The Human Element in AI-Driven Ecommerce: Courtesy of The Harvard Business Review
Reflecting on the wealth of data and trends presented in this report, one theme emerges clearly: artificial intelligence (AI) is reshaping ecommerce at a remarkable pace. Yet, amidst the statistics and technological advancements, we must not lose sight of the human element at every transaction’s heart.
The true power of AI in ecommerce lies not in replacing human interaction but in enhancing it. AI-powered shopping assistants can handle routine queries, allowing staff to focus on complex, high-value customer interactions. Similarly, AI can enhance entire product catalogues, ensuring consistent, high-quality content across vast inventories.
These advancements aren’t about creating a fully automated shopping experience but rather about empowering both customers and employees. AI-driven tools that allow customers to “ask” products about their features or care instructions don’t replace the need for knowledgeable staff – they complement it, allowing for more in-depth, value-added conversations.
Looking to the future, the potential applications of AI in ecommerce are boundless. From more sophisticated personalisation to predictive inventory management, the possibilities are exciting. However, as this report clearly shows, challenges around data integration, talent acquisition, and ethical considerations remain significant hurdles to consider and plan for.
As you move forward, remember that the most successful AI implementations aren’t just technologically advanced – they’re thoughtful, ethical, and centred on genuine customer needs. The data in this report provides a roadmap, but the destination is yours to define. How will you use AI to meet and exceed your customers’ expectations?
Ultimately, AI in ecommerce isn’t just about technology – it’s about people. It’s about understanding needs, solving problems, and creating experiences that delight, engage, and inspire. As you apply the insights from this report, I hope you keep your focus on the humans behind the data points. That’s where the real magic of AI in ecommerce lies.
AI: The Secret to a Thriving Ecommerce Business
Ecommerce has delivered unparalleled convenience to customers, allowing them to purchase products anytime from anywhere at the touch of a button. However, mounting competition in a wide range of industries and evolving customer expectations are putting pressure on organisations to deliver more than simply 24/7 access to innovative products.
“These days, consumers are looking for more personalised shopping experiences online,” says David Patterson, a partner with Clarkston Consulting, an Atlanta-based consulting firm. “They want websites to deliver experiences that resonate with who they are as a shopper. They expect organisations to know them, know their preferences, and tailor their marketing throughout the customer relationship cycle. That’s where AI and machine learning models are starting to make inroads.”
By using sophisticated algorithms and large language models to generate critical insights, AI and generative AI (gen AI) solutions promise to provide customers with relevant recommendations, personalised interactions, and fast and easy ways to find the products they’re looking for—in other words, a digital experience that captures the intuitiveness and familiarity of shopping in person.
In fact, according to a May 2024 survey by Harvard Business Review Analytic Services of 213 members of the Harvard Business Review audience, all familiar with their organisation’s decisions about using or not using AI and gen AI within their ecommerce operations, 70% believe it’s very important to their organisation to implement AI (aside from gen AI) within ecommerce operations and 65% believe it’s very important to their organisation to implement gen AI within the same function.
The good news is organisations consider AI and gen AI as more than simply back-office efficiency enhancers.
But while the majority of today’s organisations see the potential for these technologies to truly transform ecommerce, obstacles around data privacy, talent shortages, and technical integration issues can slow their adoption. Fortunately, by adopting best practices, from establishing strong leadership and modernising technology infrastructure to upskilling employees, organisations can take the next step in blurring the line between in-person experiences and ecommerce journeys.
This paper shows how AI and gen AI are important to successful ecommerce operations and explores the organisational and technological obstacles organisations must address to reap value from these innovative technologies. The paper also delves into the strategies needed to drive greater adoption of AI and gen AI in ecommerce.
Customer-Centric Applications of AI
For years, organisations have struggled to match or surpass traditional in-store experiences with ecommerce initiatives. But in today’s experience economy, it takes more than innovative web design and easy navigation to simulate the interpersonal dynamics of buying from a neighbourhood retailer. Rather, successful organisations are leveraging the power of AI and gen AI to serve consumers with unique and hyper-personalised digital customer journeys. And for good reason: 90% of respondents say that hyper-personalised customer experiences are critical to the future success of ecommerce companies.
“Technology is finally catching up to deliver what consumers expect from an online experience,” says Michelle Evans, global lead for retail and digital consumer insights at Euromonitor International, a London-based market research company.
For instance, AI models can be used to analyse vast volumes and a wide variety of customer data, such as purchasing behaviour, to anticipate customer preferences and present customers with hyper-personalised and relevant suggestions. Data-driven product recommendations and personalised search results can help increase order value and boost conversion rates. Moreover, gen AI can create high-quality content for product documentation and “help” articles for agents and consumers alike.
Hoping to capitalise on these kinds of advantages, some organisations have already begun to add AI and gen AI solutions to their ecommerce toolkits. Forty per cent of survey respondents have active use cases of AI (apart from gen AI) within ecommerce operations, while 31% have active gen AI use cases. However, these figures signal that many companies are in the early stages of AI and gen AI adoption and that there is plenty of room for greater implementation.
The good news is organisations consider AI and gen AI as more than simply back-office efficiency enhancers. Rather, organisations seem to be prioritising customer-focused use cases; 40% of respondents say their organisation is currently using AI and/or gen AI as chatbots for customer service.
FIGURE 1 Designed to provide immediate, accurate, and timely customer support, AI-powered chatbots can reduce wait times and provide helpful responses to routine inquiries.
Similarly, 36% of respondents say their organisation is using AI/gen AI to improve on-site search. The ability to find the right information in a timely fashion can have a huge impact on a customer’s perception of a brand. Although search techniques such as keyword search have stood the test of time, enhancing the accuracy and relevance of search results using AI can turn a search bar into an on-ramp to more meaningful customer interactions. Machine learning can also be integrated into keyword search strategies for learning to rank—an algorithmic technique that uses machine learning to improve ranking in site search relevancy so that queries serve up the most relevant results and rank them by relevance.
“AI has advanced to the point where consumers can start to have conversations with machines, through natural language processing and large language models, that understand what they’re looking for,” says Michael Klein, a principal at Klein4Retail Consultancy, a retail and technology consultancy in Oakland, Calif.
Another 36% of respondents’ organisations are leaning on AI and/or gen AI to generate marketing content. Gen AI works by learning from and mimicking large amounts of data to create content such as text, images, music, videos, and code based on prompts. For marketing teams, gen AI is an opportunity to create content, from email messages to promotional offers, that aligns with a customer’s needs, intent, preferences, context, and behaviour with unprecedented precision—and minimal and sometimes zero human intervention.
Despite an apparent preference for customer-centric applications, some organisations are also finding back-office implementations for AI/gen AI, such as analysing customer journey touchpoints (29%), automating demand forecasting (26%), generating website content (24%), and enhancing inventory management (18%).
Tangible Business Value from AI
AI and gen AI’s entry into ecommerce could not come at a better time. Due to budgetary constraints and talent shortages, today’s marketing teams are strapped for time and resources, making it challenging to keep pace with evolving consumer expectations and emerging target markets. Fortunately, AI and gen AI can help ecommerce executives work smarter, not harder, by eliminating segmenting in favour of AI hyper-personalising customer product offerings.
For instance, Klein says, in the past, marketing teams would be required to write business rules around distinct customer personae to help frame marketing messages for specific audiences. “Now, with artificial intelligence, marketers can understand who that persona is on the fly and assemble an experience using content, text, and imagery that differs from one audience to another without requiring human intervention.” In fact, among respondents whose organisations have implemented gen AI within ecommerce, 69% cite improved efficiency/speed of work as a benefit.
FIGURE 2 With gen AI, marketing teams can save significant time by automating marketing tasks such as content creation, campaign management, and deeper customer analytics, all of which can result in a more meaningful customer experience. For example, creating content that educates customers, provides value, and email them timely with the information or present it on your website they need to make a purchasing decision can help them along their product discovery journey.
Another advantage of deploying gen AI is reduced operational costs, cited by 48% of respondents. Ecommerce has made it easier than ever for consumers to return items they don’t want, only adding to existing labour, transport, and inspection costs. However, gen AI-powered product recommendations can minimise the risk of costly returns by ensuring customers receive the products they want when they want them so that they are less likely to return them in the first place.
In addition to operational efficiencies and cost savings, deploying gen AI can help curate customised experiences for diverse customer segments. Consider, for example, how gen AI tools are empowering organisations to educate consumers on products before they make purchases, provide quick and accurate answers to customer inquiries, and even generate product guides based on a customer’s unique online journey. Indeed, respondents point to numerous customer-centric advantages of gen AI, including better utilisation of customer data (46%), improved customer experience/engagement (46%), and improved customer service (45%). Yet only 26% of respondents cite increased sales/revenue as a benefit they’ve realised to date from using gen AI within ecommerce. That’s a missed opportunity to boost the bottom line, as gen AI can create highly targeted suggestions that are based not only on customer preferences but also on price and availability in ways that optimise customer satisfaction and profit margins.
Obstacles to AI Adoption
Together, AI and gen AI seem to be doing the impossible: blending the defining elements of traditional in-person shopping—personalised recommendations, relevant suggestions, bespoke customer service – into everyday ecommerce interactions. “Shopping online tends to be more transactional, whereas shopping in a store is more engaging,” says Euromonitor International’s Evans. “There’s more of an emotional connection with the brand. But technologies like gen AI are making that online experience feel more like shopping in person -more intuitive.”
Yet bridging the divide between physical and digital experiences requires overcoming significant obstacles, starting with the proper handling of critical customer data. “As a consumer, I need to share as much as I can about my preferences and about who I am for these organisations to echo back recommendations and personalise my shopping experience,” says Clarkston Consulting’s Patterson. “But with that sharing of data comes a lot of trust that organisations will treat personally identifiable information with care.” The consumer is fully aware of course, that any infraction will result in loss of future business.
Likely due to factors such as privacy concerns and a stringent regulatory environment, 52% of respondents whose organisations have implemented gen AI to some extent within ecommerce cite data privacy/security issues as a challenge they’re experiencing in the adoption of the technology. Other common difficulties during gen AI adoption include a lack of talent with the necessary expertise/skills (38%), unsurprising given the shortage of professionals skilled in gen AI today, and data or technical integration issues (31%), which can range from data silos to outdated data. Together, these obstacles are preventing organisations from realising benefits such as improved efficiencies, better use of customer data, and improved customer engagement.
These same issues are the top barriers preventing organisations from pursuing ecommerce gen AI adoption altogether, namely, data privacy/security concerns (55%) and a lack of talent with the necessary expertise/skills (49%), with one key exception: The third-largest barrier cited by those whose organisations aren’t moving forward with ecommerce gen AI is the lack of a clear strategy/roadmap (48%).
Ways to Win
Organisations must overcome obstacles to AI and gen AI adoption if they wish to create digital experiences that drive profitability. “It’s all too easy to move to the next brand in the online space,” says Klein4Retail’s Klein. “Loyalty and retention are challenged more than ever before because of the options that consumers have these days.”
Fortunately, establishing the necessary best practices can help address both organisational and technological hurdles to success. The first step toward tapping into AI and gen AI’s value within ecommerce is aligning IT and business teams. As with most technologies, the IT department tends to be primarily in charge of deploying AI initiatives. In fact, of all the roles listed in the survey, C-level IT executives are most often the ones primarily responsible for exploring the use of gen AI for e-commerce operations (36%), followed by a CEO, chairman, or president (27%), according to respondents.
FIGURE 3 Only 12% of respondents say a chief marketing officer is in charge of this task, and just 11% cite the head of ecommerce or product management. Despite, or perhaps because of, the IT department’s larger role in discovering use cases for gen AI, today’s tech teams do not necessarily number among gen AI’s greatest champions. Rather, when asked which group is the bigger advocate for gen AI adoption at their organisation, 45% of respondents said business teams are the bigger advocate, compared to 38% saying IT teams.
“Business leaders understand the power of AI to generate tremendous value,” says Maxime Cohen, professor of retail and operational management and director of research at the Bensadoun School of Retail Management at McGill University in Montreal. On the other hand, IT teams, he says, “have a deeper understanding of the underlying processes, what’s going on behind the scenes, so they better understand the potential dangers and pitfalls.”
One way to bring IT and business teams together to support ecommerce efforts is to establish strong leadership around AI and gen AI initiatives. “If not a chief AI officer, there should be somebody in the office of the [chief information officer] or [chief technology officer] [who] is responsible,” says Klein. Dedicated leadership can demonstrate to employees across all functions that AI is a strategic priority for the business and critical to profitability.
Viewing Gen AI as a valuable ecommerce partner can also help drive greater adoption. For instance, providing customers with compelling and personalised product offerings in a digital setting can increase the likelihood that they will purchase from a brand. In response, many ecommerce teams are relying on gen AI’s algorithms to return new, relevant, and personalised content in response to certain prompts, depending on goals and customer segments.
“The ability for gen AI to provide multiple variations of either visual or written content to support the ideation and brainstorming process is a huge advantage,” says Klein. “It’s like having a dialogue with a room full of 10 different people all with these different ideas.”
But the best genAI and human partnerships are a two-way street. Even the most carefully designed large language models can generate bias or misinformation. To produce the most accurate and effective results, experts say, human and machine collaboration, in which a human validates a model’s output, is necessary.
“To use AI techniques with some sort of human validation or human in the loop is the right way to go for most applications,” says McGill University’s Cohen. “AI can be seen as the copilot that helps humans perform better, but at the end of the day, human validation has to be part of the tool.”
The same rule applies to chatbots, which can excel in providing personalised and relevant product recommendations to online shoppers. However, for consumers who still want access to human support, especially when the issue they’re facing is unique or complex, organisations must ensure they make human decision-making and oversight a key element of the customer service equation. Negated where the AI solution is 100% autonomous.
Another best practice for extracting value from AI and gen AI is clear communication regarding the impacts of the technologies on how employees work. Leveraging gen AI to deliver personalised experiences can increase revenue and customer satisfaction in an ecommerce environment. But there’s also a flip side: By reshaping the way employees work and automating repetitive tasks, companies free employees to focus on more complex, creative, and value-added work. Some employees may even wish to improve their gen AI expertise so that they might play a bigger part in the creative process by helping generate product descriptions and marketing materials. Better yet, gen AI can empower employees to deliver better customer service by allowing them to quickly and efficiently search employee-facing systems, such as internal knowledge bases, for fast answers to important customer questions upon first contact.
“Companies need to have a clear vision and strategy for how they’ll incorporate these new generative AI tools into their day-to-day and what that looks like for an employee with a specific job function,” says Rachel Dalton, head of retail insights at Kantara, a marketing and data analytics company headquartered in London. In some cases, she adds, this approach may require organisations to “upskill their own internal talent in these areas, especially those who are passionate about AI. Being able to upskill employees and support them in pursuing certifications is important to consider.”
In addition to thoughtfully considering how AI and employees will interact, organisations must also recognise the importance of having proper data governance in place to support AI implementation. Supporting every stage of the customer journey are vast volumes of data—information that can be used to personalise customer experiences based on their unique needs and preferences. But for AI models to deliver accurate and relevant outcomes for individual users, data must be clean and carefully governed.
“Companies leading in this space are at the cutting edge of ensuring that they have a very robust and usable data set because they can’t really do anything with these technologies without data,” says Dalton.
In addition to establishing a data governance framework, data privacy policies can go a long way toward building trust between consumers and brands. If consumers know how ecommerce sites are using their data, they’ll likely be more willing to share that data, especially if it means receiving a personalised and relevant digital experience.
However, even the most closely followed best practices can’t fully eliminate many of the issues surrounding the use of gen AI in ecommerce, from consumer privacy concerns to data bias. In fact, 81% of respondents agree that achieving success with customer-facing gen AI initiatives requires calculated risk-taking.
“The challenge is really understanding what are the real-life business challenges that I can solve by leveraging AI,” says Klein. “It’s about balancing how AI will impact the business and what are the business problems we can solve by using AI.”
To strike this balance, he recommends that organisations “start small with a very confined set of use cases, whether it be to deflect 10% of calls using autonomous AI or using AI to improve on-site search.”
Small wins can be calculated using a combination of customer-centric and financial metrics. For instance, among those whose organisations have implemented gen AI to some extent within ecommerce operations, 52% rely on customer engagement metrics to gauge the effectiveness of gen AI initiatives, followed by financial metrics (47%) and customer service metrics (41%). These metrics closely align with the metrics most ecommerce companies use to measure growth and profitability.
AI as Essential for Growth
There are more than 26.5 million ecommerce websites in the world, each one vying for limited consumer dollars.1 AI and gen AI can serve as competitive differentiators, empowering organisations to provide customers with relevant recommendations across touchpoints, personalised interactions, and fast and easy ways to find the products they demand.
But these are early days for technologies like gen AI. “Things are moving so rapidly that we’re only at the beginning stages of what gen AI could bring in terms of changing the consumer experience,” says Kantara’s Dalton.
By mining both on – and off-site customer data, gen AI solutions can recommend products to consumers based not only on what’s most popular but also on their unique digital footprint as well as an organisation’s desired profit margins. Even customer service agents are gradually benefiting from gen AI by gaining fast access to relevant, accurate, and secure answers to detailed customer service questions.
Certainly, obstacles such as data privacy, talent shortages, and technical integration issues can stand in the way of creating personalised digital experiences. However, by setting the stage for strong leadership, a structured learning plan, human involvement, upskilling opportunities, a solid partnership, and clear metrics, organisations can reap continuous value from gen AI as it evolves from a buzzworthy tool to an ecommerce staple.
“There are some natural byproducts of creating these meaningful experiences in terms of more sales,” says Patterson. “But it’s also about creating the ecommerce infrastructure necessary to strengthen lasting loyalty and intimacy with your consumer.”
Methodology and Participant Profile
Harvard Business Review Analytic Services surveyed 213 members of the Harvard Business Review audience via an online survey fielded in May 2024. All respondents were employed at organisations that execute online transactions (i.e., ecommerce) and were familiar with their organisation’s decisions about using, or not using, AI and generative AI within ecommerce operations.
ABOUT HBR
Harvard Business Review Analytic Services is an independent commercial research unit within the Harvard Business Review Group, conducting research and comparative analysis on important management challenges and emerging business opportunities. Seeking to provide business intelligence and peer-group insight, each report is published based on the findings of original quantitative and/or qualitative research and analysis. Quantitative surveys are conducted with the HBR Advisory Council, HBR’s global research panel and qualitative research is conducted with senior business executives and subject-matter experts from within and beyond the Harvard Business Review author community.
