AI Is Devouring E-Commerce

There are decades when nothing happens, and weeks when decades happen.

AI Is Eating E-Commerce

The e-commerce industry has experienced steady and reliable growth over the last 20 years (albeit interrupted briefly by a global pandemic). Each year, more shoppers come online and more retailers light up their website, and together, the size of the pie gets a little bigger.

But when it comes to the technology powering these e-commerce sites, not a lot has changed in a couple of decades.

Shoppers still work their way through large catalogs of products, fiddling with filters, clicking through categories, guessing the right keywords, and swiping through product detail pages.

Shoppers still have to do all of the work to find the perfect product that will make their day (or outfit). A time traveler from the 1990s would find most e-commerce sites look and operate today just like some of the first online shopping experiences during the early ‘dot com’ days.

While shoppers appreciate access to a bigger selection of products, cheaper prices, and delivery convenience, the shopping experience still feels pretty subpar. It’s 2023; where are our jetpacks?

Well in 2023, it’s safe to say that we’re about to get our shopping jetpacks thanks to breakthroughs in Artificial Intelligence (AI).

Just like the popular saying, “Software is eating the world,” that describes the ways in which digital technology is transforming every industry, AI is about to eat e-commerce.

Wait, What?

The problem with e-commerce sites is they basically can’t understand humans. They are often a slick-looking web or mobile interface on top of a product database. So we as humans have to figure out how to speak computer lingo and tell this database what we’re looking for.

We can filter, sort, and match keywords in the product articles, but, since the database is just a computer, it can’t really understand what shoppers are looking for.

This is about to change dramatically. We just experienced a major breakthrough in AI that has transformed the way that computers can understand human language as beautifully illustrated by ChatGPT.

Top 10 Ways AI Is Transforming E-commerce

Here are 10 ways AI is going to transform e-commerce:

1. From matching to understanding
2. Results ranking gets supercharged
3. Real-time data supercharges AI
4. Everyone gets their own model
5. Personalization finally arrives
6. Learning via human feedback
7. Conversational commerce
8. Multi-model
9. AI as a service: AI via API
10. E-commerce gets scientific

1. From Matching to Understanding

Before: Search in e-commerce today blindly matches keywords typed in by a user, occasionally adding the odd synonym, before retrieving results from the database.

This leaves shoppers needing to think hard about the keyword guessing game, hoping to get the computer to return the perfect product.

Future: The search experience is about to get a lot smarter. More sophisticated AI (based on the same models powering applications like ChatGPT) now truly understands the concepts shoppers are seeking and can match through AI the perfect products regardless of whether the keywords match.

These models understand different languages, brands, colors, and synonyms, and they can pick needles out of haystacks even with the most open-ended search queries.

2. Results Ranking Gets Supercharged

Before: E-commerce sites typically sort the keyword results according to which products are most popular and likely to lead to a sale. This ends up with the most popular (often bland) or cheapest products rising to the top.

After: Ranking is super-powered by AI models. These new models create unique rankings for every individual using personalized data based on the shoppers’ behavior, how the AI understands the product based on the query, and real-time popularity and business supply and demand.

These fusion models use learn-to-rank AI models to optimize the outcome for shoppers and deliver compelling results.

3. Real-time Data Supercharges AI

Before: E-commerce sites might use some analytics dashboards to understand the popularity of different products. Data might be updated once a day to the product catalog to power popularity sorting. Merchandisers may use this data to better promote items.

After: Every search, result view, click, product view, basket addition, and sale is immediately sent to live AI models to adapt in real-time to what the customer and the market are doing.

Every experience is unique and powered by real-time learning: when the weather changes, when sales happen, or when new products land, the AI models can adapt at the speed of light.

4. Everyone Gets Their Own Model

Before: Every customer sees the same shopping experience. Vendors power every e-commerce site from the same algorithm.

Future: Every customer has their own AI model that is trained by every interaction they make on the website in real time. Every e-commerce site is powered by unique models trained and fine-tuned on the data provided by each site.

5. Personalization Finally Arrives

Before: Customers might get some basic levels of personalization after multiple visits and after registering an account. This personalization is centered on crude mainly “buy again” and some level of brand following.

After: Personalization starts after the first click and is live in every session, even for first time visitors. Personalization is used across all surfaces including home, search, browse, recommendations and even checkout.

Vendors feel confident enough with their site’s personalization that they show customers what is being personalized and let the users configure settings to their preferences.

6. Merchandising = AI Learning Via Human Feedback

Before: Merchandisers use intuition and educated guesswork to promote items and curate category pages. Often, merchandisers contradict each other’s choices, and half of the team can make things worse (but no one knows which half).

After: Merchandisers have AI superpowers. The AI automates the easy choices and hands them only the most important decisions to make. The machine then uses its feedback to continuously improve its algorithm.

Merchandisers spend more time refining the user experience, sharpening more effective intuition about their shoppers, and using that insight to teach AI.

The AI in return can rate and inform merchandisers by letting them know how effective their changes were.

7. Conversational Commerce

Before: Average keyword length is < 3 words. Most search queries are just categories (like “dress”) or filters (“blue dress”). The site has amnesia and cannot remember the previous queries.

After: Customers experience expert advice and guidance from a personalized chatbot within the site by asking questions and conversing in a human way. Conversations flow as the chatbot asks questions and refines products live in front of the user.

The bot learns about customer preferences and can provide expert advice to a customer choosing between similar products. Think: personal shopper or a genius in the Apple store.

8. Multimedia Shopping

Before: Products are simply text, listed in the title and description. Customers can only search using keywords.

After: Products are vivid and multi-dimensional. AI models can use images, videos and text to provide expert advice about them. The models use data from all over the web to understand products and brands. Customers can ask questions or use images or speech to search.

Generative AI, for example, can show custom models in clothes and recommend outfits based on previous purchases.

9. AI as a Service: AI via API

Before: E-commerce sites are powered by complex, all-in-one monoliths that are impossible to extend and have “black box” transparency.

After: Everything, including the AI model, is composable and easy to integrate via developer-friendly APIs. Individual AI models are offered as-a-service, and charged per usage, with separate endpoints for search, browse, ranking, recommendations, etc.

10. E-commerce Gets Scientific

Before: E-commerce is an art where merchandisers and developers try to drive more conversions using intuition and gut feeling. It’s difficult to understand how to move the needle and if individual changes are leading to better outcomes for shoppers.

After: E-commerce is a science, where AI models are continuously fine-tuning and improving performance through A/B tests and experimentation that provide statistically significant evidence.

The velocity and volume of experimentation accelerates, and shoppers vote with their clicks for what works and what doesn’t.

Now What?

In light of recent market breakthroughs like the rise of ChatGPT, we are now presented with new capabilities that will transform e-commerce by enabling businesses to provide seamless and personalized customer experiences like never before.

Understanding what’s possible and knowing how to capitalize on new possibilities presented by AI to meet business and customer goals is the critical next step for companies who want to evolve and thrive in an AI-tech-led world – and we can help you get there.

A good way to get started is to equip yourself with more AI search knowledge with these resources from Algolia.