Taking a More Intelligent Approach to AI in E-Commerce

Despite artificial intelligence's growing potential, many retailers currently struggle to embrace it and act on the insights it delivers. Often, this comes down to them not taking a strategic approach and deploying disjointed tools that are difficult to use or scale.

New approaches are needed if AI is to truly benefit merchants and customers alike. Retailers should start by following a four-step process:

1. Classify the different types of AI application.

AI supports merchants in multiple ways, making it easy to lose focus by trying to do too much at once. Instead, look at the different uses through a business lens and classify them based on where they can benefit your company. A simple three-way split helps clarify what you should focus on:

• The first group is where AI delivers internal productivity improvements, such as automating manual processes, freeing up your team to work on higher-value activities.
• Next there’s customer experience, covering front-end improvements that enhance the journey for shoppers, such as through AI-powered search.
• Finally, keep an eye on moonshot projects that aren’t ready for primetime yet, but could deliver big benefits in the future. For example, projects such as rapid prototyping enable you to test customer perceptions around new products by accelerating the creation of accurate digital prototypes using image-based AI.

2. Understand where AI is a good fit for your brand.

In the rush to embrace AI it's vital not to lose sight of your brand values and what differentiates your brand in the market. That means reviewing whether specific AI applications and models really are a good fit for your business.

For example, if you're aiming to foster a high-end luxury or authentic feel, then generative AI-created product images may not be right for you at this point. Instead, you could potentially look at more sophisticated uses such as virtual try-on applications. Also, check that any AI models you rely on are trained on the right data for your specific sector, rather than simply relying on generic information that doesn’t reflect your customer base.

3. Use different types of AI in specific areas.

In addition to segmenting AI by how it's used, look at how it can help you across the business and customer journey in an integrated way.

Different types of AI deliver different benefits — e.g., predictive intelligence helps analyze customer behavior, spot patterns, and predict future behavior and preferences, while semantic intelligence transforms your understanding of customer queries and intent and can be used to drive improved on-site product and content experiences.

On the automation side, you can improve efficiency through visual AI that turns images into machine-readable datasets. This enables them to be automatically organized and categorized, providing proactive recommendations to help improve the performance of visual assets, including user-generated content.

Generative AI augments your capabilities, streamlining processes, creating efficiencies, and improving performance. It’s likely to make a big difference in e-commerce site search, for example, by helping shoppers refine their queries by conversing with an AI chatbot to zero in on the right product. AI-based dynamic targeting can help merchants deliver a personalized experience to each and every customer across the web and email channels.

4. Consider the ethical dimension.

We’ve all seen the potential ethical problems that AI use can bring, from inadvertent bias to issues caused by poor training data around privacy and model transparency. Carry out a full check of any AI applications before you deploy them to ensure that you're satisfied that they won’t negatively impact your brand’s reputation — or revenues. One auto dealer was caught when its new generative AI chatbot was programmed by a user to sell him a new vehicle for $1, for example.

We’re at the start of the AI journey, making it vital that merchants take a step back and adopt a strategic approach to the technology. They should identify where it can best help them and their customers now while planning for the future as different applications mature.

Jan Soerensen is general manager, North America at Nosto, an e-commerce personalization platform.