54 million
That’s the number of search queries Google will serve in the time it will take you to read this article. That’s about 88,700 searches per second worldwide. And we aren’t even considering the hits from half a dozen other search engines millions of people use. With the sheer volume of content being created, search engines like Google, Yahoo, and Bing continually index tens of billions of websites every day, so that internet users like you and I can find the most relevant content when we need it. What’s important when you’re searching?
Historically, search engines have always provided content that matches what you’re searching for by sorting results by relevance. Much like how searching for an item on your favorite shopping site did.
And this used to be enough. Until now.
Site search is an important navigational tool for eCommerce websites, giving shoppers the ability to take shortcuts to products they want. For most eCommerce sites, the search functionality is a part of the website’s global navigation and is integrated into its wider header design.
eCommerce search systems aren’t very different from Google. Some common traits eCommerce search share with Google search are:
- Keyword match — matching your shopper’s search query to keywords within a catalog using a stemming algorithm helps the search engine suggest related/similar results.
- Faceted search — this allows shoppers the ability to refine a selection of search data with attributes such as ‘sort by’ and ‘color’. Data is sorted and grouped into tags.
- Best first — based on the assumption that the shopper wants to see the most relevant items first, constructed using a relevance score.
- Linear pagination — this option is used when the search data won’t fit on one page, such as items in an eCommerce site. The product listing page can be designed to list products via inline paging and progressive loading.
- Autocomplete/autosuggest — this allows a shopper to save time by either filling in a partial query or suggesting or populating terms after the shopper has typed an initial amount of characters into the search bar.
Relevance
Searching for a product on an eCommerce site can be overwhelming. While 30% of online shoppers use the search bar to locate a product, the percentage of stores that use data from search results is a dismal 7%.
Consider your ranking strategy based on the search term. Sites like Amazon sort by relevance and choose to list products that match the keyword closely.
Often shoppers know what they are looking for, but they don’t know how to go about searching for that product. They don’t know what it’s called and what they will need to type in to locate that product.
For instance, smartphones are fairly easy to categorize, but there could be millions of other products that are listed in categories that are not easy to guess. Your singular aim should be to help shoppers find what they are looking for as quickly as possible with as little aggravation as necessary.
Search algorithms on your site help shoppers construct search queries intuitively using a combination of autocomplete and predictive search.
But then people use different terms and expressions for different things; so what I call ‘trousers’, you could call ‘pants’. Your webpage should be intuitive enough to understand synonyms and provide the right results no matter what your shopper types in. By recognizing synonymous keywords, you also avoid leading shoppers to the dreaded zero results page. When search queries come back with results that are irrelevant or with no results at all, shoppers may feel defeated.
Sometimes, search queries will have inadvertent typos. Research reveals between 10% and 15% of all search queries are misspelled. An intuitive search engine will return results that are autocorrected. This avoids leading the shopper to a dead end and forcing them to try again.
Synonyms look at the search keyword and offer alternatives based on what that query could be. You can analyze the keywords that gave you a ‘zero results page’ to offer alternative suggestions that are closely related to their desired product, implement fuzzy logic to determine whether their search term was incorrectly spelled and recommend corrected terms as a suggestion. By doing this, you are able to match the intent of the shopper by displaying the most probable list of relevant products as helpful suggestions. This helps you seize a lost opportunity and fix an otherwise unsatisfactory experience.
How can I help my shoppers find what they are looking for?
Research found that only half of the searches on about 500 websites were successful.
Once you figure out what the shopper is searching for, you need to offer options to refine and sort results to rid your search results page of the clutter.
Sorting or narrowing down the list of products can help your shopper understand the way your search results page works and make the experience more enjoyable. Sorting rearranges results by specific preferences like, ‘Price — low to high’ or ‘Most relevant’.
On the other hand, Filters allows shoppers to limit the number of results displayed and reduce clutter like, ‘Maximum/Minimum price’, ‘Brand’, ‘Size’, and ‘Color’.
This brings us to how important usability is when designing an eCommerce search engine. If your shopper can’t navigate through your search results page, they won’t stay and worse, they won’t return.
An interesting instance of how search engines improve usability is when the results page highlights items currently in stock or only available in-store with contrasting colors.
Personalization
We want shoppers to feel unique, special and understood. Your eCommerce site should match or exceed the shopping experience a brick and mortar shop could offer.
The results page is golden: it gives merchandisers and product managers an opportunity to engage shoppers in a dialogue that meets and exceeds their expectations.
Even when shoppers are familiar with the search feature, having your site store recent searches will help them save time and effort.
By feeding your search algorithm with data generated from shoppers in your store, your search can continuously learn, adapt, and improve.
Search engines are designed to search by matching queries to keywords. Referred to as Query Understanding, this is the backbone of an effective search system that aims to establish the shopper’s intent.
What can I do to personalize my search engine?
Shoppers have evolved to search for products using a more conversational tone. For example, shoppers nowadays search for ‘Nike shoe for teenagers’.
As a result, search engines need to learn and understand human conversations to match contextual requests.
Most modern eCommerce sites don’t offer personalization for search. Because rudimentary search algorithms rely on the information within cookies stored in the local machine, results lack sufficient knowledge about who shoppers are and what they prefer. While cookies may know your recent shopping habits, your eCommerce store won’t have a clue about who you really are.
Now imagine using a search engine that knows your shopper well. Think of how different search results would be.
Using AI and ML, search engines can be personalized to make shopping less of a chore and more functional.
Ultimately, a successful search system should be able to learn and interpret shopper’s intent to display results that appear intelligent and informed.
Autosuggest
What kind of product recommendations do you want your shoppers to see when they land on your eCommerce site? How would you like them to see the right product suggestions for their queries? Suggestions are a key part of any product discovery system. The fundamental purpose of a good site search system is to help your shoppers find what they’re looking for. So, one of the ways to achieve this is to include an effective product recommendations system within your search engine. They not only help shoppers find items but also choose the most relevant search terms. Traditional recommendation systems are made up of a static list of product suggestions for specific ‘anchor’ products. How can we improve the quality of our suggestions? Suggestion systems depend on the type of business your eCommerce site is based on, for example, suggestions will be different if your store sells computer hardware as compared to fashion. When I typed in ‘Mo’ to search for Monitors on Amazon, I got suggestions like ‘Mouse’, ‘Mosquito Killer Machines’, ‘Mobile holders’, ‘Mop stick rod’, and ‘Mouthwash’ among others, even when I was searching for LCD Monitors in the same session. We can improve the search if we avoid partial suggestions when a threshold is not defined or when we don’t have sufficient information to build helpful suggestions. A good data analysis is the first step for a good suggestion system. A systematic study of search terms and the action that follows can possibly help you predict the shopper’s intent based on the search terms they use. For instance, the search engine should have picked up on my intent to buy a computer monitor and recommended only computer monitors. Armed with knowledge like this, merchants could have used sophisticated merchandising algorithms to determine the order of results. The right suggestions have the potential to influence a shopper’s purchase decisions just by virtue of the fact that your site is suggesting them.In Pursuit of Search Happiness
In a nutshell, eCommerce search is a way to get what we want when we want it. But when it comes to shopping online, not everybody is ready to buy the first thing they see because they may still want to evaluate whether it serves their needs. Online shopping is popular because of its sheer convenience. Shoppers can search for anything from virtually anywhere. As an online merchant with a large catalog, getting search to work right is especially important.
Search data is a goldmine of information for buyers, product managers, and merchandisers and help them answer two important questions:
- Who are my customers?
- How can I sell more of my products?
