Although we’re still in the nascent stages of emerging tech like Generative AI, there’s much to be said about how these technologies will forever change the way online commerce operates.
We sat down with Constructor CEO Eli Finkelshteyn to discuss ecommerce predictions on what online shopping could look like in five years when enabled by future-fit tech, like Constructor.
Q: What does product discovery look like five years from now?
“One of the more exciting things we’re starting to dig into is this idea that customers need to go offsite to conduct research for certain needs. Take, for example, someone searching for a shirt with a unique design — something that has a modern style, but is still comfortable and is great for going-out in. It’s challenging to describe that need using traditional search with keywords. Or similarly, when a parent wants to select the right Star Wars Lego set for a 7-year-old child they’re hoping to delight on their birthday. In these cases, customers have trouble having their needs met via traditional search on an ecommerce website, and most likely turn to Google for inspiration, reviews, and recommendations instead. This highlights a key limitation of the current on-site search experience. Ecommerce companies can hope the user comes back to the site after Googling, but it’s unclear whether they will, and often requires ad dollars to remind the user to come back.
There are nascent solutions to this problem, like Quizzes — which allows an ecommerce site to ask a shopper questions about their needs to better understand them and to be able to better address them — and AI Shopping Assistant, which lets users express complex needs in their own words, using LLMs and transformers to understand them in a way that was never possible before. These solutions are not yet as wide-spread or standardized as more traditional product discovery solutions like Search and Recommendations, but they are an exciting frontier that hint at how we’ll be able to shop more conveniently in the coming years.
As these solutions become more standardized, we expect to see user interfaces (UIs) for them become more standardized as well. Similar to how a shopper intuitively knows they can find a search bar on an ecommerce website, the shopper will also know where they can express their longer form needs, or have an interface where the ecommerce site asks questions to help understand their needs better. We also expect to see more retail scaffolding become created around these solutions, to merchandize within them, help them play a role in SEO, and more.
Looking ahead, we want to see a continued marriage of interests between retailers and shoppers so shoppers feel like their favorite retailer is like their digital home, and they feel less need to research on sites like Google or Amazon before making a purchase. Shoppers seek a seamless experience where they know they’re getting an advantage from buying from the retailer they’re loyal to, while retailers aim to keep users on their site, avoiding the need for them to go off site for research, which could incur costs in winning them back.
The future landscape is likely to see a holistic integration of these technological advancements, shaping a more intuitive, personalized, and streamlined ecommerce experience for both shoppers and retailers.”
Q: What are the best ways that retailers can prepare for this transition?
“Proactively experiment with new product search and discovery technology now. Don’t wait until it’s a must five years down the line because then you’ll be playing catch-up to your competitors.
And as you’re experimenting, remember that product adoption should take a user-centric approach. Encourage customers to try out new technology via an opt-in. Don’t force it on them. Let them try new experiences when they’re ready and come back only when these experiences are genuinely helpful to them. By nature, if someone visits your website, the easiest thing for them to do is what they’re used to. Offer them a new and enticing way to shop — but only if they’re interested.
Once they opt in, it’s up to you to deliver an engaging, personalized experience. If the experience is great, they’ll keep coming back, and they’ll tell their friends.
As the retail landscape gets shaped, these cool experiences will cause a brand halo effect, where companies driving them are viewed as forward-thinking and ahead of customer needs. This only further expands your customer base and loyalty organically.”
Q: What will definitely NOT be a part of product discovery in five years?
“I’m not sure five years is the exact timeline, but in the foreseeable future, we’ll see a shift in the way people express themselves while searching, mirroring historical shifts we’ve already seen.
Much like the transition from handwritten shorthand and cursive to the efficiency of typing on laptops, the advent of generative AI, transformers, and other emerging tech allow users to search using more natural, human-like language. This progression will cause older, more rigid speech patterns and keyword-centric approaches to become a thing of the past.
When we think back to how we learned to do research online on sites like Google, there was a learning curve to learning to search with keywords. It feels natural now because we’ve been doing it for so long, but it’s like writing in cursive: people learned to do it and kept doing it until typing — a better, easier way to write — came along. Online research and discovery will be the same way, and as people realize they can express themselves in more fine-grained and nuanced language now and that systems built on LLMs and transformers will understand them, I expect it will change the pattern of how we find things online.”
Q: After transformers, what is the next iteration of AI? Is there any nascent tech on the horizon?
“This can be taken in a couple of directions, but I’ll stick to AI specifically. In essence, transformers excel at leveraging all the knowledge they have to predict a reasonable next set of words. This gives them the ability to mimic a profound understanding of and ability to predict various contexts within the universe, which works well when there’s no definitive answer. The downside is that it often does not work well when something does have a definitive answer. Transformers simply generate responses that sound like they could be true, but unfortunately, sometimes are not. This has given rise to terms like “hallucination” when referring to technology like ChatGPT sometimes confidently stating answers that make great grammatical sense, but are also provably false.
A number of companies providing Foundation models like OpenAI are working on this problem, focusing on enabling AI to reason through scenarios with clear, definitive answers. Shifting from being able to understand and generate human-sounding text that’s grammatically correct to a more structured form of reasoning — like solving math problems — sounds less impressive. (We’ve had calculators for years. So, people are used to computers being able to solve this kind of problem.) But, this evolution lays the foundation for not only providing answers that sound human, but also providing the exact accurate response in situations where a single correct answer exists.
As AI enters this new phase, ecommerce companies will be challenged with finding the right application of them to commerce. The possibilities in minimizing returns (imagine being able to have a shopper find out everything they need to know before they buy) or in personalized, conversational commerce are pretty huge, for example. Figuring out how merchandising and searchandising keeps up with the new paradigms will also be full of exciting challenges and opportunities. A new Wild West is opening up before us, and those of us who seize the opportunities to take advantage of this new frontier today will be the ones who define the shopping and overall product discovery experience of tomorrow.”
*This ecommerce predictions blog post is an excerpt from our comprehensive guide, ‘Forward. Fast. The Future of AI in Product Discovery.’ Grab a copy of the ebook by clicking here.