Solving the Challenges of All-AI E-Commerce Search With a Hybrid Approach

Once the purview of science fiction, artificial intelligence (AI) has now permeated our everyday lives — from the home devices that recommend songs for our playlists, to the apps we use to monitor commuter traffic, to the chatbots that pop up to ask if we need any help. Unbeknownst to many consumers, AI has become a powerful force steering online searches — serving up results that, ideally, match exactly what we’re looking for. Now retailers are rushing to make the most of AI technology to deliver a better customer experience that drives sales. They’re also encountering the challenges that come with an all-AI approach to e-commerce search.

According to one recent industry survey, the AI-enabled e-commerce market size is expected to expand to $16.8 billion by 2030, reflecting an astounding 15.7 percent compound annual growth rate (CAGR). Over 50 percent of e-commerce retailers now employ some type of AI technology, and 84 percent of them are making the integration of AI into their businesses a top priority. Around 80 percent of retail and e-commerce businesses are currently using or plan to use AI bots in the near future.

Many of these AI investments are being targeted specifically at e-commerce search with the goal of improving the search experience. AI technology and tools offer the potential to drive automated efficiencies while enabling more personalization, convenience, and accurate search results for online shoppers. Along with making it easier and faster for shoppers to find what they’re looking for, AI promises to allow for a more natural conversation between human and computer with the use of GPTs (generative pre-trained transformers) and LLMs (large language models).

Given these smart search features, many retailers are now looking to AI-powered e-commerce search to drive sales. According to the Artificial Intelligence in E-commerce Global Market Report for 2024-2033, the use of AI in e-commerce will spur a revenue growth for retailers from $7.01 billion in 2023 to $14.07 billion in 2028. A survey of e-commerce decision-makers in North America and Europe showed that around 70 percent of e-commerce business owners believe AI will help them make better decisions about personalization. The same report also found that 54 percent of these businesses agree that AI will help with site searching functionality. But is AI really all that?

Online Retailers Confront an AI Reality Check

In the rush to embrace AI and all its advantages, retailers are bumping up against the reality that AI is complex, constantly changing, and difficult to measure and control. Research reveals that business leaders have significant concerns when it comes to leveraging AI-enabled tools and technology.

A recent Forbes survey found that over 40 percent of businesses' decision-makers are worried about becoming overly dependent on AI technology. What’s more, 35 percent expressed anxiety about the lack of in-house technical capabilities to use AI efficiently. Around 30 percent of surveyed business leaders said they were worried about AI-generated misinformation, and 24 percent had concerns about AI negatively impacting their customers.

Clearly, retailers understand and accept the need to adopt AI — particularly when it comes to delivering high-quality search required for today and tomorrow’s consumers and competitive landscape. However, retailers are also encountering the myriad challenges and complexities involved with deploying and using sophisticated AI tools and technology, especially as the AI landscape continues to shift on a daily basis.

Sure, there are an abundance of technology vendors promising that their cloud-based solution is all you need for AI-driven e-commerce search. And many of these vendors will also assure you that you won’t have to lift a finger once AI is in fully charge of your search results. As attractive as that may sound, there are some very compelling reasons not to hand your search function completely over to AI. Because, as we’ve become all-too aware, AI may be smart but it’s far from perfect.

The Challenges of AI in E-Commerce Search

To understand AI’s weaknesses in e-commerce search, we’ll need to take a quick peek under the hood. AI-powered search techniques can intuitively match terms and phrases entered by the shopper to display results, avoiding the need for exact keyword matching. For example, with a search phrase like “keep cool,” an AI tool might display air conditioners and fans among the search results, even though those exact terms weren’t used in the original search.

While this can be advantageous, it can also lead to irrelevant and inconsistent search results. For example, an AI search tool might include rat poison in the search results for a shopper who entered the search phrase “weed whacker.” That kind of inaccuracy isn’t good for business. Research from Google tells us that 53 percent of U.S. online consumers abandon their carts and go elsewhere if they can’t find what they’re looking for on a site. That search abandonment costs U.S. retailers more than $234 billion annually.

Conversely, traditional search technologies that most retailers are already using employ a keyword-based approach that matches search results more precisely. For example, a shopper who searches for “weed whacker” will be shown results that match that exact phrase — no rat poison, here. The advantage of this traditional, keyword-based approach can be more relevant and repeatable search results.

So, which method should you be using — AI or traditional? The answer is both, along with human-touch search merchandising.

Hybrid Search: The Best of Both Worlds

The good news is when it comes to search, it doesn’t have to be — nor should it be — all AI or nothing. Increasingly, retailers are realizing that a hybrid approach to search allows them to simultaneously take full advantage of AI’s automated efficiencies and smart search features, while maintaining the traditional keyword-based search methods they already know and use.

By blending exact keyword matching with fuzzy AI-powered matching, you’re able to harness the strengths of both methods. Your search becomes AI-assisted rather than AI-dominated, enabling you to ultimately get the best of both worlds.

Human-Touch E-Commerce Search Merchandising for the Win

Let’s take it a step further. Along with a hybrid approach, successful search greatly benefits from human touch. Because let’s face it: there are times when you need to be able to manually adjust and refine search results and show shoppers the products you want them to see in the order you want them to see it.

A solution that employs hybrid while also allowing a merchandiser to apply a human touch keeps you in control of your search experiences and results. And that’s the way it should be. It also gives you the flexibility and agility to meet evolving business objectives, quickly react to emerging trends, promote sales and clearance items, push high-margin merchandise, and more.

Whatever your chosen path, the bottom line is this: don’t completely relinquish your search to AI. Find the tools that allow for curation and composability along with all the new-fangled AI technology that works alongside traditional methods to deliver the experience your customers expect — and the revenue boost you seek.

Jason Hellman is senior solutions architect at Innovent Solutions, and product manager for FindTuner, an e-commerce search merchandising and machine learning solution for Solr.