Today, personalization is key to enhancing user experiences on digital platforms. Social media apps like TikTok are already leading the way by utilizing AI-powered personalization to cater to the interests of individual users, but what about online marketplaces? As a curious consumer, I wanted to delve deeper into this topic and explore how, where, and why top marketplaces are using personalization and AI. I found that on average, there were over a dozen ML-powered personalization features on leading marketplaces, which emphasizes the importance of personalization. I believe this information is crucial for anyone looking to create successful marketplaces in today's digital landscape, so I wrote this blog post to share my insights.
Scorecard
To judge, I created a scorecard with a grading rubric that would help determine which marketplace is doing a better job of presenting ideal products to its customers. While some grading points are objective, others are based on my subjective opinion.
Important Factors
Ranking Based on User Interactions (clicks, views, & searches)
Search history helps marketplaces personalize search results based on customers’ past interactions, making it easier for them to find relevant items quickly. This increases the likelihood of a purchase, improves customer satisfaction, and ultimately leads to a better shopping experience.
Search How You Think with Semantic Search
Semantic search is essential for online marketplaces as it enables platforms to understand the meaning behind search queries, providing a more advanced search experience. This gives them a competitive advantage over keyword-based search systems, attracting and retaining more users and sellers. You can experience semantic search in action at https://same.energy/.
Tailoring Search Outputs to Your Specific Location
Location plays a significant role in search results. By changing my location from Berkeley, California, to Chicago, Illinois, I observed that marketplaces offered personalized results based on the new location.
Etsy
Location
When users search for items on Etsy, the platform takes their location into account and shows them search results that are most relevant to their geographical area. This can help users find local sellers and products that are unique to their region.
Chicago and Berkeley IP addresses resulted in different sweater search results. The Chicago search yielded rib-knit woolen sweaters in basic colors, while the Berkeley search showed more funky colorways and youthful designs popular in California.
Etsy promotes local sellers by showing their products to customers in their area, allowing them to benefit from the platform.
Etsy’s Search History
Etsy's personalized search history recommends products based on user preferences, creating a personalized experience and helping customers discover new products. This benefits vendor by creating loyalty among their followers and recurring product sales.
Semantic Search
Etsy's semantic search utilizes a combination of factors, such as the listing's title, description, tags, and user behavior data, to deliver the best results to the user. For example, if a user searches for "handmade pottery," Etsy's semantic search would not only consider listings with those exact keywords, but also take into account related keywords like "ceramic," "clay," "mug," and "bowl." It would also consider other factors like the user's location, and search history to deliver the most relevant results.
The search queries I conducted on Etsy yielded significantly more relevant results compared to the conventional keyword-based search experiences common in marketplaces.
Ebay
Location
eBay does a great job of taking into account a customer’s location address and producing results relevant to the location. Here are a few search results of how eBay’s page looked like when I changed my IP address to Chicago, Illinois:
When searching for black shoes while based in Chicago, I came across completely different search results compared to when I searched for black shoes in Berkeley, California. While some styles were completely different (taking into account activity, seasons, etc), others were ranked differently to show a new form of personalization.
The results displayed on the platform were predominantly local selections. To achieve more precise search results, customers can take an extra step by filtering their location and selecting specific areas to target.
eBay’s Search History
Ebay's search history was mediocre when compared to other marketplaces. It provided relevant and quick results however, the results were basic and not personalized enough. It felt like eBay was providing me with basic results instead of more personalized ones.
Semantic Search
eBay's semantic search impressed me with its ability to provide relevant products even with limited information. In my experiment, I searched for a rare pair of Adidas shoes, and despite the shoe's scarcity, eBay successfully located the pair from resellers.
To determine if eBay could provide the desired outcome, I searched for "orange and black Adidas campus shoes." I was pleasantly surprised when the initial result on the page was the exact pair I sought.
I tried playing with this a little more….
While attempting to search for "Cryptonomicon" on eBay, I accidentally typed "cryptonomicn," but the search engine recognized it as a typo and still displayed appropriate results. This illustrates eBay's ability to comprehend and accommodate errors while providing accurate search outcomes.
Amazon
Location
I assumed Amazon would prioritize location-based search results, but my experiment using a VPN showed otherwise. Here's a search I made for men's fragrances using my Berkeley address.
The image below is the Chicago result:
Occasionally though, Amazon would change the ranking of the products and add a few newer products but overall, there weren’t that many changes to search results. Here is an example:
Overall, while there was some significance to location impact, given Amazon’s stature, I was expecting more relevance to personalization results based on location.
Amazon’s Search History
Amazon's search history is impressive because it tailors the shopping experience to your needs and past searches. This means faster and more accurate search results, personalized product recommendations, and a customized shopping experience, making it easier to find what you want.
The image above is an example of how Amazon selected top books for me based on my past purchases. Recently, I’ve been fascinated by blockchain technology and the crypto economy. Amazon noticed my interest and decided to add more books to engage my curiosity about that specific topic.
Semantic Search
Amazon's semantic search is superior to other marketplaces as it identifies synonyms and related terms relevant to a search query. For example, when searching for "cash holder," Amazon suggested related terms such as "wallet," "purse," and "money clip."
Amazon uses semantic search by analyzing the intent behind a search query, like "running shoes," to provide more relevant results that match the user's intent.
Alibaba
Location
Alibaba uses a customer's IP address to determine their location and show them relevant products and services.
When searching on Alibaba, the first thing I tried was to see trending searches in each location. The image above shows trending search results from Chicago. Seems like users were really interested in searching for 60-inch round tables and bubble maller.
However, when switching my location address to Berkeley, my search results completely changed to other products. Trending searches in my area included things like carpenter pants and shorts for women.
Alibaba personalizes recommendations based on customer location and weather patterns. For instance, if a customer is in New York during the winter season, Alibaba recommends products like jackets, scarves, and gloves. If browsing from a warmer location like Los Angeles, recommendations will differ.
Alibaba’s Search History
Alibaba did not seem to use any of my past search history. When I revisited the website after browsing for a while, the recommended and ranked products did not have a persuasive effect on me. It seems like Alibaba's search results are based on keywords to boost product rankings.
When searching up brown shoes, for example, I saw that targeted keywords were being highlighted on every result on the list. Thus, many products have exceedingly long titles like “hot sale men’s slip-on leather dress shoes” which goes to show that sellers can easily manipulate search results through specific keywords and lengthy titles to get ranked first.
Semantic Search
Alibaba's Semantic Search is unsatisfactory as it fails to provide optimal results when searching for particular items. For example, when searching for Taylor Swift's latest album "Midnights," Alibaba was unable to generate related products such as merchandise or stickers, resulting in poor search results:
Unfortunately, Alibaba only returned results relevant to one word in the search bar, which was “album.”
Above, is another example of how Alibaba was not able to deliver accurate search results. I tried looking for a green turtleneck sweater and it wasn’t capable of showing me the correct color that I was looking for even though they have those items available.
Scorecard Summary
Conclusion
Personalization is everywhere in top marketplaces. Based on my experience, I identified key personalization factors that affect user satisfaction, as well as, purchasing behavior, and created a scorecard to help founders, engineers, and product managers, gain awareness of the areas where implementing personalization can increase conversions.
Each of the major marketplaces - Amazon, eBay, Etsy, and Alibaba - has its own strengths and weaknesses in personalization. Ultimately, for me, Etsy and Amazon lead this list, followed by eBay, and then Alibaba lagging significantly behind. If you're interested in adding these types of world-class personalization to your marketplace you can contact the folks at Shaped and it out for 30-days with no obligation. Thanks for reading!
