How AI is revolutionising retail

Since ChatGPT burst on the scene, the hype about AI has gone off the scale.

And, yes, while we are just at the beginning of an interesting and exciting journey that will force us to re-examine what it means to be cognitive beings, today's relatively primitive AI is already disrupting many industries by offering highly-practical interventions that are driving profound change.

Retail is one of those industries. In 2018, McKinsey estimated the impact of AI on the retail sector was $470.2 billion. Interestingly, when looking at marketing and sales across all sectors, the range was $1.4 trillion to $2.6 trillion − a figure that is impossible even to conceptualise in rands.

These figures, which we could expect to be much higher today, were based on mapping AI and analytics to business problems they could solve.

Here in South Africa, we are somewhat behind the curve when it comes to the use of AI in retail, but we are already seeing how AI is turning purchasing from a linear process, in which the main influencers are the consumer's immediate circle and environment, to a much more dynamic one, in which accurate recommendations change the game.

AI is influential in three key areas:

The move online is making much more information available to marketers from a wide variety of sources − although Google, Meta and Amazon seem to be emerging as the effective consolidators (and thus brokers) of data. These mega-platforms are now selling space on specific pages on their sites based on analysis of the individual's actual behaviour.

Discuss a planned trip to Italy on Facebook or search for Mediterranean cruises on Google and you will soon notice ads for related products and services appearing on your various feeds. These targeted ads can be useful to consumers by opening up new ideas.

The key to successful AI, clearly, is data − the more of it, the better.

The cliché about marketing − we know it's important but not which bits − is growing outdated. Increasingly, marketers understand which bits actually did work, and can quantify that success, making it easier to motivate budgets from perennially sceptical CFOs.

The increasingly granular data available about consumers, even down to the individual level, are making the recommendations offered much more useful. Although much of what we see is still at the basic level of “customers who bought this product also bought”, it's now quite possible to be much more creative.

For example, to return to that planned trip to Italy, basic recommendations for accommodation and flight deals are useful and will ultimately be expanded. It is theoretically possible to track the kind of accommodation that interests that particular individual and then begin offering complementary services − curated experiences likely to appeal to somebody of a certain sensibility.

The key to successful AI, clearly, is data − the more of it, the better.

The online purchase process is notoriously porous, with an average cart abandonment rate just below 70%. The reasons for this high figure include an onerous registration process (24%), lengthy delivery time (19%) and complicated checkout procedures (18%).

AI has potential to address these issues and make the progression from recommendation to purchase much more certain. Powerful recommendations are obviously one element, but AI can also help to identify the individual's preferred shopping style: Is he or she a browser who wants to keep on looking, or is it all about the rapid, one-click type of checkout?

The whole point here is that there are many different ways of shopping and AI can help give a particular customer the experience he or she enjoys, all to drive down the rate of cart abandonment.

In all three areas, I should hasten to add, there is still lots of work to be done, but the general trajectory is fairly clear.

One complex issue stalling progress, and it's one that regulators are very focused on, is data privacy. Retailers need to find ways of getting consumers to consent to supplying their data, and the key might be framing the right incentives − there's no doubt that the use of AI in retail is a potentially good thing for consumers because it will provide them with wider, intelligently selected choices.

For retailers, and companies generally, of course, the benefits are very clear. The advantage of targeting marketing and sales efforts, and quantifying their effectiveness, is obvious and should ultimately lead to increased sales and a reduced sales cost.

To close, one should note an important strategic issue for data users on their journey towards more effective marketing and sales.

Simply put, it will be necessary to share data with other members of their value ecosystems in order to delight consumers with genuinely innovative and useful recommendations. That may yet prove to be the biggest hurdle of them all.