Personalized Search

Personalized search solution by Vue.ai uses AI to make search results accurate and personalized for every shopper.

When most people think about how ecommerce has recently gone nuts, they think of the impact of COVID-19 forcing shoppers to steer clear of brick-and-mortar stores. The switch to browsing and buying online en masse—both for B2C and B2B shoppers—is looking pretty permanent. Online shopping is hot. People have gotten so comfortable with it that they’re adding items to their shopping carts that they never used to buy online.  

Personalization on a retail website can take many forms, ranging from providing the most relevant search results to supplying on-target recommendations to simply remembering customers’ names and saving their item color and size preferences for the next time they log on.

Many ecommerce sites are ripe for improvement, but some businesses have yet to figure out how to apply search personalization and reap the wide-ranging benefits, including improved conversion and higher user satisfaction rates. According to a Forrester survey of digital-experience deliverers, 53% say they’re lacking the technology they need to personalize web experiences for their users. 

Estimates that ecommerce companies can increase their conversions 30% by providing an exceptional user experience. One approach is following ecommerce personalization best practices to substantively reflect each individual’s shopping needs and preferences. You can test what you do to see if it works and then fine-tune to get your strategy just right.

When it comes to applying search personalization for improving the shopping experience on an entire marketplace site, which is akin to a brick-and-mortar mall with separate stores (think Etsy), things can get a little more complex. A marketplace-oriented search engine must be able to expertly handle advanced search and browse scenarios such as dynamic faceting, content carousels, autocomplete drop-downs, and federated search. 

A major benefit of setting up personalization on a marketplace site is being able to free up your engineering team and business teams to focus on what they do best, instead of getting bogged down in the technical aspects of implementing and maintaining an effective solution.

Amazon.com uses merchandised content carousels to create an extensive online “department store” look, calling out categories such as popular deals, holiday decor, sporting goods, and last-minute deals.

Visitors to the Houzz marketplace website typically don’t realize all the shopping and content-discovery possibilities they have, so the search bar proactively provides multiple options, including “Discover design ideas,” “Shop products,” and “Suggest pros for me."

If you’ve shopped the Etsy marketplace, you can see how successful the site has been with applying personalization to re-engage customers each time they set virtual foot in the “lobby.” Based on people’s recent searching, browsing, and buying activity, Etsy showers them with sure-to-entice product and shop recommendations, even items from relatively local sellers (within their state). 

People are continually seeking all kinds of media content online—news, live-streamed or taped sporting events, cultural events such as dance presentations and concerts, songs, movies, videos. And they expect that the most relevant, freshest content will be available to them in real time, in an instant, regardless of their physical location and the time of day.

This means personalization on media sites and apps is every bit as critical as it is for  ecommerce and marketplace applications. Media site and app personalization gives users content that reflects what they’re most likely to be interested in watching, listening to, or reading on their device of choice. Personalized recommendations and other tools can keep news and entertainment junkies coming back to see what else they could enjoy on their device of choice.

Companies in the software-as-a-service (SaaS) world can put personalization to work in a uniquely important way: by giving their customer-support and business teams quick access to material they’ve accessed and engaged with the most often, as well as information that’s tailored to their roles (for example, a tech support rep vs. a salesperson). With the right details at their fingertips, company reps can improve their efficiency at enlightening users and prospective customers who need guidance. Better yet (or perhaps in addition), people seeking help can have their questions proactively answered so that they can potentially resolve their own issues, saving time and taking pressure off of the support folks. Externally facing sites are the obvious “needs improvement” targets when it comes to the perceived need for personalization, but don’t discount what personalization can do to enhance enterprise-level search and  improve enterprise workplace search efficiency. 

Personalized Search

Consider workplace search. Large organizations typically have a bunch of disparate silos: content repositories, domains, and platforms housing various functions, such as an HR portal, a tech-support ticket system, and a document database.  Personalizing workplace search could mean using features such as custom ranking, dynamic re-ranking, and query search suggestions to better meet employees’ information needs. You can also personalize search results based on what employees enter in the search bar, as well as optimize the results presented to certain worker groups, tailoring what’s returned to, for instance, a marketing communications team.  Personalized search is a customization of search engine results created by a filter that takes into account potentially relevant information such as the user’s history, location and preferences.

Google introduced personalized search in 2004, promoting it as a way to “understand exactly what you mean and give you back exactly what you want.” Google uses a combination of information from user search history, bookmarks and personalized Google+ pages, among other services that are tied to user accounts. Sites you tend to visit more often appear higher on search engine results page (SERP). Although it has less information to go on, Google also attempts to provide personalized search to users that do not have a Google account and do not have search history enabled. Microsoft Bing is another personalized search provider. Like Google, Microsoft has many pages and services a user might already interact with. Those sites and services yield information for customization of both search and advertisement results.

Personalized search is more useful for some types of searches than others. It’s convenient, for example, for finding local services and finding information on topics that you’ve researched before. For other types of searches, however, such as researching a new topic, the customization tends to yield less valid and relevant responses than would be the case for an unfiltered search.

One of the concerns raised by critics of personalized search is that it limits the user’s view of the Web. That limitation is sometimes referred to as a filter bubble, which effectively restricts the user’s perspective and the information available to them. To disable personalized search in Google, add pws=0 at the end of a search URL. Another option is to click the gear icon in the top right and go to search settings > personal results > do not use personal results. Google also provides a bookmarlet: Turn off Google personalization. In Bing, click on the gear in the top right and go to: search history > turn off search history.

A personalized search reaches beyond the user’s initial query by incorporating information gleaned from his digital history of interests. There are at least two ways in which personalized search can be delivered: modifying the search query, and re-ranking the search results. The search engines of yore catered to all users in the same way. Not so today. The search engine learns about the user profile from the individual’s search history. This profile is then mapped to a set of categories that the search engine accesses to make sense of the query. In effect, the search engine disambiguates keywords in order to serve up what the searcher is actually looking for. With this level of personalization, search engines provide different results for different searchers.

Google first introduced personalized search in 2004 and began actively implementing it in 2005 in the Google search results. Various factors come into play when personalizing a user’s search results, like the location of the user, the search language and the user web history.

Search engines today want to reach beyond what you type in the search box. They seek to understand what your intent might be in order to give you exactly what you want. Or, what they think you want. To do this they use information you have provided, either wittingly or unwittingly, to create user profiles, or personas. The search engine then utilizes the persona to target information in a way that best serves your perceived interest.

The more often you visit a certain web page, the more Google believes that you like that page. Your behavior is duly noted and used to optimize future search results. The next time you search, a page which you have been previously visiting with regularity is likely to be shown more prominently. Herein lays both advantages and disadvantages of just such a likelihood.

Finding new information becomes a major disadvantage of personalized search. You will be repeatedly shown the same results because that is what you always click on. It’s in your history. This naturally narrows your information exploration on the web. The problem is called ‘filter bubble’ and it means that people are letting search engines decide what the users want to see and what they don’t want to see. In the extreme case the user becomes confined to his own ‘filter bubble’ and is cordoned off from the rest of the information out there on the Internet.