Month: January 2025

What makes eCommerce search successful? Part of the answer, of course, is deploying a solid search engine that can accurately interpret shopper queries and display results that customers find relevant.

But simply having a great search engine at your disposal is often not enough on its own to optimize the value that search brings to your eCommerce business. To take the power of search to the next level, online retailers must also be able to leverage techniques like search optimization and machine learning to deliver personalized search results that align with sales and marketing goals.

This is where search merchandising comes in. As we explain in this blog, search merchandising makes it possible to double-down on the value of search as a vehicle for enhancing the customer experience, boosting eCommerce conversion rates and maximizing business success.

What is search merchandising (searchandising)?

Search merchandising (also called searchandising) is the art and science of presenting search results in ways that account for relevancy, shopper intent and business objectives. The purpose of search merchandising is to call shoppers’ attention to specific items or types of items that the business wants to promote. Search merchandising works by taking the results generated by a search engine, then modifying them in some way that adds value. Examples include:

• Personalizing search results by highlighting products that a specific shopper is most likely to find relevant based on their search and purchasing history.
• Dynamically presenting facets and filters based on shopper context making it easier to find products.
• Highlighting products that have a higher profit margin than other products. In this way, search merchandising would help the business boost its profitability.
• Promoting closeout items to free up space for new stock. This strategy would help the company achieve its inventory management goals.

In short, instead of simply relying on the results generated by a search engine to determine how products are presented, search merchandising leverages analytics, shopper behavior and other contextual signals, along with marketing and sales goals, to enhance the effectiveness of search results.

How does search merchandising work?

Typically, search merchandising strategies are driven by a combination of three types of data. The first is information that reflects customer goals and intent, such as a shopper’s purchasing history and click-stream behavior. The second is data from backend systems that track revenue, margin, stock status, ratings and reviews, and availability (among others). The third is data associated with marketing and sales priorities, like products that the company wants to promote.

By configuring search to deliver results based on each type of insight described above, online retailers can ensure that each search generates results that are optimized for the needs of the business and shoppers.

Why is search merchandising important?

The main reason why ecommerce search merchandising is important is straightforward enough: Search plays a central role in helping shoppers find products on online shopping sites. By ensuring that search results align both with shopper preferences and marketing goals, search merchandising maximizes the value that search brings to eCommerce businesses.

To place those statements in context, let’s dive a bit deeper into the significance of search in eCommerce and why generic search results – meaning those that have not been merchandised – often fall short.

43 percent of visitors head straight for the search box when they open a website, and shoppers who use search functionality are more than twice as likely to purchase products as those who don’t. This means that the better your site’s search is at leading shoppers to relevant results, the more likely it is to drive sales and increase revenue.

Unfortunately, not all search engines are fully optimized for the task of connecting shoppers with the most relevant products and enabling business success. To achieve those goals, online retailers need search engines that are not only capable of interpreting shoppers’ intent accurately, but that can also tailor search results in ways that help support business goals.

The three pillars of effective search merchandising

The most effective search merchandising strategies are based on three essential practices: Leveraging AI and machine learning, providing business users the ability to merchandise search results and optimize shopper experience.
Let’s dive a little deeper into what each of these “pillars” of search merchandising entails.

1. Leveraging AI and Machine Learning

   With this approach, there is no need for employees to configure curation criteria manually each time the data changes. Instead, retailers can simply define overall goals – like increasing profit margins or aligning search results with marketing campaigns – and let the search engine customize each search result accordingly.

2. Merchandising Search Results with Human Touch

   Merchandising search results with human touch brings a unique, shopper-centered perspective. By carefully selecting and promoting products and brands that resonate with shoppers, merchandisers can create more meaningful connections and guide shoppers toward the best options. This personalized approach allows online retailers to shape search results in ways that align with shifting preferences, seasonal trends, and brand identities. Doing so makes it possible to leverage search in support a variety of priorities, including:

   • Product promotions: Search merchandising can help businesses promote certain products that they want to prioritize. Just as important, it can bury or exclude others.
   • Personalized search results: Each shopper is unique. With search merchandising, retailers can dynamically optimize search results for individual shoppers based on data like purchasing history – as Migros did, for example, with help from FindTuner.
   • Shopper-specific results: Sometimes, as in B2B eCommerce, certain products may only be available to certain customers based on factors like the region the customer is located in or the type of relationship it has with the retailer. With search merchandising, eCommerce sites can ensure that the products each customer sees are actually available for that customer.
   • Collection-based merchandising: Effective search merchandising makes it possible to group certain products together. For instance, if you want to highlight products related to an upcoming holiday, you could tailor search results accordingly.
3. Optimizing the Shopper Experience

   When implemented effectively, search merchandising can help optimize the overall shopping experience.

   For example, a solid search merchandising tool can generate dynamic facets – meaning facets that are generated automatically based on the product data returned from a shopper’s query. Because dynamic facets reflect the unique context of each shopper’s journey, they can help with filtering search results and finding relevant products quicker. In turn, dynamic facets help to boost conversion rates and revenue.

   Along similar lines, search merchandising enables facet customization, either by machine learning or human touch. For instance, a site could use machine learning to display facets that allow shoppers to sort product results based on brands they’ve purchased previously or price points that align with their prior purchasing behavior.

   Search merchandising can also power visual navigation which helps shoppers find what they’re looking for faster. For instance, a search for “appliances” could provide a list of results with prominent visual elements that draw shoppers’ attention toward popular categories of appliances, while deemphasizing appliance types that may be of less interest.

   Displaying banners helps capture shoppers’ attention and guide them toward specific promotions, products, or categories, enhancing their overall shopping experience. Banners can highlight special deals, new arrivals, or trending items, making it easier for them to discover relevant products and feel inspired to explore further. Strategically placed banners add a visual, curated layer to the search experience, encouraging higher engagement and driving conversions.

   Through practices like these, search merchandising not only helps online retailers achieve their sales and marketing goals, but also enables the smoothest possible shopping experience for customers.

Search merchandising best practices

In virtually any form, searchandising can deliver major advantages to retailers. But to get the very most out of the it, eCommerce companies should consider best practices like the following.

1. Start with a Strong Foundation

   Effective search merchandising begins with search results generated by a well-tuned search engine. When the search engine accurately interprets shopper queries and delivers relevant results, merchandising efforts can focus on enhancing rather than correcting those results. By starting with a strong foundation of precise, reliable search results, businesses can use merchandising techniques—like product promotions, banners, and custom ordering—to add value, spotlight key items, and create a cohesive shopping journey that truly resonates with shoppers.

2. Combine Automation with Merchandiser Control

   Combining machine learning and AI with merchandiser’s insights enables search merchandising to achieve scalable, impactful results. AI and machine learning analyze vast amounts of data quickly, identifying patterns and optimizing search results at scale, while merchandiser control enables alignment with brand values and evolving trends. This approach allows retailers to manage large product assortments effectively, balancing efficiency and personalization to create search experiences that feel both relevant and thoughtfully curated at scale.

3. Perform A/B testing

   A/B testing is essential in search merchandising because it allows retailers to understand what truly resonates with their shoppers. By comparing different versions of merchandising strategies—such as product placements, banners, or promotions—businesses can identify which approach drives the most engagement, conversions, or satisfaction. This data-driven approach enables continuous optimization, ensuring that merchandising decisions are informed by real shopper preferences, which ultimately leads to a more effective, responsive search experience that aligns with evolving shopper behavior and business needs.

4. Implement no-code search merchandising

   To get the most out of search merchandising, businesses shouldn’t have to devote the time and expertise of their software developers. Instead, they should adopt no-code and low-code solutions, like FindTuner, that allow business teams to configure the criteria that shape search results.

   With this approach, anyone involved in eCommerce – merchandisers, product managers, search managers and beyond – can configure search without having to learn special technical skills or tools. In addition, no-code or low-code search merchandising helps ensure that retailers can update search strategies quickly. They don’t need to begin preparing months in advance to update their search engine’s configuration so that they’re ready for a holiday sale, for example. They can also react quickly to changing market or inventory conditions by factoring them into search results.

Getting started with search merchandising

Because search merchandising involves multiple techniques and complex technology, it might seem challenging to begin benefitting from the practice at your business. But when you choose a solution like FindTuner, that’s not the case.

FindTuner’s search merchandising capabilities enable anyone to define the criteria that power personalized search. Whether you want to promote important products, tailor the search experience to different types of customers, target certain shopper attributes or all of the above, FindTuner’s low-code/no-code approach makes it easy to do so.

Learn more by requesting a demo.

Large language models (LLMs) often feel like having the world’s greatest toolkit for fixing a problem you can’t quite identify. This gap can feel particularly acute when working with search indexes such as Elasticsearch, Solr, or OpenSearch. Existing search technologies are exceptional at making rapid textual matches but are unable to match documents without direct text matches.

Consider the Humble Search Bar

Finding the best search results often means going beyond what users put into the search bar. When considering a search bar, it becomes clear it inherits context. In other words, a search bar on a web site for purchasing auto parts should seek ways to apply an auto parts perspective to whatever the user enters. You shouldn’t expect the user to input something like “I’m searching for an auto part called a bumper.” Moreover, people often don’t have quite the right word to describe what they are looking for. Sometimes, they may have the right word, just not the word matching key results that would best serve them.

Large Language Models (LLMs) can bring the power of contextual inference into the search bar. A simple way to leverage that power is to use it for identifying and surfacing potential synonyms. This blog will discuss one simple approach for managing this process, considering limitations in how LLMs function as well as the volume of queries and potential synonyms that need to be considered. Human oversight will be considered a critical component necessary to avoid irrelevant or counterproductive synonyms. Due to this necessity, an overarching goal is to make human interaction as efficient and effective as possible.

A Quick Primer on LLMs

By using context inference, LLMs can make associations beyond direct text matches. In the LLM data, words is represented in a sort of coordinate system with hundreds, or even thousands, of dimensions. This coordinate system effectively creates innumerable conceptual vectors. There’s no way to know what concepts are represented in the data. They could be as fundamental as “gender”, as weirdly specific as “astrophysics”, or as nebulous as “things people think about at 3pm”. This is generally processed through a mechanism known as cosine similarity. Visually, the word “signal” might be thought about like this:

In this example, the red vector (1) might represent “things related to radios”, the green vector (2) might represent “cars”, and the blue vector (3) could be “astrophysics”. The word “signal” has a relationship to “wave”, “blinker”, and “pulsar” in each of these contexts, respectively.

Finding the Right Perspective

The diagram above is a limited representation of generalized word embeddings from LLMs such as Google’s Gemini or Meta’s Llama. They are both built from massive sets of data with no particular focus. However, as a user of one of these LLMs, a search for “signal” on the aforementioned web site of an auto parts store would ideally just utilize vector 2. Similarly, for a search on scientific papers, vectors 1 and 3 might both be applicable. As a search professional, the goal is to make use of vectors best match the user’s perspective when they fill in that search box.

Is This a Problem?

It might not seem like this is a big deal. After all, there’s no car part called a “wave”. Unfortunately, it’s possible the store could sell a window cleaner called “Bright Wave” or a product description for a window flag might include the text “waving in the wind.” A search for the term “signal” could conceivably return window cleaner and sports team flags when the user most likely desires a replacement blinker. In many ways, these represent a typical synonym issue in Solr, Elastic, OpenSearch, and other BM25 based tools. Since synonyms apply universally on a field, it’s important to find that sweet spot where documents relevant to a user are returned without generating too many false positives.

Developing a LLM to Search Synonym Pipeline

Since you never know what matches an LLM might return, validating the synonyms before implementing them is critical. In a large commerce site, you might have tens of thousands of words users search on and hundreds of thousands of potential synonyms for those words. Clearly, making this task into something a single subject matter expert or search professional can manage will take some planning.

Scoring!

The simplest mechanism to reduce the potential synonyms a human needs to consider is to look at the vector search score. All LLM based searches are done with vector searches. These kinds of searches generate a score based on distance or relevancy associated with each potential synonym which is always used to determine sort order. Each vector search will return exactly the number of values requested because, technically, every document in the vector database is a “match”, just with increasingly bad relevancy.

Let’s consider a log with 250,000 historical user searches. A search team could go through this log trying to think of potential synonyms that may find suitable results in a product catalog. Assuming they are doing due diligence and averaging roughly 15 seconds on each search, that log will require about 1000-man hours to process. So, the first thing to do is limit the maximum number of potential synonyms returned for each user search to something reasonable, like 5. Now the search team doesn’t need to try to imagine possible matches, they can simply look at the 5 that came back for each query and think “yes” or “no”. Less thinking means less time, this alone could shave off a solid third of the time.

That still leave about 650-man hours, but what if 5 options are just the maximum. After all, not every word searched by a user will have 5 words that are even in the ballpark to be used as a synonym. In fact, having 5 suitable synonyms is rare. It’s far more likely that no suitable synonyms are available. Just because the LLM thinks one word is more like the user’s query than any other word doesn’t mean the closest word is even worthy of consideration. This is where applying a threshold value on scores can drastically reduce the possibilities a search team needs to consider. Even a reasonable threshold value could reduce the word/synonym pairs up for consideration down to less than 5% of the total “matches” made. Now the 650-man hours becomes a much more manageable 30-to-35-man hours. This is the kind of effort a first-time evaluation might require, but iterative cycles might just take an hour a week after common synonym pairing have been evaluated by the search team.

Reduce Potential Synonyms

Another way to improve LLM results is to exclude words that should never be used as a synonym. It doesn’t matter if the LLM thinks “wave” is a good match for “signal” if the vector database doesn’t contain the word “wave” as an option. Pruning potential synonyms could be done automatically as a function of how potential synonyms are gathered and filtered before being added to the database.

For instance, a good way to determine a pool of potential synonyms is to draw them from a description field for products. However, description fields contain a lot of terms that LLMs consider highly interchangeable. This is notoriously true for numerical indicators. Setting up a process to block numerical words (such as “one” or even the actual number 1) removes the risk of it being offered as a synonym entirely.

Custom Source Embeddings

This is a little technical, so the details won’t be discussed here, but an LLM can be used to generate a reduced set of data based on an overarching phrase or concept. If a search is very specifically for auto parts, the LLM could generate a smaller model that is tuned based on that idea. A well-constructed phrase will create a much better set of embeddings to base the searches on, which will improve score results and make the threshold filtering more effective. However, developing a good set of embeddings is a non-trivial task.

In Conclusion

LLMs can bring suggestions to the attention that may have been overlooked. However, some human intervention will be required to ensure synonyms are not added that will adversely affect user experience. The key to a useful implementation is to reduce the amount of user searches and potential synonyms a search team needs to evaluate. Instead of 250,000 searches with an undefined set of possibilities, a good system may have the team needing only to consider a few hundred legitimate pairs in a more readable format. It may be helpful to create your own synonym management application based on these principles or take advantage of an existing product such as our FindTuner application. Good luck on your own path in turning an LLM into a synonym generating machine!