Fix on-site search with Shopify Search & Discovery: solve no-result searches with synonyms and no-click searches with boosts, and learn the filter limits and a weekly routine.
A visitor who uses your search box is a visitor who knows what they want. But if they type "hoodie" and see "no results found", they leave, even when your "hooded sweatshirt" product sits on the same site. This article walks through the synonym, product boost, filter and report settings in Shopify's Search & Discovery app, with the official limits, in an order you can actually follow.
Why is on-site search a separate job?
Menus and collection pages put visitors on a path you built. The search box lets visitors draw their own path with their own words, so the gap between your catalogue names and your customers' vocabulary shows up here directly. If your product is called "hooded sweatshirt" and the customer types "hoodie", search may not know they mean the same thing.
The good news: Shopify provides the tools to close that gap, and most settings need no code. The bad news: the settings only work if they are not set once and forgotten, because the words customers search for keep changing. That is why most of this article is about the measurement routine that feeds the settings.
Measure first: Search & Discovery reports
According to the Shopify Help Center, the Search & Discovery app offers five search reports: click rate, purchase rate, searches by query, searches with no results, and searches with no clicks. The metrics shown cover the last 30 days; for other date ranges you go to Reports under Analytics. The reports do not include predictive search activity.
These reports separate two different problems:
- Searches with no results: either the product does not exist, or the customer used a different word. The second is fixed with synonyms.
- Searches with no clicks: results exist but the customer did not see what they wanted. This is usually a ranking problem, handled with product boosts.
Looking at the top 10 rows of each list every week is enough. Do not chase long lists; catch the repeating words.
Hypothetical example: the no-result rate
Suppose your store had 1,200 searches in a month and 180 returned nothing. The rate is 15% (180 / 1,200). If 90 of those no-result searches involve just three words (say "hoodie", a misspelt variant and "tshirt"), connecting those three words to the right products brings no-result searches down to 90, a rate of 7.5% (90 / 1,200). The figures are invented only to show the arithmetic; your own store's report will differ.
How to build synonym groups
Shopify's "Modifying search" page spells out the limits. Each synonym can be a single word or a phrase of up to 5 words. A group can hold up to 20 synonyms, and the whole store up to 1,000. A synonym must be unique across the store, so you cannot put "hoodie" in two groups. Multi-word synonyms work as phrase searches. Group titles do not affect results; they are only for your own organisation. Synonyms are not used for matches on SKU and barcode fields.
A practical order of work:
- Pull the repeating words from the searches-with-no-results report.
- Match each word to a product group, and name the group in your catalogue's language, not the customer's.
- Put spelling variants and loanword usage (for example "hoodie" and its local spellings) in the same group.
- Do not link identifiers such as brand, model or internal stock codes to synonyms; they already match through product fields.
- Once a month, update the groups against new no-result searches.
Shopify's developer documentation says the Storefront API search offers typo tolerance, partial match and automatic synonyms. These do not make manual synonym management unnecessary; for words specific to your sector, building your own groups is the safer route. Test on your live store to see which words are matched automatically in your language.
Product boosts: which product should rise?
A product boost moves a product you choose to the top of results for specific search terms. Shopify says up to 10 search terms can be added to one product boost. A boost only applies to products available for sale; sold-out products go to the end of the results. Boosts are not applied when a query contains search syntax. They can be managed through the Bulk Editor or the "Search product boosts" metafield, which can also be edited with Shopify Flow.
Use boosts by asking:
- Which terms appear in the searches-with-no-clicks report, and which product is the right answer?
- Is the product you want to feature for brand or campaign reasons actually in line with the customer's intent?
- Does the boosted product have enough stock? A boost on a product that has sold out does nothing.
You can also choose what happens to out-of-stock products in the app's Settings: according to Shopify they can be displayed, hidden or placed last, and the default is to show them after other matching results. The right choice depends on your catalogue; for products that sell out often but return, placing them last is usually a safer start than hiding them.
Result types and Semantic Search
By default, storefront search can return product, page and blog post results. Predictive search can return products, pages, blog posts, collections and query suggestions. These are set in the app's Settings, and the published theme can override them, so always test on the live store after changing them.
Shopify's Semantic Search is opt-in. According to the help page, it is currently available only on English storefronts with fewer than 200,000 products, on the Shopify, Advanced and Plus plans, and it does not apply to predictive search. So if your storefront is in Turkish, do not count on it; synonyms and boosts are the main tools you control. The scope may widen over time, so check the Help Center page occasionally.
Filters: narrowing after the search
After searching, the customer should be able to narrow the results. In Search & Discovery, standard filters exist in every store: availability, category, price, product type, tag and vendor. Custom filters are based on product options, metafields or metaobjects. Product options such as size, colour and material are natural filter candidates.
The official limits to know when setting up filters:
- A store can have up to 25 filters, and each filter source can be used only once.
- Collections with more than 5,000 products do not display filters.
- Search results with more than 100,000 products hide filters; filters and value counts on search results are based on up to the first 1,000 results.
- A filter shows at most 100 values on the storefront; the rest are not displayed.
- The price filter does not display for currencies other than the shop's default currency.
- Different filters combine with AND. Values within one filter combine with OR by default; tag, metafield list and metaobject reference list filters can be switched to AND.
Filters disappearing on very large collections can be a surprise. Splitting collections into sub-collections before the catalogue grows is how you stay under that limit. For the SEO side of titles, descriptions and filters, see our Shopify collection page SEO guide.
The SEO side of filter and search pages
Applying filters can generate a practically endless set of URLs for Google. The Google Search Central page "Managing crawling of faceted navigation URLs" calls this an infinite URL space and recommends preventing crawling of filtered URLs if you do not need them indexed. It presents blocking the relevant parameters in robots.txt as the primary option, and describes rel="canonical" and nofollow as less effective in the long term. It recommends a 404 status code when a filter combination returns no results.
That page makes no specific recommendation for internal search result pages. It would therefore be wrong to present "block search pages in robots.txt" as a Google requirement; the decision depends on your own indexing policy. Google's robots.txt help also reminds us that disallowed URLs may still be indexed without being crawled, so do not rely on robots.txt alone to remove a URL from the index. For the wider picture, use our e-commerce SEO technical checklist.
A weekly search maintenance routine
Setting things up once is not enough; this short routine keeps search current:
- Open the top 10 rows of the no-result and no-click reports in Search & Discovery.
- Add words that have a product but were not found to a synonym group.
- Note words with no matching product; if they repeat, that is a product or content opportunity.
- For no-click queries, add a boost to the right product and check its stock.
- If you added new product options or metafields, review whether you need a new filter.
- Test the changes on the live store on both desktop and mobile.
To follow the same routine alongside your organic performance on Google, our weekly Search Console guide helps. Part of the repeating maintenance, such as updating the boost metafield, can be automated with Shopify Flow.
Conclusion
On-site search is a conversion area that needs no design but does need regular upkeep. Read your customers' real words from the reports first, fix no-result searches with synonyms and no-click searches with boosts, plan filters around the limits, and manage filtered URLs for Google separately. A ten-minute weekly routine carries most of the work.
Sources: Shopify Help Center, "Modifying search", "Search & Discovery reports and analytics" and "Search & Discovery filters" pages; Shopify.dev, "Search and discovery" (Storefront API); Google Search Central, "Managing crawling of faceted navigation URLs" (last updated 2025-12-18) and "Useful robots.txt rules" (accessed 2026). The calculation example is hypothetical.