E-commerce site search has the power to boost visitor experience, build customer loyalty, and grow on-site conversion rates. According to one study, visitors who use search can generate around 30-60% of all ecommerce site revenue.
Despite the potential impact of improving on-site search, only 15% of companies have resources dedicated to optimizing it.
E-commerce site search is complex. In this article, we’ll talk about some of the challenges for product search, opportunities for improvement, and solutions available.
Who is this guide for?
We wrote this guide for brand, retailer, and marketplace professionals at B2C or B2B companies. Any kind of e-commerce business with a sizable product catalog can benefit from high performing site search to provide a better customer experience.
Amazon.com and site conversion rates
On-site conversion rates hover at around 3% industry-wide, but Amazon.com enjoys a conversion rate that is five times industry average (it’s even higher for Prime members). Search is vital at the world’s largest marketplace for finding anything, so naturally Amazon invested heavily in search engineering for 20 years and today has 2000 people working on site search.
Whether you’re a marketplace, retailer, or brand that competes directly with Amazon, there’s good news: You don’t need to hire 1,500 search engineers. Today’s off-the-shelf e-commerce search engine technology can give you a leg up without breaking the bank.
The modern e-commerce stack
In 2021, about 50% of retailers said they planned to spend more time developing their own site search capabilities.
Newer search solutions can replace the default search engine that shipped with your e-commerce software, or allow you to build search from scratch on a new headless commerce platform. Either way, search engines can easily be plugged into a highly interdependent tech stack that may include some combination of:
- E-commerce platform / CMS: the core engine of your customer-facing e-commerce solution that helps manage everything from product page design to customer ratings.
- PIM: product information management system for managing and sharing content between systems.
- Tagging: product tagging solutions can enrich medata for more accurate search.
- Inventory: inventory management is often handled with an ERP solution.
- CRM: whether you bought or built your CRM, you need a central place to store your customer information across all touchpoints and interactions.
- Payments: an online payments processing solution.
To be effective, search needs to work across the different systems that make up your e-commerce business. It needs to be able to search across your product catalog, check your inventory for out-of-stock items, leverage visitor characteristics from the CRM, and display updated pricing information. Plus, it needs to do all of these things in milliseconds.
The challenges of e-commerce search
Successful e-commerce search requires a multi-faceted product and approach. Below, I’ve outlined some of the bigger problems that retailers face along with the kinds of solutions broadly available today.
As Baymard Research points out, even many of the largest retailers and brands in the world still haven’t solved some more basic search challenges like spell checking. There are many other basic and advanced capabilities required for successful search. We touch on some of these topics here, and we have other guides you may be interested to read such as our 12 Tips to Increase E-Commerce Search & Discovery Conversion Rates.
With that brief introduction, let’s take a look at some areas you might consider as you hunt for a new e-commerce search platform.
E-commerce search engine features
The modern e-commerce stack is multi-layered and includes systems for payment processing, returns, inventory management, product information management (PIM), headless commerce, and more! Providing high-quality, high-converting, up-to-the-minute content with search results delivered instantly is tough act to do well. We've compiled a list of features that companies should consider — and ask their vendor about — when selecting a search solution.
When rolling out a new product, collection, sale, or campaign, time is of the essence.
We met with the head of e-commerce for a fashion company recently. She told us their brand new site search solution — implemented less than 12 months earlier — takes 30 minutes or longer to update whenever they added new products or updated existing listings.
Worse still, some systems take up to 24 hours for the index to refresh.
With instant indexing, any new content you publish or update on your site will be added to your index — or search collection — immediately. Similarly, any time you delete content or mark a page with a valid 404 error code, a search engine should automatically remove this page from your site or collection.
Search engines will index a site in one of two ways: either through a web crawler or via API. Crawlers work well for smaller, static websites, but APIs are the way to go for e-commerce tech stacks where data is coming from a variety of systems and needs instant updates.
While many search providers will claim they offer instant search, it’s worth doing some research; the only real way to know exactly what it can do is to take it for a test spin.
Artificial intelligence (AI)
There are many different kinds of queries that search engines need to contend with including:
- Symptoms “fix a drip”
- Broad searches “hammers”
- Specific product name or brand name “Anderson retractable screen door”
- Category searches “garden equipment”
- NLP queries “¾ in pvc fittings”
- Synonyms “resin vs epoxy”
Relevance — trying to match the user’s query with the best results — is tough. Many retailers write dozens of rules and synonyms to help match search queries with the right content. This can work to some degree, but it can be extremely difficult, time consuming, and expensive to maintain.
AI learning models known as “vectors” are used to encode datasets into mathematical representations which can be used to deliver exceptionally relevant results.
Vectors actually eliminate the need to write synonyms or create rules. By comparison, keyword searches (how most search engines work today) don’t work efficiently when queries are vague. Vectors offer a way for e-commerce search engines to leverage AI to find closer matches in less time. However, vectors have a problem: they're slow and expensive. Fortunately, there's a solution in Neuralsearch which accelerates vectors massively while also offering the power of keyword-based search.
In 2019, if you searched for “face mask” on Amazon.com you would mainly get a list of anti-aging creams and night masks. In 2020, to be effective that same search would need to include N-95 disposable face masks to protect against Covid-19.
The ebb and flow at which new products are introduced or new trends (fidget spinners, anyone?) come into the market make it essential that retailers have a solution that’s ready to go.
Signal boosting is the process by which search engines leverage user behavior such as clicks and conversions to optimize search result ranking. As customers do more searches, machine learning models can use the signals (clicks, conversions) to automatically rank content.
Learn-to-rank and reinforcement learning are two techniques that help search solutions adjust result rankings continuously and automatically.
Even though you can’t always predict the next trend perfectly, a search engine with machine learning can help by making adjustments for you.
Personalization and merchandising
Brick-and-mortar merchants have learned how to maximize every inch of real estate — from the aisles to the fitting room to the check-out line — where similar items and packaged deals can be sold.
It’s different for online sellers. For a while, recommendation solutions gained some popularity to help retailers suggest similar products to customers. If you like X, you might like Y. The problem is that these recommendations were mediocre, un-personalized, or way off the mark.
Up to 43% of shoppers will use your site’s search box, so personalizing results and showing related products can be a better way to merchandise.
Personalized search includes search results that are tailored to each individual based on their profile, which can include past search history, purchase history, brand preference, location (geo), product ratings, gender, and more.
Personalization starts with data. The more demographic and psychographic data you can collect about your customers and visitors, the more sophisticated your personalization can be.
Search metrics and testing
It’s not always easy to know which search terms and queries are moving the needle. Even if you have the list of most popular keywords and search revenue numbers, what do you do with that information?
A separate but related problem is: how to know whether the changes you’re making to search results are working?
Most businesses and e-commerce websites run Google Analytics, and the data they contain about your online store is very valuable, but it’s not always helpful.
Most third-party search solutions should offer some kind of search analytics. Daily, weekly, and monthly search trend information can help you adjust search results and know in what areas to invest. Ineffective searches can help you identify no-result searches and what searchers weren’t able to find.
But there’s another way: A/B testing. Some of today’s e-commerce search platforms offer split testing to help you to determine which search algorithm delivers the most relevant results and highest ROI.
Armed with this data, you can focus on search optimization to provide a better overall user experience for your store.
Amazon has shown that millisecond lags can cost millions of dollars in lost revenue. E-commerce store search needs to be fast to provide an optimal user experience and reduce site abandonment.
When it comes to product search, speed matters. For ideal usability, it’s best to:
- Add autosuggest (also called autocomplete) to the search bar to display results as users type their queries
- Offer instant search (with product images)
- Dynamically refresh results as users select different filters and facets (while displaying the list of selected filters so people can uncheck their options too)
All of the above assumes you have a search solution that is able to deliver results in milliseconds — even on Black Friday.
Easy-to-configure responsive UI
Just 68% of shoppers say they’re satisfied with the experience when searching on a retailer’s mobile site or mobile app.
Mobile commerce is only getting bigger and while many retailers have built responsive websites, search hasn’t evolved at the same pace.
Many search providers now offer libraries to design mobile front ends with a mobile-first approach. Mobile search can still be full-functioning with features such as autocomplete (or predictive search), instant search, product previews, filters, sorting, and more.
Top 10 e-commerce site search solutions
Here are a number of platforms you may choose from. It includes SaaS-based search and e-commerce platforms that integrate search. This list is far from exhaustive, but it should provide a good sense of the range of solutions available.
Search.io was built from the ground up with AI at the core to offer powerful and easy-to-configure search without the complexity of traditional search engines. Try Search.io free for 14-days.
Searchnode is a European headquartered, fully-managed e-commerce search solution. It offers customers an easy-to-update index to adapt search to their site.
Bloomreach is an e-commerce platform for building headless commerce solutions with search built into the core product.
Searchanise is ideal for smaller stores and it’s the highest-rated third-party search engine on the Shopify App Store.
Headquartered in Finland, Klevu is an e-commerce platform with AI and NLP technology.
Attraqt is a headless commerce solution that recently acquired FredHopper search to offer alongside additional e-commerce solutions.
Originally developed as a mobile search engine, Algolia has expanded its capabilities to cover AI-powered site search and e-commerce search.
A new entrant to e-commerce search, Wizzy offers a Magento search extension alongside its own search solution.
Adobe’s original e-commerce search solution, Adobe Acquire and Connect, has been put out to pasture and customers will be encouraged to consider Magento.
Prefixbox is an AI-powered e-commerce provider with a modular service for adding e-commerce features.
- Blog: How to get a competitive edge with smart (and surprisingly easy) e-commerce site search solutions
- Tutorial: Advanced e-commerce search in 10 steps
- eBook: How AI Can Improve E-commerce Search Conversions