Search has become a ubiquitous part of the eCommerce customer experience. 69% of customers use the search bar to look for products online, making it the most common way to discover products and a key part of the online purchase journey. However, product discovery is still difficult for online retailers, and there is a gap between merchandisers and consumers that existing eCommerce search solutions have not been able to solve.
This gap is the reason why GroupBy was founded. It was the impetus behind our original product, and it is the reason GroupBy was the first to market with a search and product discovery solution entirely powered by Google Cloud Vertex AI Search for Retail. Today, we're going to walk you through our decision to go all in on a new, AI technology, why we made the switch, and the progress we've seen so far.
eCommerce search engines have a unique and difficult job to do. When a customer types a query into the search bar, the search engine has to interpret that query, determine which products are relevant, and then bring those specific products back to the customer.
Product search is different from other kinds of search in that product search tools scroll through a limited product catalog, where each product has a certain number of attributes, such as size, color, style, etc. However, while web search has gotten better over the years, eCommerce search experiences have routinely struggled to bring back relevant products for customer queries.
Irrelevant search results, as we know now, create a poor customer shopping experience, and have a direct negative impact on sales. It was this problem that GroupBy first set out to solve.
Founded by a group of Endeca alumni, GroupBy began in 2013 with the express intent of solving the problem of search for the eCommerce space. In our first quarter of existence, GroupBy was approached by Google with a problem of solving their search for the retail market demand. Given the success of Google’s Internet Search, they wanted to create a Google-like experience for the Retail market. Out of this idea, our first product, Searchandiser, was born.
However, the original Searchandiser was an on-premise solution. With the cloud quickly rising, GroupBy realized that cloud-based technology was the way of the future. As such, we would need to leverage a search technology that was built for the cloud from the ground up.
By the end of 2014, we made the decision to focus on becoming a fully SaaS, API-first, cloud-based solution provider. And that meant choosing a search engine to build our cloud-based Searchandiser on.
The two prevailing cloud technologies for search, at the time, were Elasticsearch and SOLR search. Both were backed by the Apache Lucene library. Where SOLR was open source, Elasticsearch was architected slightly differently, in a way that really allowed it to take advantage of the elasticity of the cloud. Because of this, GroupBy chose Elasticsearch to power the cloud-based version of Searchandiser.
However, like all search engines of that time, they were built on keyword matching technology to search and index text within documents. Meaning, they were not custom built for an eCommerce environment.
The solution? Build logic layers to make these base technologies deliver better eCommerce search experiences. We layered semantic search understanding and AI on top of the engine to improve outcomes for our eCommerce clients. And while our clients were very happy, our engineers were not.
The fact is, whether you're using Lucene-backed or SOLR or Elasticsearch, the underlying technology is several decades old. These engines use inverted indices, keyword matching and relevance calculations to deliver relevant search results. But when your foundation is built on a technology that is decades old, there is only so far you can get by layering new technologies on top of it.
While our clients were happy and our technology was providing excellent customer search experiences, GroupBy knew there was an underlying problem. Client demand was always increasing, and while we were able to meet that demand, we knew there was a technical issue to address.
If we wanted to truly innovate and really solve the problem of eCommerce search and product discovery, bolting AI on top of a legacy search engine was not going to get us there. And that's not just because the technology is "old." The inverted indices and keyword matching technologies that underlie both SOLR and Elasticsearch, were originally developed by the Library of Congress for document look up and full-text search. It was not developed with eCommerce in mind at all. Diverse product catalogs, faceted search, even search analytics -- all are add-ons to this original technology.
Given the rapidly growing eCommerce market and rising client demand, GroupBy knew that we were essentially building on top technical debt. There are only so many layers that can be added over top of a legacy search engine, as the limitations of the underlying technology itself will eventually hinder ability to deliver results. If we wanted to truly innovate, bolting new technologies on top of legacy search wasn't going to get us to the product the market was really demanding.
The Product The Market Was Demanding
As mentioned above, eCommerce sites have complex search needs. In order to deliver a user experience that improves customer satisfaction, retailers need advanced features to avoid irrelevant search results.
An effective eCommerce search engine must crawl through databases not of documents and text, but a database of product catalogs, filled with different product categories, product details and product names, and then display products to potential customers.
This is why legacy search technology struggles to deliver accurate search results for eCommerce retailers, wholesalers and distributors: it simply isn't built for it. And as the market grows, the demand for relevant search results and better eCommerce search experiences continues to rise. Good search functionality is no longer a "nice to have" for retailers -- it's an essential to providing the kind of online shopping experience modern consumers have come to expect.
This is why GroupBy turned to true AI to solve the problem of eCommerce search.
In 2019, Google informed GroupBy that they were working on a new initiative: they were building a new search engine, specifically for eCommerce use cases. To do this, they were completely rebuilding search from the ground up, and this new engine would be built entirely on true AI technology, instead of AI layered over legacy tech.
As an existing Google partner, it made sense to GroupBy to go all in and completely revamp our platform from the ground up to work with this new engine. Over the years, Google has sunk a billion plus dollars into creating technology that understands search relevancy. Their dominance is such that not only is Google a verb, but it was added to the dictionary all the way back in 2006!
This engine that Google was working on would also specifically leverage learnings from Google.com and Google Shopping in its search algorithm, and would include Google’s advanced machine learning models. By leveraging this new engine, GroupBy knew we could achieve a fundamentally better experience for our customers and deliver an unmatched online shopping experience.
Google Cloud's search engine -- now known as Discovery AI -- is unmatched in the market. By leveraging the power of natural language processing and large language models, this new engine is able to understand user intent and context for any search term. With this underlying technology, GroupBy is able provide far more than just eCommerce site search. Our suite of product discovery tools spans the entire purchase journey, going beyond search solutions with product offerings like Recommendations, Merchandising, and Analytics.
- Recommendations: by leveraging the power of machine learning models, this engine can also deliver hyper-personalized recommendations across all channels, for a true omnichannel shopping experience. Multiple machine learning models let you customize the experience for your customers.
- Merchandising: curate campaigns and optimize your merchandising for key eCommerce metrics, such as conversion rate. Merchandisers have all the tools and controls they're used to, and can tune search queries if they really need to, even though the AI powering the GroupBy platform helps dramatically decrease the number of search rules required to deliver a well-optimized search experience. The Google Cloud search engine is able to provide highly personalized experiences, surfacing the exact products a customer is looking for with minimal manual intervention.
- Analytics: Search analytics let you understand user behavior at a glance and are important for optimizing the search experience.
Plus, the API-first, composable architecture enables seamless integration with your tech stack in as little as 7 to 12 weeks! But that's not all the GroupBy platform can do.
GroupBy's dedication to innovation goes beyond simply being first to market with a true AI eCommerce search and product discovery experience. We've continued to innovate, adding on key pieces that make search solutions more effective.
Data Enrichment
Enrich, GroupBy's data enrichment product, is essential to a quality search experience because your search function is only as good as your data. Bad data negatively impacts search, no matter what engine you're on. Enrich helps GroupBy build on the power of Google engine by curating, normalizing and attributing your product data so that all of your product attributes are easily searchable and indexed.
By incorporating Enrich into our Search and Product Discovery Platform, GroupBy addresses one of the core problems that retailers face when optimizing their search experience.
Fitment From The Search Bar
Fitment -- searching by year/make/model -- is a highly complex search function that many retailers require. Any wholesaler or distributor, whether they're B2B or B2C, has this problem. Parts can range from automotive to computer and electronic components, to MRO parts to HVAC installations and more. Given that a single part regularly fits dozens of products (and can fit thousands of products at the higher end), fitment is extremely complex.
Traditionally, eCommerce companies solved this problem with extensive faceted navigation. These long menus were clunky, and made for a not-so-great buying experience. But, by leveraging the power of Google Cloud Vertex AI Search for Retail, GroupBy has been able to solve fitment from the search bar, delivering guaranteed to fit results in as little as 2-3 clicks. Legacy solutions require 7-10.
Switching to an AI-first engine has been a key part of GroupBy's success. And by adopting technology that is built for eCommerce from the ground up, GroupBy has been able to innovate beyond what legacy solutions enabled us to do. In order to stay true to our vision and keep innovating, we've looked at the core issues that impact search experience and built out solutions for them.
AI-first technology is dramatically changing the eCommerce landscape, and eCommerce site search solutions must evolve along with it. In the same way that Lucene-backed search solutions were once the superior option in the market, GroupBy sees a future where the market is primarily backed by Google. The revolutionary engine that powers the GroupBy platform provides an unmatched experience for customers and can deliver relevant results for any query at any stage of the customer journey.
Completely re-coded from the ground up, built on AI for eCommerce purposes, this engine is the way of the future. eCommerce is continuing to evolve, and the technologies we use must evolve alongside the market.
For more information about how you can leverage this AI-first search engine to revolutionize your eCommerce search and product discovery experience, book a demo.
