The digital world is ever-changing, and search technology has come a long way from basic keyword searches. Now, AI (Artificial Intelligence) and large language models (LLMs) are stepping into the spotlight to provide even better search results (in theory).
What does this mean for eCommerce brands today? Well, likely, a hybrid approach – which I’ll detail below.
But first, here’s a quick refresh on…
Remember when search was all about keywords? The first (and now antiquated) search engines used statistical methods to match queries to indexed items.
Web search providers like JumpStation or AltaVista were fast and worked well for exact matches, but they struggled with long-tail queries, synonyms, and concepts. So, we started to add AI-driven features like synonym generation to make things better.
Semantic search engines came along to understand the meaning behind words and phrases, using natural language processing (NLP) techniques. Instead of just matching keywords, semantic search engines delivered more accurate results for broader queries.
For example, "What is the tallest building in the world?":
A literal processing of this search is confined to web properties that have keyword elements that match the user’s search query — so, great win for “TallestBuildingInTheWorld.com”! Semantically, though, we know the user’s intent, and we know the specific information they’re seeking.
So search engines became much better at cutting to the chase.
With AI-driven search, we started using various machine learning techniques for query processing, retrieval, and ranking. Large language models like GPT made AI more popular in search, helping us understand and process natural language text more effectively.
Companies like Google pioneered the technology that allowed LLMs to create vectors for comparing queries to results, making the search experience even better.
Just look at this great example from OpenDataScience.com. Let's say we have these common query words: "cat," "chicken," and "apple." We can turn them into vectors like this:
Vectors are like magic codes that describe what the words or phrases are about. When we change words into vectors, we can easily find similar words or phrases by comparing their magic codes.
Generative AI, like OpenAI's ChatGPT and Google's Bard, takes search to the next level by allowing users to ask a question in natural language and receive a direct answer, instead of having to scroll through pages of search results. This new technology creates unique personalized content for each query and could impact the types of content publishers produce, as they will need to optimize content for generative search queries.
Generative AI may also affect the media ecosystem by changing audience behavior and creating new commercial models. While it has the potential to revolutionize search and change how we consume information online, the industry still has work to do to improve the accuracy and quality of generated content.
For enterprise organizations, there isn’t the luxury of having a “black box” AI solution that you can set and forget. Rather, business leaders need to collaborate with partners that can deploy AI efficiently in the right places across their enterprise.
Here are a few of the strategies I am helping our clients navigate.
The optimal retailer site experience will combine the best of both worlds - keyword search and vector search - to give users an amazing experience. Hybrid search is fast and accurate for exact matches and simple phrases, while vectors improve long-tail queries and provide more relevant results.
Your marketing and merchandising teams should craft product detail pages that address both use cases. Keyword relevancy in headlines, bullet points, and descriptions are still important. You also need to include semantically related text to help search engines place your products in the correct related queries.
For example: Product Detail Pages (PDP) need to speak to the queries of “Travel coffee mug” and “What’s the best brand for leak-proof coffee mugs?”
As generative search becomes more popular, think about creating content that's optimized for long-tail and natural language queries. Make sure your product descriptions, social media channels, and blog posts are informative and conversational to cater to this new search trend.
You’ll want to think about common questions or prompts that users may input to AI queries, then craft clear and concise answers in your copy.
Keep an eye on emerging search technologies, like neural hashing (search vectors on steroids) and generative AI. Stay updated on new tools and platforms that can improve your eCommerce search experience and be ready to embrace change.
This process is moving forward rapidly, so it’s important to be flexible and adaptive. And remember: your specific use case for leveraging AI will be unique. There is no one-size-fits-all.
Work with AI and machine learning experts to help you navigate the complex world of advanced search technologies. This investment will help you deliver an enhanced user experience and stay ahead of the competition.
At Pacvue, our team is working day-in, day-out partnering with the largest brands and agencies on navigating the AI technologies they’ll need to stay competitive and profitable. If you want to understand the industry’s transformation for our leading experts, contact our team.
Timely trends and need-to-knows delivered straight to your inbox.