Navigating an Evolving Search Landscape: A Decade With Bloomreach
By Irshad Allana
Over the past decade, the e-commerce landscape has undergone a profound transformation, marked by seismic shifts and the emergence of new technologies to fuel growth. The modern e-commerce leader must recognize how these changes have reshaped the industry and ultimately forced a redesign of your digital strategy.
A decade ago, e-commerce was in its infancy and dominated by Amazon. Major retailers cautiously entered the online realm using basic solutions. The focus was on simply establishing an online storefront, but the landscape evolved significantly over time, with startups like Bloomreach introducing automation and machine learning to address specific industry challenges. Meanwhile, e-commerce's rapid growth extended across diverse verticals, demanding a comprehensive approach to discovery use cases. User-friendly tools such as Squarespace, Stripe, and Shopify democratized establishing an online presence, fostering further growth and D2C adoption.
Today, the e-commerce market has transformed significantly since Bloomreach's inception. The rise of data science is evident as experienced teams now demand tools for customizing storefronts, leveraging data for optimization, and experimenting with new technologies. In this era of the savvy customer, transparency and control are paramount, shifting away from "black box" algorithms and vague promises toward customer-driven online success.
AI at the Forefront
Fast forward to November 2022 — the introduction of ChatGPT by OpenAI would cause a huge shift in the AI landscape. In a very simple, accessible, chat-like interface, it showed the world what was possible with the next generation of large language models (LLMs).
Since then, the ecosystem to support the research, development, and deployment of LLM and generative AI-based applications has seen unbelievable growth. Platforms like HuggingFace provide easy access to ML models, while technology vendors such as Pinecone, Weaviate, Qdrant, and Nvidia have seen soaring valuations. Facebook has embraced open source and made its LLAMA/LLAMA2 models open. Microsoft and Google have also been hard at work to infuse generative AI (GenAI) technologies into their lucrative search and productivity applications.
But what does all this mean for product discovery in e-commerce?
The Future of the Market
In short: The experience for the shopper is about to get much better! AI unlocks a whole new world of use cases. Consider the prospect of replicating the expertise, service, and guidance provided by an in-store sales associate — something previously believed to be beyond reach. Gone are the days of delving into lengthy FAQs, reviews, and spec sheets to understand if a product aligns with your needs. Now you can get:
- Intelligent Agents: With the integration of generative AI technologies, we’re likely to see the introduction of intelligent agents embedded into e-commerce sites to quickly answer questions, compare products, and summarize reviews to help you make the best purchase decision more quickly.
- More Personalized Experiences: With embedded personalization, these agents could learn your specific preferences to help give you advice for your next summer dress or new couch. It will be even better than walking into a store and talking to a sales agent you’ve never met before!
- Alternative Ways To Search: Plus, the onset of multi-modal models will seamlessly blend and transition multiple inputs of text, chat, voice, and image to truly not just recreate, but go beyond the experience that a shopper could get in-store.
This experience will be a win-win across the board — more satisfied customers who can discover new products or make better purchase decisions, while also translating into more revenue for online merchants.
The Next Frontier
While there’s a lot of potential, it’s important to keep in mind that these technologies are still in their infancy. There’s a lot more work to be done on the development of these models, and vendors will also need to do their part to ensure that customers are experiencing outsized returns. Here’s what e-commerce leaders should keep an eye out for:
- Cost Considerations: While we’ve seen what these technologies can do, they’re still only truly accessible to a small subset of companies with deep pockets. There’s a lot of work in the space to make these models more economical so they can be democratized and deployed at scale. There are a lot of eyes on this problem, so solutions are likely on the horizon.
- Evolution of Business Models: The appearance and functioning of GenAI applications may differ significantly from previously developed applications from a business standpoint. We anticipate the emergence of new business models that enable companies to deliver value to customers while safeguarding their own margins.
- Critical Role of Trust: As GenAI becomes integrated into various applications, maintaining a high standard is paramount when introducing products powered by GenAI. If "intelligent agents" cannot provide reliable answers, they risk user disengagement. In applications such as search, misrepresentations by a shopping agent about certain product capabilities can lead to dissatisfaction and erode trust.
- Traditional Search Will Remain: There’s no doubt that LLMs and vector-based search systems can understand intent and respond in a way that can help customers get to the answers/products they’re looking for faster. However, the primary form of product search will still take the form of well-established patterns (e.g., <attribute><attribute><product> in English) and simple search terms (e.g., searching for “2% milk” instead of a more elaborate query like "I’m looking for milk that has 2% fat”). The likely evolution of GenAI involves responding to these tail and torso queries, presenting an opportunity for better relevancy all around.
Building for the Future
In a rapidly evolving e-commerce landscape — where the market is expanding, the consumer is more discerning, and technology is advancing at an unprecedented pace — Bloomreach is committed to staying at the forefront of innovation. This is why one of the key vision drivers of our Bloomreach Discovery roadmap is next-gen experiences.
Bloomreach's approach to next-gen experiences is about empowerment and greater customization. For example, imagine an entire ranking algorithm built specifically for your customers that combines the e-commerce learnings of Bloomreach’s default ranking algorithms with your own unique signals and business goals. Advancements in LLMs will help us go beyond traditional product searches and enhance the user experience in novel and engaging ways — think meaningful conversations about your products with a customer via MMS. “Next gen” is all about tailoring your e-commerce platform to your specific needs and preferences, ensuring that you're not limited by any off-the-shelf solutions.
Introducing the Latest in Search Innovation at Bloomreach
Our product and engineering teams at Bloomreach have been busy building for this new future of search and have launched three exciting features this quarter that affirm the market’s evolution toward customization, deeper relevance, and user-friendly search alternatives:
- Algo Weight Customization: Gain unprecedented control over ranking algorithms by assigning weights to specific signals directly on the dashboard. This gives you the flexibility to not only assign weights to user behavior signals (e.g., views, conversions) but also pair those with the signals that your data science team has created and sent to Bloomreach.
- LLM-Based Precision: Drive deeper relevance with a new LLM-based precision mode that utilizes LLMs to reduce noise in the recall set. For example, Bloomreach’s semantic understanding, now supercharged with LLMs, will not only identify and extract the proper product type from a query like “white nike shoes,” but will also match it against related products such as cleats, boots, runners, etc.
- Visual Search: Shorten the customer’s path to purchase with image-based search. For example, a customer unsure of exactly how to describe what they’re looking for can now upload an image with a similar product to search a merchant’s catalog. Furthermore, if they like a specific style on a site, they could just use the site’s image to look for similar products to the one they are interested in — it’s kind of like pointing out an item in a store to an associate and asking for “more like this one.”
In the ever-evolving realm of e-commerce, Bloomreach's decade-long journey has not only mirrored the industry's transformation but has actively contributed to shaping its future. We've witnessed the shift from merely standing up the store to the aspiration of standing out from the crowd through personalized experiences that rival the giants. With our commitment to next-gen experiences powered by large language models, we’re poised to usher in a new era of e-commerce, empowering our customers and partners to curate unique, engaging, and highly relevant shopping experiences.
The next chapter in e-commerce promises to be even more exciting, and Bloomreach is dedicated to leading the way, delivering exceptional tools and experiences that meet the evolving demands of the market. Be sure to keep up with all the latest innovations we’ve introduced in Bloomreach Discovery.