How is AI Making Retail and eCommerce More Human-Centric

From empowering the research stage to helping teams craft personalized discovery experiences, Artificial Intelligence (AI) is making its way into developing digital products. We have picked the brains of our awesome Head of Design, Johannes Dornish and Director of eCommerce, Mateo Melichar.

In what ways do you see AI driving the transformation of the design process for digital products?

The integration of AI amplifies capabilities during the conventional development of digital products. The design process usually starts with understanding the people, the business, and the environment and then moves on to problem definition, idea generation, prototyping, and testing. Then, once you show a new solution to the world, you receive feedback and sustain an iterative process of further testing and constant learning.

AI's impact spans every stage of this process. It gathers and analyzes data to understand the problem while learning from a business’ historical data and existing research. If prompted correctly, Generative AI can provide diverse problem framings and unique perspectives for idea generation. This helps teams create infinite prototypes and designs, which is valuable at this stage, enabling them to go very broad. Then, those ideas are swiftly tested and refined with AI-powered coding assistants so businesses can launch multiple options, judge which ones are the best, and continuously adapt based on feedback.

We are already using AI to drive efficiency and quality while scaling. And it doesn't only help us in the development process – it helps customers too, as they can obtain a broad range of options, and the product validation process becomes much easier.

What future possibilities do you envision for AI and the shopping experience?

Many developments are currently focusing on stock optimization and UX improvements. Tailored interactions must be prioritised, and AI can be beneficial when integrating them into the discovery experience. AI-powered solutions can deliver relevant and personalized content in real time, making it easier for users to find products that match their interests.

We will see shopping assistance evolve thanks to search engines powered by generative AI and the use of easily interpretable data. Users will enjoy a "size finder" feature that will suggest, “OK, based on this percentage, this T-shirt should fit you”.

Automated customer service might also become a reality, including voice interactions based on machine learning. It's just a matter of time before these advancements become integral to the shopping journey.

So, can we expect in the foreseeable future to visit a brand’s website and get tailored recommendations for the right outfit based, for example, on specific event characteristics, user preferences, and seasonal weather?

Absolutely! It's just a matter of time and investment. We are currently experiencing a paradigm shift. ML has enormous potential for optimizing data analysis, mainly to provide personalized recommendations and better-targeted PDPS (product detail pages). These advancements and integration of more intuitive AI-powered interaction models genuinely revolutionise user interaction patterns. Brands in the fashion realm are already adopting new shopping assistants powered by ChatGPT, offering their clients the chance to talk or chat with tech instead of using the search field.

Every user-facing company should be experimenting with Conversational AI. Unlike the old chatbot hype and artificial interactions from a few years ago, this technology allows us to address context and emulate human behaviour through its pre-trained model, ensuring more natural conversations.

Speaking of personalization, how can AI help to analyze user behaviour and get relevant insights to offer customized content?

When it comes to understanding actual user preferences, primary research remains crucial. But AI can be a game-changer by training generative models with customer data, simulating human behaviour. Imagine being able to ask AI, “Would a 35-year-old male lawyer living in Madrid like this T-shirt?” or “Would he like this new website feature?”– and getting accurate answers!

The accuracy of AI suggestions, often exceeding 90%, has impressed even the best data scientists in the world. It's theoretically possible, but we must be mindful of data privacy and consider implementing opt-in options to respect users' choices. Striking a balance between innovation and safeguarding user privacy is crucial in this AI-driven world.

Now that you are mentioning data privacy, handling data is one of the crucial challenges for companies when implementing AI-driven solutions. What key aspects should they consider?

There are several vital aspects that brands should bear in mind, with bias in AI decision-making models at the top of the list. These models can learn from biased data sets, leading to inaccurate and discriminatory outcomes. The existence of attack vectors targeting data sets themselves is a critical security topic, as it means that an attacker might try to manipulate the training data sets to affect the results of a trained model.

Protection and verification of data sets are essential and must be handled carefully. To address these concerns, brands must be proactive, regularly auditing AI outputs and ensuring the AI's behaviour aligns with the brand's values through relevant training data. Protecting sensitive data with proper encryption and access restrictions and using secure data transfer protocols is crucial to avoid any data leaks and to enable the secure handling of sensitive information. For instance, technical and organizational measures are essential for working with personally identifiable information.

Measures must be adequate for data processing, which is particularly sensitive for AI systems, especially during the training and improvement of models.

Another thing to keep an eye on is monitoring AI decisions across different user groups, which helps guarantee that the system treats everyone fairly and impartially. Providing human oversight for critical decisions is essential at this stage.

How do you balance the use of AI with human expertise?

AI and human expertise can harmoniously coexist. You need human creativity skills to combine things and find new ways to build successful digital products, and AI can help you forge those new paths. AI can provide this initial dot creation, but you'll need to prompt it so that it connects the dots, okay? And if you want to solve a more complex problem, you need the human factor. In a business context, AI is a copilot, supporting and updating tasks, particularly in repetitive areas.

However, tasks that require the human factor must be carefully handled, as verifying the accuracy and feasibility of AI-generated results remains vital. Understanding, validating, and confirming the results of AI-generated solutions is essential to ensure optimal outcomes. By blending AI into our decision-making processes while valuing the wealth of human knowledge and judgment, we get the best of both worlds.

Looking ahead, how will commerce and retail business models transform as AI uses continue to expand?

For e-commerce players, AI promises to make unstructured data more accessible so they can leverage it to optimize the user experience. The power of AI lies in its ability to help teams personalize the shopping journey through infinite testing capabilities, allowing for tailored suggestions and optimizing conversions for individual user types.

Beyond delivering the best possible experience and expanding business sales, I hope AI will lead to a broader transformation in business models, with a strong focus on process optimization, sustainability, personalization, and making interactions easier and more accessible. From improving returns logistics to finding more efficient and eco-friendly routes, the future of commerce and retail is set to be driven by AI's capacity to deliver personalized experiences and foster sustainability across the industry.

Conclusion

Remember, implementing AI is more complex. Data often needs to be more cohesive. You should take a step before you take a step further.

Suppose we have learned anything this year from engaging with various brands. Theoretically, the omnichannel approach looks very easy, but connecting different data points can be complicated.

Your infrastructure might need to be more cohesive, which prevents easy connectivity. Many brands have codependencies on suppliers and obtaining data there.

Systems like Product Information Management (PIM) can play a crucial role when we work with many products; updating these is still a manual job for many companies, as they need to catch up on the critical infrastructure. Your CRM and CMS systems are essential to lead the way to see what content customers interact with.

Monolith solutions have reigned for ages, but with the new MACH uprising (microservices, API first, cloud and headless) approach, many quickly try to update their infrastructure with new features to be more flexible and scale up, scale down, as the demand rises or subsides.

The challenge is that many of these flagship systems must be utilised more and can be clunky sometimes.

Also, since they miss critical design thinking, they do not focus on the end customer and focus too much on the technology and dealing with bug fixing.

Another challenge is siloed thinking; departments make collaborating difficult, and projects stagnate.

Hence, there is light. AI can bring new innovative ideas, and like every change, it can bring fresh new thinking into teams fresh new blood as we need to reskill and apply new technologies.

Remember that it starts with mindset first, team second and technology third. With enough perseverance, focus and dedication, we can use technology to transform our organizations.;-)