Since the early 2000s, artificial intelligence (AI) technology has advanced at break-neck speed and many South African retailers have since realised its huge potential.
Globally, AI has been used to analyse customer purchasing behaviour to enable them to fine-hone their offerings to their target market. AI can serve retailers through automating most processes and, in some tasks, replace human labour altogether. With AI, retailers can predict demand and thus, increase and streamline overall productivity and generate more revenue while enhancing security.
Aside from banking apps that use facial recognition in many of their transactions, eg Discovery Bank, there are facial recognition tools that also analyse body language, giving insights to consumer preferences. According to an article on BizCommunity: ‘In-store cameras detect shoppers’ facial expressions, which can reveal if they think a product is too expensive or if they are unimpressed with its specifications.’
Then there is the area where Virtual Reality — once the domain of Gamers — bridges the online and offline divide. Offering customers immersive, in-store experiences that showcase products and services in innovative ways, can be key to attracting and retaining customers.
Uptake deploying AI within the retail sector in South Africa, however, poses its own set of challenges.
To make accurate calculations of customers’ purchasing patterns, preferences, and similar consumer habits, AI needs access to immense amounts of ‘clean’ data and is therefore dependent on high levels of connectivity among consumers. To glean accurate insights, machine learning (ML) software requires unbiased, consistent information.
In South Africa — where more than 30% of the population is not connected via the internet — this is not possible; thus, AI cannot perform optimally owing to a lack of data.
Even though our population diversity is commendable, comprising a variety of social, ethnic and racial groups, it creates a challenge when it comes to creating informed and accurate customer profiles and the problem is, any incomplete or duplicated information can compromise the value of the data, possibly resulting in decisions being made based on inappropriate/inaccurate information.
In addition, not having sufficient or accurate data can jeopardise a retailer’s chance of tapping into areas that are under-serviced.
Overlay that with our diversity of languages and then tools such as chatbots become challenging, so too, face recognition programs.
An AI tool that operates on the agreement of consumers to provide certain information may encounter a stumbling block as the more wary (some would argue, less-informed) South African may baulk at this, with it being viewed as an invasion of privacy.
Despite its potential, executing AI solutions could be an intimidating undertaking for any retail executive. ‘Why fix what isn’t broken?’ has often been the approach to AI.
A good starting point, however, would be to automate decisions that are repeated often and then ramp up AI deployment in other areas as efficacies are recorded.
There is a long road ahead for South African retailers who need to closely consider whether the deployment of AI will offer value to its business and how best to implement it…
