« Blogs & Insights

AI and personalization: Enhancing shopping, recommendations and service

| June 19, 2024

In today's retail industry, digital transformation rules are fueled by AI and tech advances. Consumers want personalized experiences, driving a rise in targeted marketing and custom products. Online shopping remains strong, complemented by a renewed interest in physical stores. Sustainability and ethics now heavily influence buying decisions, pushing retailers towards responsible practices. However, challenges like supply chain issues and changing consumer needs demand agility and innovation for continued success.

The Kestria Consumer & Retail Practice Group has convened esteemed professionals from various fields to discuss the evolution of AI in e-commerce, ethical considerations and future customer service innovations.

Key takeaways:

Evolution of Personalized Shopping: Personalized shopping has advanced from basic recommendations to real-time, interactive systems using AI, machine learning and deep learning, enhancing customer engagement and satisfaction.

Balancing Personalization and Privacy: Ethical AI use involves data minimization, transparency, robust security and compliance with regulations like GDPR and CCPA to maintain user trust and meet legal requirements.

AI's Role in Customer Service and Marketing: AI is revolutionizing customer service and marketing with large language models, predictive analytics and better integration of online and offline experiences, enhancing customer satisfaction.

The transformation of personalised shopping over the years

For Dr. Elaheh Momeni from Austria, CTO and co-founder at eMentalist.ai, the history of personalised shopping has evolved from basic recommendations to sophisticated, real-time and interactive systems, enhancing customer engagement, satisfaction and loyalty. In the early 2000s, algorithms like market basket analysis and collaborative filtering were used. By the 2010s, machine learning, AI and deep learning allowed for processing more data, including unstructured data like text, images and voice, advancing personalisation. By the late 2010s, retailers tailored entire homepages for each customer, incorporating real-time personalisation and conversational assistants. Recent advancements in virtual assistants, predictive analytics and the integration of augmented and virtual reality have further enhanced the shopping experience. 

Bill Tolany is an Executive Advisor at the Business Development Bank of Canada (BDC), with experience in retail and with brands in North America, specifically in the US and Canada. In the past, small merchants knew their customers personally, making real-time recommendations. While much of that has changed, most business is still conducted in person. For instance, China leads in e-commerce at 50% of retail, the UK and South Korea are at 30% and the US is at 15%. Over the last 30 years, there have been three big waves in the digital revolution:

  1. The widespread availability of the consumer internet.
  2. Mobile experiences that allowed more customization and targeting.
  3. The current wave of AI, promises faster more personalized interactions.

AI is set to mimic the one-to-one relationships of traditional retail by integrating digital benefits into the in-person environment. This will revolutionize how retail is conducted, bringing customizations like facial recognition to physical stores. This is the future of retail in the near term.

Per Konstantinos Frantzis from Costa Rica, VP Finance North Zone Latin America at The Coca-Cola Company, AI has recently evolved, impacting all industries and business areas. Companies now use sophisticated algorithms to explore vast data from sources like browsing history and social media, aiming to provide personalized product recommendations and improve customer satisfaction. Today's consumers expect tailored recommendations, such as eco-friendly product buyers receiving Earth Day promotions. This creates a win-win: customers enjoy personalized experiences and companies gain tools for targeted recommendations, brand loyalty and increased sales.

As stated by Emad El Sonbaty from France, Head of Global Rewards, Strategic Workforce and HRIS at Orange, operators and Telco companies monitor the dynamics of data through their networks. Over the past 15 years, the data journey has evolved from data center servers to the cloud and now to big data. With big data, artificial intelligence has become crucial in managing vast amounts of data, leading us toward a world of robotics. AI revolves around three key elements: analytics, interaction and understanding. These elements enable real-time predictions of consumer behavior, pricing dynamics and machine learning models. This leads to faster decision-making to meet customer expectations and product recommendations. Filtering massive data helps design patterns to enable personalized product displays, offers, demand forecasting and future behavior predictions.

Renee Hartmann, an American living in Portugal, founder of CLA, has extensive retail experience across North America, Europe and Asia. She currently consults for retailers, brands and retail technology startups and leads Coresight Research’s AI Council which includes 250 retail leaders who are implementing AI in their organisations. Her work focuses on AI strategy in retail, especially in personalizing consumer and buying experiences. In a recent discussion with VCs and retail executives, the consensus was that traditional search is merging with chat, evolving into conversational commerce. This approach aims to understand consumer needs and inspire discovery, rather than just providing search results. At a recent visit to Criteo in France, their AI lab head mentioned that if someone wants to buy a bicycle helmet online, it's not just about buying the helmet. He explained that understanding a consumer's motivation, like commuting safely, helps recommend related products, enhancing the shopping experience. Another example is an MIT-funded startup that uses e-commerce conversion data to dynamically update product descriptions based on consumer preferences, improving conversion rates by highlighting desired features like sustainability. The goal is to better understand and quickly meet consumer desires.

Ethical considerations and best practices in using AI for customer data

Emad El Sonbaty emphasizes that balancing personalization and privacy involves technical measures, transparent practices and regulatory compliance. These three dimensions are crucial for leveraging AI to provide personalized experiences while protecting user privacy. This approach enhances user trust and aligns with ethical standards and legal requirements. To summarize, in a world of big data, balancing privacy and personalization involves:

  1. Data minimization and anonymization: Preventing identification of individuals through techniques like data masking and tokenization.
  2. Transparency and consent: Complying with regulations like GDPR, providing privacy policies and allowing users to control their data.
  3. Data security and access control: Implementing robust security measures, limiting data access and regularly auditing access.
  4. Preferences and user control: Respecting and managing user preferences.
  5. Ethical AI practices: Ensuring fairness and explainability in AI systems.
  6. Compliance: Staying informed about global regulations such as GDPR and CCPA (California Consumer Privacy Act in the US).
  7. Governance and accountability: Establishing data governance frameworks, accountability and training to ensure adherence to privacy policies.

These elements are vital for creating a personalized yet privacy-respecting user experience.

For Renee Hartmann, the challenge of AI and consumer data varies by location. Regulations differ significantly between Europe, the US and regions like China, impacting how data is handled. Another challenge is determining which data to clean and use within an organization. Various data functions are used, but integrating them isn't always clear. In-store retail, for example, uses RFID and other data points to improve inventory management, sales forecasting and understanding consumer needs. Retailers face an overwhelming amount of data, needing it to be prepared and structured usefully. An emerging area is synthetic data, which helps train models and learn about potential consumers while avoiding privacy issues. This new approach is gaining traction among retailers.

To elaborate on the previously mentioned points, Bill Tolany notes that AI is essentially a predictive engine and understanding how the model was trained is crucial. Organizations use both public and private models, but for privacy purposes, everything should be on a private instance, even though training often uses public data. AI's use of vendors is similar to other customer data practices. Direct-to-consumer or retail businesses often rely on vendor partners for data services. It's vital to understand the vendor's privacy policies, intellectual property and liabilities. These best practices in consumer data sharing are especially important with AI, where data and interactions scale exponentially. Working with vendors appropriately and keeping as much internal as possible is recommended.

How algorithms accurately understand and predict customer preferences

As for Konstantinos Frantzis, it is widely recognized that AI uses sophisticated algorithms to collect information from multiple sources, understand it and provide customized experiences. This dynamic process involves machine learning algorithms continuously adapting to predict customer behaviors and offer personalized experiences. Examples include Netflix, Amazon and Spotify. Netflix recommends movies based on viewing histories, Amazon suggests products from purchase histories and Spotify creates music lists from listening habits and moods.

In the corporate world, AI modernizes marketing. QR codes on packages allow companies to gather real-time data and optimize messaging. Companies can adjust campaign timings based on consumer scanning patterns. Another example is targeting products to consumer behaviors, like offering electrolytes for hangovers. Algorithms identify peak times for going out and tailor promotions, like discounted combo meals on Fridays.

The key is this dynamic, continuously learning process that helps businesses optimize their approach while providing consumers with a more personalized experience, increasing customer satisfaction.

Following what's been said by Emad El Sonbaty, Dr. Elaheh Momeni also highlights the importance of data. ‚First, we need to collect lots of data, like customer history and product characteristics. This data must be converted into a numerical form through feature engineering, which is critical for the algorithm's learning process. There are different types of algorithms. Simple ones, like Market Basket analysis, find similar products based on past purchases. More complex algorithms use predictive techniques, incorporating historical and contextual data, such as geographical location, to improve accuracy. Evaluation is a crucial part of training. It involves mitigating bias, teaching the algorithm about different customer types and products, using feedback loops and basically integrating human in loop to enhance accuracy. This ensures diverse product and service recommendations. The algorithm learns from human preferences and incorporates more data. Thanks to new neural network algorithms, we can now handle massive amounts of data, which was previously impossible,’ says Dr. Elaheh Momeni.

The future of AI in customer service

Renee Hartmann notes that retailers are seeing the most short-term benefits from AI in marketing and customer service. These areas show immediate benefits from AI applications. Many companies are creating large language models trained on brand data to customize and maintain brand voice, addressing customer personas and intents. Retailers fear customer service going off-brand, but customizable language models can help control this. 

Another key aspect is making decisions on the fly. Training customer service to interact effectively with customers, offering refunds or discounts and using AI to reduce returns by making smarter decisions helps improve profitability.

‚Finally, adding a human in the loop is crucial. Having a fail-safe system where real people step in when needed ensures nothing goes wrong and prevents hallucinations, maintaining brand integrity. Live commerce and one-to-one sales support integrated with AI are also becoming important. It's essential to use AI effectively and know when to involve humans for support and oversight,‘ adds Renee Hartmann.

For Emad El Sonbaty, the innovation is fundamental to ensuring automation and efficiency, aiming for a unified customer experience. Continuous advancements promise to revolutionize business-customer interactions, enhancing engagement and satisfaction. Some examples include:

  1. Developing natural language understanding.
  2. Implementing emotion recognition and response in conversational AI.
  3. Utilizing predictive analytics for hyper-personalization and customer journey mapping.

Emad compares AI to a baby: ‚You teach and feed it information to produce desired outcomes. For instance, in Omni-channel integration, unified customer profiles are crucial. Reflecting on my retail experience, I recall that successful retailers design stores to attract customers. Similarly, AI should drive automated, efficient workflows and self-service solutions to improve customer experiences, as Renee shared.‘

Bill Tolany believes we can expect better integration between online experiences and physical stores, where most transactions occur. Currently, interactions don't fully link online and offline behaviors, making the Omni Channel approach increasingly important. Customer behavior varies with context. For example, shopping with kids differs from shopping alone. AI can help by considering factors like companions, time of day and weather, influencing behavior in real-time. ‚These innovations will lead to more context-specific recommendations. It's about aligning both retailer and customer goals making trips or purchases more efficient and beneficial. This co-pilot environment will be highly advantageous,‘ states Bill Tolany.

Agreeing with Bill, Dr. Elaheh Momeni believes that natural language's impact on processes and responses is crucial for customer satisfaction. Understanding and responding to customer emotions can enhance their experience. Services like self-service and human-like voice assistants will significantly improve interactions. AI-enhanced copilots, combining machines and human agents, can manage tasks such as answering complaints, scheduling appointments and handling inventory logistics, thus improving customer satisfaction.

‚Ethics, transparency, security and privacy are critical areas. This includes reducing biases in algorithms, understanding model limitations and ensuring transparent data use. Customers should control their data, with options to remove, edit or not share it. Protecting customer data is essential for satisfaction,‘ says Dr. Elaheh Momeni.

Konstantinos Frantzis mentions that his company recently implemented a machine learning system to predict sales using data such as weather, macroeconomic factors and past history. This reduces planning time and boosts employee satisfaction. Predictive Analytics will evolve significantly. ‚Many companies use suggested orders, where algorithms predict products a retailer will need. If the customer agrees, it becomes a real order, freeing salespeople for more effective tasks. Lastly, combining AI with augmented reality and virtual reality is exciting. Using virtual reality to test products and AI to suggest purchases creates an immersive experience, optimizing consumer satisfaction. This benefits both businesses and consumers, making the future with AI very exciting.‘

Key roles and talent for organisational evolution

When discussing future talent, Emad El Sonbaty highlights upskilling and reskilling as key areas. These are crucial for learning and development due to ongoing transformation. Organisations face resistance and require buy-in to maintain resources, bring in new talent and ensure a stable environment. Another point is using new technologies in daily activities. For instance, in data management, upskilling and reskilling are needed for roles in data collection, analytics, engineering and architecture across all industries. Leadership must believe in collective intelligence and embrace disruptive ideas. Fortunately, these elements are part of the DNA in the Orange group, helping us thrive with diversity amidst the evolving landscape.

According to Bill Tolany, companies most likely to succeed have a learning culture and adapt to new market developments. The current winning model uses AI to augment, not replace, people. Like a human on a bicycle versus walking, AI helps you go faster and work better. In customer service, employees using AI as a supplement are more productive. Organisations need these skills at all levels, including the executive team, to use AI as a tool and adapt to the new world. It's about mindset and the ability to use these tools. The younger generation, familiar with these tools, might have an edge. Ensuring everyone can adapt to new tools and a changing environment is crucial.

Renee Hartmann states that the new perspective is AI won't take your job, but someone using AI will. It's not about fearing the technology, but fearing not adopting it. Another piece of advice is to use AI like your best intern: let it assist you, but don't let it make decisions for you. Regarding young people adopting AI, it's quite user-friendly, unlike social media, Discord or the metaverse. There aren't significant age barriers. Once people start using it, they realize its ease of use. AI isn't polarizing across generations; everyone can use it. It's about developing the habit of using and experimenting with it.

‚It's about transforming the corporate culture to turn functional leaders in finance, HR or marketing into holistic business leaders. They need up-to-date critical experiences and capabilities to meet business needs. This is the key aspect of cultural transformation. How do we offer these capabilities to the organisation? How do we provide the learning curve, through training or on the job, to enhance capabilities?‘ adds Konstantinos Frantzis. Digital and AI have been discussed for years, but the speed of change requires enhanced organisational capabilities. Senior leaders should aim to develop 360-degree leaders who understand business needs and go beyond their functional silos to remain relevant in the future.

Dr. Elaheh Momeni highlights the need for data engineers and AI engineers, not necessarily AI scientists, who can use existing models and technologies. We don't need to develop everything from scratch but require experts to understand company data and apply available technology. Alongside cultural change management, these roles are crucial: data thread and data engineer.


AI and digital transformation significantly impact retail, driving personalized shopping, ethical data use and customer service innovations. Shopping has advanced from basic recommendations to real-time, interactive systems. Ethical AI balances personalization with privacy through data minimization, transparency and security. AI revolutionizes customer service and marketing with predictive analytics, conversational commerce and better integration of online and offline experiences, enhancing customer satisfaction. Experts highlight the importance of adapting to these changes for future success.

The Kestria Consumer & Retail Practice Group understands the demands of standing out in the highly competitive business environment of today. Recognizing the impact of digital transformation, we partner with clients to identify leaders capable of creating an omnichannel 'one customer' experience across online and offline platforms. By staying in constant contact with the dynamic consumer and retail world, we navigate the evolving landscape where technology and enhanced consumer experiences are reshaping the sector daily.

Blogs & Insights