Generative AI: The secret sauce for retail innovation

Insights

• Nearly 75% of retail businesses have already started their generative AI journeys, as per Infosys research.
• Generative AI is transforming this sector by enhancing customer experiences and streamlining operations.
• It drives personalization, efficiency, and innovation, setting new service standards.
• However, challenges related to delivering value, mitigating costs, ensuring data privacy, and addressing AI bias must be overcome.
• The future of this industry depends on ethical and effective implementation of technologies.

Retail is an industry that is keen to reap the benefits of generative AI, with 75% of the companies Infosys surveyed in its generative AI research already having started to explore how to use this transformative technology in areas such as personalization, efficiency, and innovation. However, the implementation of this technology is not without challenges. To harness the benefits of generative AI, the retail industry must overcome several obstacles, including delivering value through use cases, safeguarding data privacy and security, and addressing AI bias.

The retail industry expects its generative AI spending to increase by 88% this year – as per Infosys research – ranking it above sectors such as consumer packaged goods, logistics, telecommunications, life sciences, healthcare, and energy, mining and utilities. Nearly 75% of retail businesses are either experimenting with or implementing generative AI (Figure 1). This figure is mostly greater than the overall sample, yet another plus for these sectors. Also, 78% of business leaders in the US selected generative AI as the top emerging technology over the next few years.

Figure 1: Most retail companies already use generative AI

Fields of application

In the past couple of years, several key areas have emerged where generative AI is making a significant impact. They include:

• Content creation and marketing: Brands such as Valentino and Ralph Lauren are leveraging AI to produce content for e-commerce site landing pages, blogs, product descriptions, social media updates, and other collateral for marketing campaigns. The Infosys study shows that enhanced user experience and content generation are the key focus areas in generative AI for retail businesses. More of these (29%) expect generative AI to have a positive impact on content generation than the overall trend (19%). They also expect generative AI to have a positive effect on user experience.
• Trend prediction: The success of fashion brands relies on how well trends are received by consumers. Zara and H&M are using AI algorithms and predictive analytics to analyze their audiences’ desires, and predict trends that have a greater chance at being accepted. These insights feed into their design, production, and inventory. More accurate forecasting means both more sales, and sustainability gains thanks to less wastage. For example, Mango uses AI to scan social media images of their products shared by consumers to determine the popularity of fabrics and patterns. The brand then replaces products that are not faring well instead of replenishing them, and decides the quantity to be produced.
  

  Zara and H&M are using AI algorithms and predictive analytics to analyze their audiences’ desires, and predict trends that have a greater chance at being accepted.

• Customer support and product recommendations
  Retail depends on delivering excellent customer support to ensure repeat customers. Generative AI can transform customer support operations for this sector. For example, eBay’s AI virtual assistant called ShopBot allows customers to use text, voice, or image to express the item they’re looking for, and makes product recommendations accordingly. This has helped the brand increase its conversion rates compared to the use of standard search. Quick service restaurants (QSRs) have also benefited from the technology. Deep Brew, Starbucks’ proprietary AI platform, creates personalized menu recommendations and marketing messages through customer data analysis.
  Yum Brands is testing a generative AI feature on its mobile app that is used by its restaurant managers to ask questions related to work processes and get responses easily rather than pour over training material. Wendy’s is testing voice bots driven by generative AI for its drive-thru ordering, to take customers’ orders more quickly and accurately.
  
• Fraud detection
  Retailers frequently lose much of their revenue to fraud be it return fraud, credit card fraud, or employee fraud. AI algorithms can detect and highlight fraudulent activity by examining data from point-of-sale systems and access logs. This alerts retailers into initiating steps such as enhanced security measures, or employee training to increase awareness about the impact of fraud, to combat retail fraud issues. Amazon uses generative AI to detect financial fraud, and manage its finances better. QSRs can also use generative AI to analyze transaction data and identify indicators of fraud.
• Inventory management
  Generative AI offers retail businesses real-time visibility into stocks present at different stores, by analyzing sales data swiftly and effortlessly, which helps prevent understocking or overstocking. Domino's is developing a generative AI assistant that store managers can use to save time spent on routine tasks such as inventory management and ordering ingredients, and utilize their time in more challenging areas.

While the use of generative AI in retail is exciting and rewarding, businesses also face challenges related to its implementation.

Obstacles linked to generative AI

• AI is expensive to implement whether it is related to building an AI system, acquiring high-quality data for it, or setting up and maintaining the system, not to mention integrating it into existing hardware and software, and hiring AI talent or reskilling existing talent in AI to optimize usage outcomes. Financial constraints are a significant hurdle, especially for smaller businesses, and those wanting to use AI must be clear about the upfront costs and the potential for creating value and ROI. Infosys research shows that only 5% of retail businesses among those surveyed create value with their use cases, which is much lower than the overall trend (13%).
• Data privacy and security are critical issues. AI systems depend heavily on consumer data to operate efficiently, which brings up concerns about data protection and adherence to regulations. Retailers will need to implement strong security measures to safeguard sensitive information and uphold customer trust.
• Ethics and bias are as big a concern for retail as data privacy and security. The data that the AI models are trained on needs to be of the highest quality and accuracy. If they are trained on biased datasets, they are likely to produce distorted results, which can be damaging to the organization. Nearly one-third of retail is concerned with ethics, bias, and safety — much higher than the overall trend (21%), as per the Infosys study.
  
• The study also shows that half of retail companies do not think their workforce is ready for generative AI (Figure 2). While less confident than the overall trend, retail was more neutral about managing generative AI than the overall trend, which indicates a need for improvement in this area. Businesses will require multi-disciplinary teams with deep knowledge in both domains and industries to get the most of their AI implementation.

Figure 2: Retail workforce is less prepared for generative AI

Dodging the roadblocks

Businesses wanting to develop and deploy generative AI and generate value from it must have an AI readiness framework. This helps employees, customers, and users feel confident in the business, and establishes the groundwork for a successful AI-first business transformation.

• Businesses need a clear vision of their goals that aligns with business objectives, and they must gain stakeholder support. This alignment is key to delivering projects that add real business value. Organizations should measure this value through metrics like increased revenue and profitability, or greater market share.
• A robust data strategy is essential, particularly for a business such as retail that handles sensitive information. The data usage must comply with all relevant regulations and guidelines. Establishing solid data governance means businesses can address data privacy and security issues through encryption, access controls, and regular security audits, all of which are established as part of data governance. While implementing best-in-class governance frameworks can be expensive, best practice for data governance and preparation ensures that data being used for AI is consistent, reliable, and secure. It also aids companies in adhering to data privacy and security regulations, mitigating the risk of data breaches and resulting penalties. Formulating data governance includes setting up a dedicated team, processes, and tools for implementation.
• To reduce AI bias, it is crucial to train models with diverse and representative datasets. Conducting regular audits for bias, and involving teams from varied backgrounds in the development process can help achieve fair and unbiased results. Ensuring transparency in AI decision-making and maintaining continuous monitoring are essential to address potential biases and enhance the accuracy and fairness of AI systems. Gathering user feedback is a useful approach to uncovering and addressing biases. A strong ethical framework is essential, outlining the company’s ethical standards and potential dilemmas to ensure clarity for all AI users, and establishing KPIs that ensure these ethical standards are consistently upheld.

  Gathering user feedback is a useful approach to uncovering and addressing AI biases.

  
• To control expenses related to implementing AI, companies can consider scalable AI solutions that support gradual investment. Numerous AI providers offer modular systems or pay-as-you-go options that allow businesses to begin on a smaller scale and grow as their requirements and budgets permit. Furthermore, running pilot projects can help businesses evaluate the potential return on investment before committing to larger financial outlays. While the use cases for AI are many, companies need to prioritize use cases and using AI only for relevant use cases that deliver the most value. This can help with reducing costs.
  
  Partnering with technology providers who offer integration support can also help streamline the process and reduce the complexity of implementation. Studies indicate that retailers are adopting generative AI at different rates, with 21% planning to use the technology via solutions from software vendors. Government subsidies to promote technological development are also an option that can help reduce initial costs significantly.
• Even the most sophisticated technology can go untapped if the employees using it are not adept at using it and are unaware of its value, that is, its benefits to the business and their own roles in making it all happen. Investment in training and upskilling programs can help companies bring their employees up to speed in AI skills. With AI changing the roles humans play at work significantly, it’s important for businesses to train them in techo-functional skills that brings together technical expertise with business knowledge. Companies can also outsource know-how from consulting firms to fulfil their talent needs.

Retail is ahead of most other industries in its generative AI implementation, and beginning to enjoy the rewards. As the technology advances, and this sector gets the hang of resolving the challenges related to it, it will increasingly benefit from it. A thoughtful and strategic approach is critical—one that balances technological capabilities with ethical considerations.

Note: The data in this article comes from the Infosys Knowledge Institute’s Generative AI Radar survey. The sample of Retail respondents also includes a small number of hospitality businesses with retail characteristics, such as Quick Service Restaurants.