AI is on the shopping list for retailers

There is no shortage of reports saying artificial intelligence (AI) and machine learning will transform how we shop. Researchers at Juniper Research forecast retailers will pump $12 billion into AI by 2023, compared to $3.6 billion today. But, how will we get to this vision of AI powering retailers?

AI provides retailers with a variety of benefits

AI could help retailers be more competitive and to do so by cutting costs while increasing the quality of experience and engagement with customers. All kinds of processes will be affected including jobs – though AI will end dull jobs of little value to the customer or the employee.

Within physical stores, AI could remove what makes the experience mundane for example how smart checkouts use computer vision to enable “just walk out” purchasing and eliminate queuing.

Furthermore, taking away tills frees up staff from undertaking low value tasks for high value customer-facing activities like personal styling. These higher value activities can then be augmented by AI to benefit retailers further. For example, artificial intelligence can take a massive catalogue of products and filter them down into a manageable amount of recommendations, which the personal stylist can then present to the customer. The combination of human and artificial intelligence – people skills from human beings with data-driven insight provided by AI – provides a better customer experience than either one used alone, helping the retailer drive sales more effectively.

AI is already making waves in e-commerce

Online commerce is clearly where AI is having a huge effect already. For example, how machine learning sorts out product catalogues for online retailers. For decades, category managers and developers have been trying to create the perfect – high quality, relevant – product catalogue. With advances in machine learning, these day-to-day tasks could be automated.

There is a huge potential for intelligent algorithms to make sense of an increasing amount of data collected as part of digital commerce processes like search, fraud detection or churn prediction.

Demand forecasting is where AI will be playing a transformative role in the background. Understanding consumer demand patterns, predicting supply chain pressures and proactively making changes are highly complex tasks that need to be done in real time. This carves out an obvious role for AI that can learn and meet customer demands faster. Unsurprisingly, the number of retailers relying on AI-powered demand forecasting is expected to triple over the next four years.

Where do chatbots come into play?

However, no matter how intelligent AI becomes they can’t be useful without the ability to take actions on behalf of customers. A good example of this is chatbots, about which big promises were made, but with a failure to deliver in most cases.

The reasons for this include how the industry hasn’t fully comprehended the amount of training or data needed to make chatbots truly useful. Too many of today’s chatbots lack the ability to change an order address, buy on behalf of a customer or do other things that a customer could do themselves or would expect a human agent to do.

This lack of usefulness isn’t a failing of AI system but the systems they sit on top of. These systems were built for an old retail model, with two primary touchpoints: storefront and catalogue. E-commerce came along in the 1990s and the catalogue was largely replaced with the desktop webstore. The software that supported both online and physical retail was logically monolithic in nature. E-commerce was treated as just another store in many instances and it was fed the data it needed.

Most retailers still sit on these monoliths. As needs have changed, software vendors added layers on top of the monolith to expose APIs for use by external applications. This was meant to modify processes to be more agile but was never intended to fully support the high-speeds and scalability required by chatbots.

How can retailers make the most of investment into AI?

Because the API was an afterthought, it was never meant to be a truly high-speed entity. Typically, any large load on an API causes the internal processing of the monolith to slow or fail due to the unexpected processing burden.

Thankfully, modern cloud native and API-first commerce platforms are designed to scale, to have one system about product or customer of truth rather than siloed data and to meet the demands of any customer interaction, be that a chatbot, a VR experience or an in-store mPOS transaction.

The growth of AI in retail doesn’t need to end in disillusionment. Its success will depend on how retailers have migrated from legacy to new modern commerce platforms that can accommodate and maximise new technologies like AI and omnichannel strategies seamlessly.

Website | + posts

James Dye is responsible for commercetools' business activities in the UK. James came to commerctools from Red Ant, where he was responsible for devising innovative omnichannel solutions and introducing customers to their digital store platform. Previously, James ran a retail innovation agency, winning several awards, including the Retail Week Customer Innovation of the year for the Argos Christmas Gift Finder.

AI Readiness - Harnessing the Power of Data and AI

Newsletter

Related articles

A Practical Guide to the EU AI Act

Disclaimer: This article is opinion-based; please seek legal advice...

Building a Smart City

If you ask me how I picture the future,...

Mastering Hypervisors for Enhanced Business Efficiency

The cloud computing landscape is a complex ecosystem characterised...

Cloud Computing’s Role in Digital Transformation

Definition of Digital Transformation Digital transformation refers to the process...

The hidden costs of technical debt inaction

With technology moving at a rapid pace, you would...