The Ultimate Guide To Increasing Conversions Through Site Search
32 Minute Read
What is the goal of this guide?
This guide is designed to help online retailers of every shape, size, and industry to increase their conversion rate and revenue by increasing the number of visitors that use search, and making the search results more relevant.
Why optimize for search?
The neverending quest for better conversion rates is often frustrating, and can waste a lot of time if you’re not calculated in your approach. Many retailers spend hundreds of hours every year optimizing colors, buttons, ad copy, landing page layouts, and checkout flow. Of course, all of this is necessary, but at a certain point, further optimizations won’t produce meaningful and long-term results.
However, optimizing for search conversions is a worthwhile pursuit, and represents the single most undervalued optimization strategy available to ecommerce retailers today. Our research indicates that even major online retailers like BestBuy, Walmart, Sears, and Amazon could benefit from making seemingly minor adjustments.
So why is search largely ignored?
Few retailers see search as a priority because, in the world of ecommerce, less than 10% of site visitors utilize search. In general, it wouldn’t be wise to spend much time optimizing for such a small subset of users, but search is special.
What makes it special?
Search visitors are simply more valuable than other types of visitors. While the average ecommerce conversion rate is just 1.72% for non-search visits, visitors that interact with search at least once have a conversion rate 3.5x higher at 5.98%.
What is the overarching strategy?
The scope of this guide is to show you how to increase the number of visitors using search, optimize the search & autocomplete UI, and improve the relevancy of the search results that follow.
On average, visitors who use search to navigate an ecommerce site will convert at a rate 348% higher than those who do not interact with search. Search visitors also produce UPT (Units Per Transaction) figures 38% higher than non-search visitors. To be fair, part of this has to do with the intention of the visitor. However, at the end of the day, visitors who use search provide a precise input that tailors the shopping experience to match their needs and interests.
Convincing more users to search will automatically increase the relevancy of their personal shopping experience, leading to better engagement and conversion. While it’s more difficult to measure, customer lifetime value will also increase as a result of the fact that this makes the shopping experience more intuitive and enjoyable.
Of course, increasing the number of visitors who use search won’t help much if the search results are irrelevant. Search relevancy is one of the biggest pain points for online retailers, particularly those using out-of-the-box solutions from major shopping cart platforms. It becomes important, then, to find ways to make the search experience better for your shoppers.
This guide will walk you through the above strategies, and should result in drastic increases to the volume of visitors using search on desktop and mobile, and will then help you to improve the search experience for your shoppers.
How to use this guide
Many of the optimizations in this report will need to be tested for optimal results. The facts and figures we provide are averages, and may not work exactly as outlined in every industry.
With that in mind, it is important to formulate a plan on how you will test proposed changes. While it’s not the scope of this guide to teach split-testing best practices, you’ll want to follow these basic principles:
- Test one change at a time.
If you make multiple updates simultaneously, you will not be able to analyze what’s working, and what’s not.
- Provide a large sample pool.
While it may be scary to drive a large portion of your visitors to untested designs and layouts, it is necessary to get useful data. Provide a pool of least 1,000 visitors.
- Segment mobile/desktop visitors.
Mobile visitors and desktop visitors will behave very differently. Design and layout changes should be optimized for these segments separately.
- Segment paid traffic and organic traffic.
Paid traffic visitors will generally behave differently on your site.
- Segment new and returning visitors.
Returning visitors may be more familiar with your previous design, and may react differently compared to new visitors.
There are many resources available online that can provide additional best practices regarding split-testing, but these are the most important and most relevant for ecommerce. Start here, and branch out if you can.
Chapter 1: Understanding Search Psychology
Why do people search on ecommerce stores? The answer is ultimately fairly simple; they want to save time. The act of locating the search bar, entering a search query, and then locating the correct set of products must be the fastest path to those products if you want to satisfy the needs of these visitors.
In addition to saving time, search visitors have come to an ecommerce site with a specific objective or goal. They are not “window shoppers” who are just killing time and waiting for something to pique their interest.
That said, all search visitors are not the same. There are multiple types of search queries which are important to understand. The psychology and intention that motivates these different types of queries should inform design decisions, and how search results are ranked and organized. It’s also important to support these types of queries from a technical standpoint. What are the various types of searches?
Type 1: Broad spectrum
The first type of search visitor knows that they are looking for a product within a broad category such as “laptops”. They know they need one, but they don’t know the brand, size, CPU, storage type, storage size, etc. Depending on how broad the category, it may be just as simple for them to navigate using the main navigation menus. This may be the case with “laptops” but perhaps not with something like audio receivers on an ordinary consumer electronics site.
What is the psychology of this user? This type of shopper is somewhat motivated to make a purchase, but may still be in the research phase. They may know almost nothing about the products in this category and will have questions such as:
- What is the average price of these products?
- What types of these products exist?
- In what styles are these products available?
- What features and configurations are available?
- What is the price for a premium version of this product?
- What is the lowest price for a version of this product?
- What sacrifices will I make if I purchase a cheap product in this category?
- What will I gain if I purchase a premium product in this category?
- Do these products have warranties?
- How long will this product last?
The questions that arise may have more to do with the industry than the shopper’s personal savvy, but these are the types of questions they will be asking themselves.
Most search visitors fall into this category. With that in mind, it’s important to answer these questions as quickly as possible to keep these visitors engaged. The concerns of these shoppers should inform the design of the search bar, autocomplete, and search results pages. Specifics will be discussed on how to do this in chapters four and five.
Type 2: Precise searches
Whereas a broad spectrum search visitor would ask the search engine for a “laptop”, a precise search visitor might ask for “Late 2016 MacBook Pro 15 inch”. These precise searches are usually product names or even product numbers or SKUs.
What is the psychology of these visitors? Unlike the broad spectrum search visitor, these visitors are not in the research phase. These visitors know exactly what they want, the price they are willing to pay, and are generally ready to buy. If they find the product at the right price, and your store meets the rest of their expectations (shipping price, shipping speed, return policy, warranty, etc.) they will make a purchase. These visitors are the reason why search conversion rates are so much higher than non-search conversion rates.
Catering to the habits, wants, and needs of these visitors is also extremely important. Their primary concerns are:
- General availability
- Color/configuration availability
- Item condition (new/used/refurbished)
- Relative ease of checkout experience
- Shipping speed
- Shipping cost
- Return policy
More will be discussed on how to cater to these types of visitors in chapters four and five.
Type 3: Exploratory searches
Exploratory searches fall in between type 1 and type 2. Psychologically, the user has the same intentions as type 2 users, but doesn’t know how smart the search engine is. So rather than risk wasting time with a precise search, they search using broader terms with the intention of using filters, sorting, or autocomplete suggestions to help them narrow the results.
As an example, rather than searching for “Late 2016 MacBook Pro 15 inch”, a type 3 search user might instead type “MacBook Pro”. Depending on the site, they might see autocomplete product suggestions that would help them locate the right product. If they don’t see suggestions, they’ll generally complete the query and narrow results using the available filters and sorting options. In this case, the user might sort by “newest” or “price: high to low”.
Type 4: Use case/symptom searches
This type of search allows the user to describe the reason they need or want the product. So, rather than searching for “MacBook Pro” they might search for “work laptop”, or “video editing laptop”.
While this type of search isn’t common in every industry, industries such as health & fitness see as much as 40% of their searches fall into this category. For example, “sore throat” is a common search on many health and pharmacy websites.
Whether or not this is important enough to devote resources should be researched. Examining common search queries via Google Analytics can help you identify and classify the types of searches that are most frequently happening on your site.
Type 5: Feature searches
Feature searches are queries that allow the user to specify further details about the product that they need. For example, a search for “UHD TV”. At the same time, it can also allow the user to describe other needs that the user might have that wouldn’t traditionally be thought of as features. “$500 laptop” is another example of this type of search.
These searches are used fairly commonly in most industries. In apparel, an example of a feature search could be a search for the material that the shopper wants (such as “wool sweater”). These users are typically fairly familiar with the product they want, and how much they’re willing to pay. Psychologically, they’re much more likely to make a purchase than someone using the broad spectrum type of search.
Chapter 2: Underlying search technology: How it works, and where it’s going
Search technology was originally developed decades ago to search large databases of files.
All the open source search engines in use in ecommerce today are based on Apache Lucene. This search algorithm is a full-text index, which means that it builds an index for each word in a document. This was originally developed before the internet existed, and as such, it has had to evolve over the course of multiple decades in order to continue to be useful and to expand its reach beyond the office, and into homes and personal devices.
Solr is one such evolution of Lucene. Solr was developed by CNET in order to power faceted navigation. It’s also much easier to deploy, and has strong multi-language support. It’s flexible in terms of how relevancy is determined, but requires development expertise in order to be implemented.
Like Solr, ElasticSearch was also built on top of Apache Lucene. ElasticSearch is open source, and can also be modified for many different applications. Heavily modified versions are currently in use by Netflix, Facebook, and other large media companies.
All of these technologies are designed to search databases. They are not specialized tools, but rather, provide a general search framework that can be built upon for more specific use cases. These technologies have been adopted by popular ecommerce shopping cart platforms, and modified to support the basic functionality needed to search those databases.
Do these technologies go far enough?
Large ecommerce platforms such as Magento, Shopify, and BigCommerce use modified versions of these open source technologies to enable generally useful search engines on ecommerce stores. Inexpensive ecommerce plugins available from the Magento, Shopify, and BigCommerce app stores are generally built upon these same technologies but provide additional controls. But have they gone far enough?
Unfortunately, the answer is no. Building a specialized search tool is extremely resource intensive, and for that reason, major shopping cart platforms have found this existing technology to be “good enough”. This is likely to change at some point in the future, but at present there are serious limitations to what these technologies are able to provide. The diverse array of stores that are served by these platforms is also a complication as more precise algorithms would require tuning to work well for each individual store.
As consumer behaviors evolve, these algorithms would also need to be refined. Even Google provides frequent updates to their algorithms to help search results become more natural and relevant. This too, is very expensive today.
What are the limitations?
At current, all ecommerce platforms use basic algorithms to match user queries to products. These algorithms match keywords in a user search query to keywords in a predefined set of product fields. The more frequently these keywords appear in the product fields (title, description, image alt text, etc.), the more relevant the search algorithm deems that product.
Right now, these basic technologies are not built to support the 5 major query types. They simply look for matching words. While broad searches may yield somewhat relevant results, symptom use/case searches are unlikely to return any relevant products.
Some platforms provide controls that allow retailers to determine which fields are searchable, but very little control outside of that is available. The same goes for many third-party plugins. While they can improve matters with custom weighting, the underlying algorithm is a hurdle that generally requires resources and upkeep.
This can create serious problems for relevancy which are very impactful to retailers in most industries. In the electronics industry, for example, a search for “4K TV” can return results for all kinds of items with either of those keywords. Blu-Ray players, TV stands, universal remotes, computer monitors, Blu-Ray discs, digital movies, and much more will all be mixed in with relevant products.
In another example, a basic search for ‘red nike running shoes’ will return results for any products that contain the words “red”, “nike”, “running”, and “shoes”. So, instead of seeing a product that matches their query, the shopper will generally see a collection of shoes, Nike branded items (which might include gym bags, shirts, etc), and other items that contain these words in their searchable fields. While some custom weighting can improve the relevancy of the search results, it can never completely fix the innate problem that the algorithm does not understand what the shopper wants.
Why? The search engine doesn’t understand the relationships between the separate words in the query. Instead, it does a separate lookup for each and every word in the user’s search.
While most major ecommerce platforms provide manual controls for managing synonyms, most don’t go a good job automatically. This is another area that creates serious user experience challenges.
A common synonym issue is that of measurements. There are many ways to type out “inch”, and if each variation is not in your product data, the search engine won’t find results, and won’t return expected products.
What does this weak support for basic query types mean to retailers? Based on comparisons between ecommerce retailers with simple keyword-matching search engines and those with custom solutions, these irrelevant results can result in as much as a 20% loss in annual revenue.
Chapters four and five will discuss methods that can help retailers increase engagement with the search bar and improve out-of-the-box relevancy.
Where is search technology going?
Even today, custom solutions provide additional layers of intelligence that work to make search results more natural. While there are many algorithms at play, the most popular are machine learning algorithms that examine user behavior trends to understand which products are most relevant. For example, these algorithms analyze user paths after searches for “4K TV” or “red nike running shoes” and can see which products have the highest engagement and purchase rates. While products with extremely low engagement (clicks, page view time, add-to-cart, conversion rate, etc.) will be eliminated from results, those with the highest engagement rates will be displayed first. This intelligence is also applied to autocomplete, product recommendations, and other pages.
In addition to the automated aspect of these custom solutions, merchandising controls are also becoming increasingly popular, allowing retailers to manually sort products on search results, category, brand, and promotional landing pages. As marketing campaigns become more responsive and sophisticated, these controls are becoming invaluable.
More advanced features are also on the horizon, such as options to apply custom boosting rules that automatically increase the visibility of products based on specific business objectives such as profit margin, or product seasonality.
Chapter 3: Web design philosophy: Why search is now the #1 priority
Over the years, ecommerce web design hasn’t evolved much. The basic philosophies and goals have remained largely the same. For smaller, lesser known brands, it’s generally understood that the primary goal is to help a user understand the Unique Selling Proposition (USP) as quickly as possible.
As such, home pages are generally used to drive home the USP for people unfamiliar with the brand, while providing navigation controls for those who know what products they want.
The rise of Amazon has caused widespread adoption of more advanced navigational features such as sorting and filtering. Amazon’s influence is so profound that these features are essentially a must-have for anyone not selling specialized products that aren’t available on that platform.
More recently, intelligent site search technology has become commoditized. This is largely as a result of the widespread availability of natural search experiences on personal devices. In addition to phones and tablets, hundreds of millions of home computers, game consoles, and even smart speakers allow consumers to control their home electronics and order products without pressing a single button. As a result, many consumers expect smarter search technology that understands them more intuitively.
While the industry is likely still years away from evolving to the point of equivalent technology living on individual stores, shoppers want to see the right products as quickly as possible. 40% of ecommerce website visitors will abandon a site within 3 seconds if the page does not completely load, and forecasts show that this percentage will likely increase rapidly. That said, the expectation that products can be found nearly instantaneously is also on the rise.
This has motivated many ecommerce retailers to prioritize search in their designs, making it easier to find and use.
The discovery of “choice overload”
Several years ago, many online stores prioritized “easy” category navigation, providing quick access to the most popular categories. Far less emphasis was placed upon search at that time. However, in the last decade, researchers discovered what’s referred to as “choice overload”, wherein an individual has difficulty making decisions due to the fact that too many options are available. Entire books have been written on the subject, such as The Paradox of Choice – Why More Is Less by American psychologist Barry Schwartz. Web designers and conversion optimization professionals all over the world have confirmed this, showing that fewer options leads to better engagement in most cases.
Staples, for example, now uses search as the primary navigational component. They even feature a search bar within the category navigation menu.
This simplification of the UI has increased user engagement, and has been adopted by most of the top 500 internet retailers. It is expected that adoption will continue to spread and trickle down to smaller retailers as more intelligent search becomes more accessible.
Removing friction from the buyer journey
In addition to simplifying navigation, it’s long been known that adding steps to the buying process (this is known as “friction”) dramatically reduces conversion rates. Navigating through categories to find the right product often adds several steps, increasing the risk that the visitor will exit.
In the last few years, the most successful online retailers have adopted simpler navigation, and an increased emphasis on the search bar for the reasons outlined above.
Notice the screenshot here displaying the BestBuy.com home page. While there are a number of controls and navigation options, the primary means of navigation are the search bar, and just four top-level categories.
Compared to designs they used in the past, this design drives far more users to search.
As mentioned in earlier chapters, site search isn’t just better for shoppers, it’s also better for retailers. The average ecommerce site makes 38% of their revenue from search visitors, even though those visitors make up only 5-7% of total traffic. Prioritizing search should, at the very least, be tested by every retailer with a large catalog of products.
Additionally, as advertising gets more expensive, optimizing the conversion flow needs to be a top priority for most retailers as it is twice as effective as increasing traffic. Search is often overlooked as a way to increase conversion rates for new visitors.
Chapter 4: Design Optimizations
There are many ways that search can be improved in order to help you convert more visitors. The easiest way is simply to improve the design of your search area with the goal of convincing more of your shoppers to use it.
Isn’t this complicated, expensive, and time-consuming?
In reality, the average retailer could spend only a few hours making a handful of design changes and see drastic results. One SearchSpring client, MXGear, volunteered to have their search UI critiqued in an effort to increase usage. What were the results?
Notice that their original design is not bad by any means. However, it’s difficult to locate the search bar as it blends in with other design elements in the header. We suggested only two minor refinements to their design which took only a few hours to implement.
In the new design, the search bar is much easier to locate, but still not intrusive. What were the results?
Search usage doubled.
Prior to the changes, 6% of monthly visitors were engaging with search, but after the changes were made, the number of search visitors increased to 13%. Since their search visitors were converting at a rate 3x higher than non-search visitors, this change brought immediate benefits in terms of sales and revenue.
Importantly, these design suggestions were not made on a gut feeling, nor were they guesses based on years of design experience. Rather, careful and thoughtful research informed these suggestions, and it performed exactly as the data suggested it would.
1) Use a search box
In order to create a minimal aesthetic, many websites began removing or hiding as many navigational elements as possible a few years ago. By and large, these changes are beneficial as they eliminate unnecessary distractions. However, the search bar suffered as a result, and was often hidden under a menu, or replaced with a small magnifying glass icon.
The icon may be more attractive, but it’s simply not as effective as an open text entry field in terms of helping visitors shop.
“The transition to the search icon is not as smooth and without perils as designers may want… The magnifying glass alone makes it much harder to locate the search.”
Exactly how much worse is an icon when compared to a search bar? The difference is quite a bit more drastic than we expected. In our research, we found that the average ecommerce site using an icon had search usage of just 4.44%. Sites that use a search box, on the other hand, had usage more than 3x higher at 14.14%.
Why is the difference so drastic?
For one, an icon occupies less visual space, making harder to locate when searched for, and easier to miss by everyone else.
Secondly, the open-text entry field is simply more inviting. Modern search icons behave differently on different sites. While some will slide open with a cursor hover, others need to be clicked. Still others will actually take visitors to a new page to search. These additional clicks (and possibly page loads) add subconscious friction that scares shoppers away.
This topic has been thoroughly researched and confirmed by web designers in recent years.
“When used without an open-entry text field, the icon takes up less space. Visually, it’s less prominent and, therefore, less noticeable. We don’t recommend the icon-only pattern for desktop websites. Icon-only search makes sense on mobile devices, because there’s less screen space and fewer icons and labels in general. But on desktop, there’s more to look at, and thus, it’s easier for the stand-alone search icon to get lost in the crowd.”
Still, an icon does have its place in certain circumstances. Mobile devices, for example, are typically held in portrait orientation, and have far fewer horizontal pixels. While an open search entry field will still outperform an icon in most cases, this could certainly come at the significant cost of hiding more critical elements.
2) Increase contrast
The example we showed of MXGear earlier in this chapter is a great example of why contrast is so important. A lack of contrast from other navigational elements camouflages the search box. While individuals that have the specific intention of searching are likely to find it, others are much less likely to use it.
There are different ways that you can increase contrast. It’s not just about the search box and the background color. The search box also needs to contrast other buttons, images, and links around it.
Again, the MXGear example is a great showcase.
In addition to contrasting with the background, the search box is in contrast to everything else around it. None of the other buttons have an accented outline, for example.
Customize the following elements to help your search bar stand out:
- Background color
- Button color
- Text color
For example, if your site uses a white background as most do, use a black or accented border to help it stand out. Including a search button with an accent color can also help it to stand out from other nearby elements.
Contrast is a bit harder to quantify, but we analyzed site performance based on their use of the elements above. Sites that utilized these elements (not necessarily all of them) did see significantly higher search usage.
This was much more impactful than we expected, and should certainly be tested. It should be noted that this contrast doesn’t have to be in conflict with the design language and color palette used on the rest of the site. Whereas low contrast search bars led to 8.78% search usage, those with high contrast colors and borders generated 18.82% search usage.
This extraordinary effect is achievable in only a few minutes by any web developer.
3) Optimize the text entry field
As was mentioned earlier in this chapter, open text-entry fields get a lot more usage than icons, but opening up the field for text entry isn’t enough. There’s a lot more that can be done to encourage usage.
Use a white background
In the previous section, we discussed using background colors to increase contrast, but that strategy should not be applied to the entry field in most cases. When an entry field is grey, that’s universally used as an indicator that text entry is blocked. If the background of your site is any color other than white, using a white text field is a great way to increase contrast.
But what if your site’s background is white? Won’t using a white text entry field decrease contrast compared to grey or another color? The statistics show that this doesn’t matter. Use other methods to increase contrast, but you should always use a white entry field.
Sites with entry fields that were any other color saw usage of 4.79%, while sites that used a white text entry field (even on a white background) saw usage of 14.57%.
Once again, MXGear is a great example of this optimization in practice. It’s easy to see from the before and after screenshots that the white text entry field made the text box not only more visible, but also a lot more inviting. People simply expect that areas where they can type will be blank or white.
Many sites have greyed-out text inside of the search box. Is this useful? Yes. We found that sites with sample text or CTA text (such as “search” or “start typing”) saw about 2% more of their total visits use search. 2% may not sound like much, and indeed this is a less impactful optimization, but is still worthwhile as it should be easy to implement.
It should be mentioned that we did not test or account for the exact words that appeared in the search bar. It’s reasonable to think that certain phrases such as “start typing” might be more effective than others, but there were simply too many variations for results to be conclusive with regard to which phrase is best.
Note: Since search icons do not have a text entry field until they’re clicked, any site that used an icon instead of a search bar or box was scored as not having a white background or CTA text.
4) Negative space
Negative space (or whitespace) is the empty space in between page elements (images, buttons, links, etc.). It’s not often thought about by anyone who is not a designer, but is extremely important. It gives the various elements room to breath, and makes them easier to discover.
When items are too close together, it’s harder for users to pick out specific items, making everything harder to find. Unfortunately, it’s far more difficult to implement than most of the other optimizations discussed thus far.
The best practice with regard to the search bar is to leave a large amount of space between it and the other nearby elements. The results of our study here were also surprising. Sites that did not use whitespace effectively (whether using a search box or icon) saw usage of 10.35%, while sites that gave the search area plenty of room to breath saw usage of 18.46%.
While harder to implement, this should be kept in mind the next time that your site receives a redesign.
The area where the search bar is placed can greatly affect its usage. There are some common industry conventions and related user expectations, however, so testing is suggested to determine the best placement for your site.
Desktop users search about 50% more than mobile users do, so it’s a bit more important to get this right. Research shows that the best location for the search bar is actually in the center in the header. Center placement resulted in 15.86% search usage, whereas top right delivered 13.43%. Top left was, by far, the worst option with just 7.72% usage.
Center placement in the header isn’t always the best option, however, especially if there are a lot of categories or other distractions. One fantastic option is to place it below the header. Our client, Candy.com uses this to great effect.
They’ve given the search bar it’s own space on the screen, with effective but not exaggerated use of white space. Not only is it center aligned, but it’s also extremely wide, and effectively uses a bold, accented border and button.
Candy.com also has some of the highest mobile usage of any site running our software.
While most sites won’t want to use such a large portion of their limited mobile screen real estate for the search bar, it’s something worth thinking about, especially if your mobile search visitors are converting at a high rate.
Most sites can get away with an icon in the top right on a mobile site, though. Use of the icon in the top right gets usage 5x higher than a hiding the search feature inside of the hamburger slide-out menu.
Again, in an effort to make sites appear more minimal, many designers are foregoing the use of a button. However, this is a barrier to usability as many users do not know that they can initiate the search using the “Enter” button on their keyboard. This is particularly unfriendly on mobile devices as the software keyboards don’t always provide the “Enter” or “Return” options.
Sites that did not provide any button averaged 6.01% search usage. In stark contrast, sites that did have a button of some sort averaged 14.49%.
The button text didn’t seem to make much of a difference, but when compared to a button that used a search icon instead of the word “search” or “go”, usage took a significant hit. Text buttons saw usage of 16.17%, whereas the magnifying glass icon averaged 13.44%.
Overall, among the sites that we examined this study, the ones with the highest usage always followed most of the above mentioned best practices.
Candy.com had the highest usage of all, and had the most unique search bar placement. It’s obvious that giving it priority in design can have major effects. Looking at the top twenty five sites (in terms of search percentage), we found that 40% used a top center placement, and the remaining 60% used top right placement. 100% of these sites used a search box rather than an icon, and eight of the top twenty-five sites had a perfect score, following all of the above-mentioned best practices.
Convincing more users to search will inevitably increase conversion rates. This influx of new search users often drives some other numbers down, though. Average order values, for example, tend to drop slightly as search usage increases.
However, the overall effect of increasing search usage is an increase in conversion rates and revenue.
Even after making notable changes in all of these areas, you may not see the results you’re looking for. What can you do next? Experiment with the blink test. Give people who are unfamiliar with your site three seconds to find the search box. If they can’t find it in that time, you’ve failed and need to make some additional refinements. Generally speaking, though, the above best practices will enable you to pass the blink test with flying colors.
Autocomplete is a feature that is expected by ecommerce shoppers. It’s everywhere. From their desktop operating system down to their smartwatch, autocomplete is helping people complete dozens of searches every day. Autocomplete suggestions of some sort are now found on 82% of major ecommerce sites as well.
Due to the overwhelming commonness of autocomplete, users have expectations of how it will look and work, so it’s important that it be optimized to measure up.
General autocomplete UI best practices
- Style various suggestion types differently
We’ll discuss more specifics later in this chapter. It’s important that the user be able to intuitively understand what they’re clicking.
- Do not use a scroll bar
The autocomplete widget should expand in size as the number of suggestions grow, but should never have a scroll bar. Users are not accustomed to removing their hands from the keyboard during an autocomplete interaction.
- Use a maximum of 10 suggestions
Between the various types of suggestions, do not include more than 10 total. Depending on the UI, less than 10 might be more ideal. In any case, the suggestions should not expand beyond the user’s view.
- Highlight the active suggestion
Autocomplete users should be able to hit the “enter” or “return” button on their keyboard once they see the suggestion they want to use. Highlight the top suggestion to help users understand that clicking “enter” will confirm that query and take them to results.
Display query suggestions
As mentioned above, the majority of top ecommerce sites today provide suggestions for search users. These suggestions help users in a number of ways.
- Query completion
A query completion helps the user by completing the words, making it faster for them to enter their search.
These suggestions help users who aren’t sure how to spell the keywords they need.
Pictured above, many sites now suggest searching within specific categories to help narrow search results down to more relevant items.
All of these autocomplete features make it easier for the shopper to search. They should all be supported in order to remove friction for the shopper.
Display product suggestions
As discussed in Chapter 1, there are different types of search queries. Product suggestions cater primarily to precise search users. These are the individuals who know exactly what they want and how much they are willing to pay for it.
These are the most valuable type of visitor, so it’s very important to cater to these visitors wherever possible. Product suggestions (pictured above) display either exact product matches, or general matches (depending on how specific the search query) for the user’s query. What are the best practices for these suggestions?
- Display between 3-5 products
Displaying too many products can result in choice overload, preventing users from seeing what they’re looking for.
- Display product images
Product images help to delineate products from search query suggestions, and help direct the user’s attention.
- Separate product suggestions in the UI
It should be obvious to the user which of the results are typeahead (search) suggestions, and which are product suggestions. It should be intuitively understood that clicking a search suggestion will take the user to a search results page, and that a product suggestion will take them to a product page. Separate these types of suggestions in the UI using a different background tone, a horizontal line, or by other means.
Display pricing information
Prices provide an answer to one of the shopper’s primary concerns. Whether the shopper is a precise search user, or a broad spectrum user, pricing information will be helpful to them.
Beyond setting pricing expectations, displaying price has an additional value. Properly displaying price along with other product data also helps shoppers to understand that what they are seeing is not a query completion which will take them to a list of similar products (i.e. T-shirts), but a specific product. In testing, removal of the price meant greatly increased bounce rates since the user believed they would be taken to a list of search results rather than to a product page.
In effect, the addition of price information to auto complete results does more than just inform the user of the price, but also intuitively helps them understand the functionality available from the search box. This can eliminate search results pages entirely for many shopping sessions in certain industries. This may also convince users to be more precise in their search query, perhaps using additional keywords so that they see the product they want within the autocomplete UX.
Provide an add-to-cart button
While most conversions will come from product pages, removing steps from the buying process will always result in increased conversions.
While 8 out of every 10 shoppers are researching products before they buy, a small percentage have already completed their research and know precisely what product they want. For these few, reducing the number of steps it takes for them to buy will increase conversion rates. While Amazon hasn’t shared how successful their 1-click ordering button or physical Dash buttons have been, we can imagine that they have been huge successes. Including an add-to-cart button increases conversions from search by 8% on average.
Provide filter options
Saving time and reducing steps for shoppers will generally have a positive effect on their satisfaction and on conversion rates. Modern web technology allows for results to update live as the user types. SearchSpring is now bringing the ability to filter search results within autocomplete, eliminating search results pages from many search queries.
15% of search visitors filter results using this autocomplete UI, preventing at least one page load.
Chapter 5: Relevancy Optimizations
Making the search experience more relevant is incredibly important for search visitors. Users expect that the result set they see will display the right products at the top of the page. While it’s not the scope of this guide to explain how to develop custom algorithms to improve overall relevancy, there are optimizations that most retailers can make with the help of a developer.
Handling synonyms and redirects
Synonyms are a common problem for retailers. People have different ways of searching for or defining the same thing. While a human is easily able to understand the similarity between the words “50 inch tv” and “50in tv”, a search engine usually needs some help. In a general sense, any versions of a product title that aren’t in the metadata won’t appear in search results without help.
Review synonyms monthly
The most obvious best practice when it comes to managing synonyms is to spend time programing them into your shopping cart or search engine dashboard. Make it a practice to review synonym needs once per month. Using your search dashboard or Google Analytics, find common searches that return zero results. While some of these searches will be for products you don’t carry, many of them will be synonyms for products you do have.
Support content searches using redirects
You may be surprised to find that some of your most common searches aren’t for products at all, but for content pages. Whether or not you have a blog, you certainly have pages that don’t have any products. Some of the most common content searches are for “return policy”, “shipping policy”, and “contact us”.
Setting up redirects to handle these searches should not take much time, and will help you support the majority of content searches.
Review popular searches with low conversion rates
This is a huge opportunity to increase conversion rates that many retailers are missing out on. As you review search queries in Google Analytics, take a look at searches that have very low conversion rates, and especially those with zero conversions. Likely, there’s a problem with the results that are loading, even if it’s not resulting in a “no results” page loading.
Creating redirects for synonyms with better results or custom-merchandising a landing page are great ways to increase conversions on these pages.
Create redirects for special landing pages
In chapter 1, we discussed various types of searches. The most common type of search is the broad spectrum search. The example we used in that chapter is for the term “laptop” which is a very common electronics product category.
Properly catering to the needs of these visitors can have profound benefits. One of the best ways to handle these searches is to use a form of guided selling by redirecting these searches to a landing page. Rather than displaying a list of laptops, many of which will not appeal to the shopper, provide them with a highly visual guide that will direct them to the type of laptop that they want.
This strategy can also be applied to brands if brand names are a popular search term on your website.
Support Persistent Queries
Persistent queries follow the user, and retain the keywords in the search box after the user hits “search”, rather than removing the keywords. This allows them to modify their search without needing to re-enter their entire search. This is important because the average search visitor enters more than 2 iterations of their search, often adding or removing a keyword in order to improve their search results.
There are 3 common reasons that a user might want to add or remove keywords:
- Not enough products in search results
Perhaps after entering a search query that is too complicated or unsupported by the engine, the user sees only a few products in the result set. Removing keywords may increase the number of results.
- Irrelevant results
After entering a search query, the user sees results that are not what they expected. Adding or removing keywords may improve matters. Spelling or other corrections may also be made.
- Too many results
After entering a search query that is too broad, the user sees results that are generally relevant, but can’t immediately find products that suit their needs. Entering additional keywords may narrow results.
Most users make these modifications when searching, but nearly 70% of ecommerce stores do not support this behavior by retaining keywords across pages. This forces users to re-enter their search, which may be several words in length. This adds friction to the searching process unnecessarily, making it harder for users to find relevant products.
Handling “no results” pages
No results pages are huge problems for many large websites. This is partly due to dated search engine technology, and can often be mitigated by integrating smarter search technology. However, no results pages are inevitable, and there are good opportunities for improvement on most ecommerce sites.
What can you do to make these pages more shopper friendly?
Prevent “no results” pages
Wherever possible, prevent shoppers from seeing these pages. The first step is to find common searches that are resulting in the display of these pages. Google Analytics can be configured to display this information, and some search providers display this information automatically.
While some of the search terms that are not loading results can be ignored, there are likely to be some that can fixed by creating a synonym.
As we discussed earlier in this chapter, many searches that aren’t loading results can be fixed with a simple synonym. For instance, if a search for “PJs” isn’t returning results, program a synonym for “pajamas”.
Create a custom results page
Taking it a step beyond programming a synonym, if the search is popular or important enough, you might actually want to create a custom results page with a manually selected result set and layout.
Follow UI best practices
Some searches will inevitably end up on no results pages, so it’s certainly helpful to optimize the UI. The goal of the no results page is really to get the user off of that page, and onto a page that’s more relevant. That should be the #1 priority.
Confirm their search
Often times, people land on these pages due to incorrect spelling or typographical errors. Displaying their search in large, bold font near the top of the page gives the user an opportunity to see the search they entered and realize their mistake.
Even when the user did not make an error, knowing what they searched for will prevent them from entering the same exact search a second time. This helps them to understand that they need to make a modification to their search in order to find the products they want.
Provide relevant product or query suggestions
If your search engine is smart enough, it may be able to identify possible search terms and products even though there isn’t an exact match for the user’s search. For example, while there won’t be any products with the word “teee shirt” in their product data, some engines would be smart enough to recognize that the user probably meant to type “tee shirt”.
Provide helpful query suggestions
Most no results pages can’t find any relevant products or search query suggestions. In that case, help the user by suggesting paths they can take to find products, such as by providing the following:
1. Search tips:
Show users how to improve their query. For instance, ask them to:
- Check for typos
- Use different keywords
- Widen their search
- User fewer words
2. Category suggestions:
Suggest that users browse using popular categories on your store.
Now is the time to improve your search. Even if you can’t implement all of these strategies, find a few that are simple and get started right away. Most retailers are surprised by how much an effect moving their search bar or adding a border can have on their bottom line.
Some of our clients who implemented just one or two of these changes saw search usage double or triple within just a few days. Use the included checklist to help your designer find the best opportunities to improve your site. Start with the simple, easy changes, and work your way up to the more difficult changes with your next UI refresh.
Lastly, be sure to implement only one change at a time so that you know exactly what affect that change is having.
Others have also read these articles
Have a question or have an interesting idea for an article you would like to see?