Articles / Ecommerce and Shopify
Ecommerce site search best practices: what our A/B tests show
What we found
Shoppers who search often convert 2 to 3 times more, because they know what they want. In our tests, a bigger or louder search box never lifted conversion. The gains come from what search returns: the closest match first, clear filters, and no dead ends.
When a Shopify store gets built, search is usually an afterthought. The team spends its time on menus and collection pages, and the search box gets whatever room is left in the header. That is a miss, because shoppers who use search often convert at 2 to 3 times the rate of everyone else.
I have tested search on ecommerce sites for years, and the results surprised me. Making the search box bigger or louder never lifted conversion. What matters is what search returns once a shopper uses it. This guide covers why searchers convert, what our A/B tests show, and a checklist for getting search right.
Shoppers who search convert 2 to 3 times more because they know what they want
Search matches the shopper’s mindset. People who search usually know exactly what they are looking for. They are not browsing other products. They type a product name, a product code or a brand, and a good result gets them there faster than any menu. Shopify reports the same pattern: shoppers who search are 2 to 3 times more likely to convert, based on Forrester research cited in its guide to search and discovery.
There is a second kind of searcher too. This shopper already spent time trying to find a product in the menus and gave up. They will not go back to the navigation, so in frustration they search. If search fails them as well, they leave.
I saw this on a content site with over 20,000 pages. Visitors who searched reached the page they wanted in fewer clicks than visitors who used the menus.
Desktop shoppers preferred the compact search box by 17.6%
Since searchers convert better, the obvious move is to get more shoppers into search. We tested that on a desktop homepage. The variation widened the header search box and its preview dropdown, and added hint text inviting shoppers to search by product type, color or style.
Shoppers preferred the original compact box. The bigger box cut conversion 17.6% over 18 days, and we paused the test. The larger box pulled attention away from the main calls to action on the page, so more shoppers browsed instead of taking a clear next step.
Evidence · Homepage · Lost
Does clearer search hint text lift desktop homepage conversion?
18 days · desktop · We widened the search box and wrote clearer, more helpful hint text.
-17.6% conversion, -36.2% revenue per visitor
Read the full testThe lesson: shoppers who want to search already find the box. Keep it visible on every page, but it does not need to be the loudest thing in the header. Our homepage tests cover other ways to give shoppers a clear path to products.
An outlined search box moved revenue per visitor just 1.0%
A second test on the same kind of page widened the search box and added a border for contrast. Conversion fell 9.7% and revenue per visitor fell only 1.0% over 15 days, and neither change reached our confidence bar.
Evidence · Homepage · Not significant
Does a wider, outlined search box lift homepage conversion?
15 days · desktop · We widened the header search box and outlined it for contrast.
-9.7% conversion, -1.0% revenue per visitor
Read the full testWhat we took from it: improve the search results before drawing more eyes to search. If the results are weak, more searches just mean more frustrated shoppers.
This is where I see the most room on Shopify. Shopify’s built-in search is good, but it can rank results too loosely. A shopper can search for “red dress” and get white dresses near the top, because the word red appears somewhere in the description. Search platforms like Searchspring or Elasticsearch let you weight the product title, metafields and page content so the closest match comes first.
The best way to spot this is a usability test. Watch for shoppers who have to search again with different words, or who leave right after seeing the results. Both are signs that search is returning the wrong products.
Sending searches to collection pages added 6.2% more browsing, but no sales lift
We also tested sending the most common searches, about 50% of all searches, straight to the matching collection page instead of a results page. The idea was to shorten the path for high-intent shoppers.
Shoppers viewed 6.2% more pages per session, but conversion fell 3.4% and revenue per visitor fell 3.1%. Neither sales result reached our confidence bar, so we paused the test and kept the results page. The collection pages were broader than what shoppers had typed, so they had to browse to find the product they searched for.
Evidence · Collection page · Not significant
Do search redirects to matching collection pages lift conversion?
22 days · responsive · Common search terms redirected to the best matching collection page
-3.4% conversion, +6.2% pages per session
Read the full testIf you use redirects, save them for very specific terms where a single collection is clearly the answer. For everything else, a results page ranked by the closest match does the job better.
A louder search box does not help until the results do.
Across our search tests, making the box wider, outlined or more prominent never lifted conversion. Shoppers who want to search already find the box. What changes the outcome is whether the first results match what they typed.
Stores with over 200 products need filters on search results
How much filtering you need depends on the size of your store. Count how many products a shopper has to choose from. That number should shape how you tag products and whether you need advanced filters.
Stores with over 200 products, and a lot of variety, often see shoppers use filters on search results pages as well as on collection pages. These shoppers still convert at a higher rate, because they chose the path to the product and narrowed the results themselves. People trust their own choices more than a site’s navigation.
So make filters clear, and make them work on both desktop and mobile. Let shoppers narrow by the attributes they care about, like size, color and price. This is one of those places where the job is to stay out of the shopper’s way. Our collection page tests show the same thing on category pages.
Searches with 0 results show you what to fix or what to stock
Pull a report of searches that returned no results, every month. Each one is either a search problem or a product opportunity. Some are misspellings or different words for products you carry, which you can fix with synonyms. Shopify’s Search & Discovery app lets you add those. Others are products shoppers assume you sell. That is demand you can measure before you stock anything.
A failed first search costs more than it looks. Nielsen Norman Group found that almost half of users whose first search failed gave up right away.
Nielsen's research found search should be a simple, visible box on every page, and that queries are short. In an ecommerce study, 51% of first searches succeeded, and almost half of users whose first search failed gave up right away.
Jakob Nielsen May 2001 · www.nngroup.com ↗
Our tests agree the box should be easy to find, but they show a bigger box is not the fix. The first search has to work, so the gains are in the results.
Shoppers who search have already told you what they want. The job is to show it to them, not to make the box bigger.
Ecommerce site search checklist: 7 things to check
- Keep the search box visible on every page, on desktop and mobile, without making it the loudest thing in the header.
- Name and tag products the way shoppers search: product names, codes, brands and colors.
- Search your own top terms and check that the first results match the words used.
- Add filters to search results if you carry more than 200 products, and test them on mobile.
- Review searches with 0 results every month, then add synonyms or note the demand.
- Watch real shoppers search in a usability test. Repeat searches and exits after results are the warning signs.
- Judge any search change on revenue per visitor, not just clicks or pages viewed. A change that adds browsing can still cost sales.
Frequently asked questions
Do shoppers who use site search convert better?
Usually, yes. Shoppers who search are often 2 to 3 times more likely to convert, based on Forrester research cited by Shopify. They know what they want, so a good result gets them to the product faster than the menus.
Should I make my search box bigger?
Make sure it is visible on every page, but bigger is not always better. In our desktop test, a wider search box with hint text cut conversion 17.6%, because it pulled attention from the main calls to action. Improve the results before you make the box louder.
Is Shopify's built-in search good enough?
For many stores, yes, especially with Shopify's Search & Discovery app for synonyms and filters. If shoppers get loosely matching products near the top, a search platform like Searchspring or Elasticsearch lets you weight product titles, metafields and content.
What should I do about searches with no results?
Review them every month. Misspellings and different words for products you carry can be fixed with synonyms. Searches for products you do not sell show demand you can measure before you stock anything.
Should site search redirect to collection pages?
Only for very specific terms where 1 collection is clearly the answer. In our test, redirecting common searches to collection pages added 6.2% more browsing but did not lift sales, because the collections were broader than what shoppers typed.
The evidence
The 3 tests this article leans on
Evidence · Homepage · Lost
Does clearer search hint text lift desktop homepage conversion?
-17.6% conversion, -36.2% revenue per visitor
Read the full testEvidence · Homepage · Not significant
-9.7% conversion, -1.0% revenue per visitor
Read the full testEvidence · Collection page · Not significant
Do search redirects to matching collection pages lift conversion?
-3.4% conversion, +6.2% pages per session
Read the full test