The rapid adoption of generative AI (GAI) tools like ChatGPT represents a pivotal change in how people search the internet. While there are some early signals of this shift, it takes time to change user behavior — especially for something people have been doing for 25 years (happy birthday, Google). This change may happen more quickly on search engines, where the average search was already three to five words in 2021, but it will be slower on e-commerce websites, where the majority of searches are still up to two words. For example, shoppers are more likely to search for “black pants” on an eCommerce site than “dark bottoms to wear to work.” But e-commerce brands think the change to conversational commerce is on the horizon, and they’re updating the search functionality on their sites in preparation.
Recommended AI ML Blog: Why Architecture Matters with Generative AI and Cloud Security
Vector Search versus Keyword Search
Today, keyword search provides the fastest and most accurate results for short queries like “black pants,” but it struggles to find matches for long-tail searches where the items aren’t tagged with the exact keywords used. That’s where vector search comes in. The best vector search products use natural language processing (NLP) and machine learning (ML) to translate text, images and audio into a vector. Then, vector embedding identifies specific keywords from the search database to a corresponding vector to find synonyms, intent and ranking. Vector search aims to return relevant results even for queries without an exact keyword match, helping reduce instances of “no results” pages.
Some shopping optimization platforms that offer vector search today combine it with keyword search to provide the most accurate results, no matter the length of the query. This helps e-commerce brands match their customers’ search styles as they evolve, which is especially important for brands that sell to multiple generations. If the usage of ChatGPT indicates who will adopt conversational search first, it will be Gen Z and millennials, as 60 percent of ChatGPT users are under the age of 34. But it may take longer for older generations to change their behavior and for vector search technology to surpass and completely replace keyword search.
Why Vector Search Is Essential for Conversational Commerce
Conversational commerce refers to a shopper interacting with a brand through a chatbot, an application like Facebook Messenger or a smart device like Google Home. It makes it easier for shoppers to chat with a company representative, get support and see personalized recommendations, mimicking the in-person shopping experience. Although this technology’s been slow to gain traction, it’s picking up speed — spending via conversational commerce channels is expected to reach $290 billion by 2025.
A consumer engaging in conversational commerce assumes that someone (human or bot) on the other end will respond, prompt follow-ups and make recommendations.
Here’s where the user’s behavior changes. e-commerce sites aren’t asking a shopper to change the way they search in a search box.
They’re presenting a new way of finding information where it’s more intuitive to use natural language.
When shoppers use natural language, the website they’re on must use NLP to analyze the text and give it meaning and structure. Vectorization converts words into numerical vectors that encode their meaning and can be mathematically processed. This technique identifies synonyms and intentions and groups related words together logically. Vector embeddings are a common technique in this process, which utilizes neural networks. These networks are modeled after the brain’s neurons and use deep learning to recognize complex patterns through mathematical functions trained on large datasets. This technology is essential for the success of conversational commerce, where keyword search alone wouldn’t find a match to a long natural language query.
The Future of e-Commerce Search and Discovery
The rise of vector search and the adoption of conversational commerce are changing how people search and shop online. Vector search provides an alternative, augmenting approach to keyword search, with the goal of improving relevancy for long-tail natural language queries often used in conversational commerce. As search capabilities evolve, some eCommerce brands are adapting to support natural language queries that may become more common through conversational commerce interfaces. While adopting these new technologies may be slower for some generations, the benefits they offer in terms of convenience and personalization are likely to make them a staple of online shopping in the years to come.
Top News: Resulticks Carves Out New Niche in Audience Engagement–Connected Experiences Powered by RESUL
