Search technologies aim to return relevant results, but their effectiveness varies. Short keyword queries are common, but they limit exploration and discovery. Generative AI can enhance searches by better understanding intent and meaning, improving results and user experience. It offers opportunities for businesses to engage customers directly on their websites and compete effectively in e-commerce. GenAI can be integrated into existing search systems or automate the generation of product synonyms.
Search technologies serve one purpose, to return relevant results. That’s it. The amount of time and effort we spend searching is entirely dependent on how relevant those results are. Assistive technologies like voice recognition and digital assistants may help us execute searches more easily, but whether the results are helpful is an entirely different matter. When the search query is simple, like the name of a brand, product, or key feature, search engines do a really good job returning matching results. But when those queries become more complex, when your search intention is exploration and discovery, legacy search technologies have some critical limitations.
Customers have learned adaptive strategies to compensate for these limitations. One example of these adaptive strategies is keyword and phrase optimization. Customers do not ask questions in complete sentences, but rather in short and precisely-worded phrases. If those keywords are not present in the target content, then conventional search technologies often fail to return helpful results. Much of user intent and context is conveyed in grammatical nuances and when these are omitted we limit the ways customers can make requests. For example, these search queries have very different intentions and contexts:
- Coffee shops
- Organic coffee shops
- Organic coffee shops that have 5-star ratings
- Organic coffee shops that have 5-star ratings near me
- Organic coffee shops that have 5-star ratings near me that are open right now
The average search keyword is two words. Long tail keywords, phrases consisting of 3-5 words, represent 91.8% of all search queries, but only represent 3.3% of search volume. This means that users are conditioned to use shorter keywords more frequently. They attempt the long tail searches, but when they aren’t rewarded with the desired results, they revert back to shorter phrases.
E-commerce customers who do not have a specific product in mind often begin their search in a web search engine. This is because the webshop onsite search does not have direct access to the boundless content available on the web. For instance, most customers would not bother trying to search in a webshop for, “best dorm room gifts for first-year college students.” Unless that shop specializes in those types of products, it is highly unlikely that they have written a specific article or curated a collection of products that are indexed for these keywords. Even if they had, that webshop would not have the resources to produce articles for every thematic search.
Web search engines do a little bit better here because they have access to millions of websites that have precisely these types of articles. But just because thousands of matching articles may exist, it still does not mean that conventional search technologies can find all of the most relevant results. This is because web search engines also contend with the problem of understanding context. This is difficult to do when your search phrase is just a few keywords. This is further complicated by the fact that search engines rank results based on engagement, which means that popular sites dominate in a virtuous cycle which makes it even more difficult for smaller businesses to compete. When your potential customer begins their shopping journey in a web search engine, then your shop competes with the thousands of other shops that might be returned in a web search.
Generative AI (genAI) can now give every webshop the opportunity to engage their customers directly on their site at the beginning of their shopping journey. GenAI augments search intelligence by broadening the types of queries customers can submit. It can account for minor mistakes like misspellings or handle synonyms. For example, most webshops today return very different results across queries like “light brown,” “tan,” “beige,” or “khaki.” They may also return very different results if you search for “polka dot bow tie” or “polka dot bowtie.” Go ahead, and try this search on the world’s biggest e-commerce website by visiting the links provided. You will get very different results that range between hundreds and thousands. And that company invests TENS OF BILLIONS in research and development.
How can genAI improve this? Because relevance depends on two things, intention and meaning. Intention is the purpose or goal of the customer who initiates the action. Meaning is the interpretation and significance attributed by the receiver of that action; in this case, that receiver is the search engine. Users can express their intent more broadly and more comprehensively with genAI instead of short, 3 or 4-keyword phrases. Large Language Models (LLM) that power genAI can derive meaning from those intentions more effectively by accounting for misspellings, synonyms, and relative context. Generative AI and Large Language Models (LLM) can better understand the customer’s intent and meaning when they search to ensure that you are returning the right products at the right time. That could be by augmenting your existing search technologies with LLM to respond to queries real-time, or by automating the generation of your product synonyms to be uploaded to your search indices. Engenai can help you find the solution that works best for your business.
