learn from mistakes and actually get better with time,” he said at the IBM Amplify conference.

African context

For the African context, this is exciting because there is no need for translation into local languages – you don’t encode translation software, you literally teach the machine a new language, and by trial and error it picks up on nuance, tone and context.

A recent study by McKinsey found that African consumers are crying out for local content to suit their distinct tastes, cultures, and languages. This is particularly the case in non-English-speaking countries like Algeria, Angola, and Senegal, where more than 40% of survey respondents said local content was the key change they wanted to the Internet. 

The very fact that the machine has a name – IBM executives nearly exclusively used the pronouns “him” and “his”, not “it”, when referring to Watson – highlights the explicitly humanised space that they want their system to embody.

But there’s a darker side to cognitive computing, not just on privacy concerns, but the almost evocative sense of the “uncanny valley” that experts in artificial intelligence and robotics that scientists are still trying to get a better understanding of.

The uncanny valley is a term used to describe the mental uneasiness that one feels when robots or virtual characters try, but just fail to mimic a realistic human.

People react positively the more human-like an artificial figure is, but only up to a point. The sense of familiarity drops sharply into the uncanny valley once the realism just misses the mark.

Scientists believe that that distress is a evolutionary mechanism for survival – we are wired to detect minor distortions that indicate disease, mental or physical problems; human-yet-not-human figures also trigger our fear of death because they remind us of a corpse, which could carry disease.

Related Content


blog comments powered by Disqus