A customer has the right product in mind, types it into your website’s search box and sees nothing useful. Meanwhile, the item is available to buy through a category page or direct link. Before spending more to bring people into the shop, check whether the search inside it can find what you sell.
For a UK retailer, the practical question is where the failure starts: product visibility, the words being matched, the suggestions shown while typing, or the results page itself. Start with a small set of repeatable searches, then fix the cause you can demonstrate.
Find out which search your customers are using
Ask your developer to identify what powers the search box on the live website: the platform’s built-in search, an extension, an external search service or custom code. Check the mobile search as well as the desktop version. Record the search provider and theme or extension version before changing settings.
Shopify’s current search guidance explicitly distinguishes its native search from third-party apps and custom storefront experiences. A third-party app can control the results and may handle visibility differently. Changing a native setting is therefore not evidence that your actual search has changed.
For WooCommerce, ask the same ownership question rather than assuming every search box uses identical matching rules. Get the documentation for the component actually serving your results. There is little value in configuring a feature that the live search never uses.
Build a test set from real customer language
Choose a manageable sample of products that matter commercially. Include a frequently requested item, a recent addition, a product with similar variants and something a customer has already struggled to find. Use public product information and anonymised enquiry wording rather than customer details.
For each item, try:
- The exact product name.
- The everyday name a customer would use.
- The complete stock-keeping unit, or SKU, if customers order by code.
- A relevant size, material or compatibility phrase.
- A likely typing mistake.
For example, a cabinet-hardware retailer might test a handle’s catalogue name, “cupboard handle”, its full code and a phrase describing its fixing centres. This is an illustrative test, not a claim that every platform searches every attribute. Ask which fields your search supports before deciding where the information belongs.
Record the query, expected product URL, actual result and device. Capture suggestions before submission and results afterwards. An exact name that fails suggests a different investigation from a product that appears by name but disappears when the customer uses an ordinary alternative.
Check whether the product is eligible to appear
Shopify: publication and hidden status
A product’s presence in the admin area does not establish that shoppers can find it. Shopify’s searchability documentation says products must be published to the Online Store channel to appear in storefront search. It also identifies Unlisted status and the seo.hidden metafield as reasons a product can be excluded.
Check these before rewriting the description or adding a boost. Establish why a product was hidden before changing it: an item intended for a private offer should not become discoverable simply to make a test pass. If an app manages the field, ask its owner how to prevent the unwanted setting returning during the next update.
WooCommerce: catalogue visibility
WooCommerce documents separate catalogue visibility choices. “Shop only” allows a product on shop and category pages but excludes it from search; “Search only” does the reverse. “Hidden” leaves the individual product page accessible while excluding it from those listings. “Shop and search” is the normal option for products intended to appear in both.
That gives you a useful diagnostic: opening a direct product link successfully does not prove that the product is eligible for search. Check publication and catalogue visibility, then repeat the customer’s exact query while logged out. Preserve any deliberate trade-only or customer-specific restrictions.
Check the words and product codes being matched
If a product appears under its official name but not the term customers use, review its searchable information. Add accurate, useful wording where it helps someone understand the product. A proprietary range name alone may not explain that an item is a replacement hinge, a drawer runner or a cupboard handle.
Keep distinctions that affect the purchase. Different fixing centres, fittings or materials should not be treated as interchangeable just because the products look similar. Test the customer phrase against both a suitable item and a near-match that should remain distinguishable.
For code-based ordering, Shopify documents full SKU and barcode searches. Partial-code matching is supported when the code contains hyphens, rather than for any arbitrary fragment. Test the full code before treating a failed fragment as a fault. Confirm the rules separately for other search providers.
Ask your developer to explain any proposed matching change with examples. A repair should make relevant products easier to find without filling every search with loosely related stock. Compare the original failures and a few searches that already worked before accepting it.
Test suggestions and submitted results separately
The dropdown shown while someone types is only one part of the journey. Submit the same query with Enter or the search button, then inspect the full results page. Repeat this on a phone, including opening a result and returning to the search.
Shopify’s current Search & Discovery guidance documents separate result-type settings for predictive search and the submitted results page. A theme that supplies its own type parameter can override those settings. It also allows unavailable products to be shown, hidden or placed last. Check both the setting and what the live theme actually requests.
The same guidance says semantic search is included for eligible stores on Grow, Advanced or Plus with fewer than 200,000 products, and does not apply to predictive search. Do not assume that a broader match on the results page must also appear in the suggestions. Use the current requirements rather than an old activation tutorial.
Product boosts can promote selected products for particular terms within Shopify’s store search; they do not improve Google rankings. Use them for a justified relevance decision, after checking visibility and product information.
Finally, test the interaction itself: can a customer reach the search, submit it and select a result using the keyboard? Our website keyboard accessibility checks provide a practical starting point for that part of the journey.
Use search data without mistaking activity for success
Keep manual tests even if you have analytics. A recorded search tells you that something happened, not that the right product appeared or that a purchase became easier.
Google’s GA4 enhanced-measurement documentation says view_search_results uses recognised URL query parameters to detect site search. The defaults are q, s, search, query and keyword; other parameters can be configured. The search_term parameter supplies the Search term dimension.
Compare this with your live search URL. If results update without an appropriate query parameter, do not assume automatic collection covers the interaction. Have the developer verify the implementation. Google also requires that personally identifiable information is not collected. Check whether free-text searches could expose names, email addresses or order details before enabling collection.
The documented automatic event does not provide a result count. To measure zero-result searches, ask for a separately verified count from the search provider or a suitable custom implementation. Confirm that it counts the customer-visible result set, rather than suggestions or an earlier response.
Use the evidence to prioritise repeated failures involving products you actually sell. For wider measurement work, our analytics and reporting service connects website behaviour with useful commercial outcomes.
Give unsuccessful searches a useful next step
A search for something you do not stock can legitimately return nothing. Make that state helpful: retain the query, allow easy editing and offer relevant category navigation or a clear contact route. Distinguish suggested alternatives from an exact match, particularly where fit or compatibility matters.
On mobile, check that the keyboard, search overlay and empty-results message do not trap the customer. Someone should be able to revise their search without reloading the homepage or losing the words they entered.
What to ask for before signing off the repair
Request the original test set with before-and-after results, the component changed and the reason for the change. Confirm that intended products appear, unsuitable variants remain distinguishable and deliberately hidden products stay hidden. Repeat the tests after a product update to check that the catalogue workflow preserves the fix.
Agree who will review failed searches and when: after a catalogue import, a theme change or a search-extension update is a sensible starting point. Keep the sample small enough that someone will actually run it.
If customers cannot find products you know are available, talk to BuzzBoost. Send the search phrase, expected product link and what the customer sees. We can help identify the break in the journey and plan a focused website repair.
Featured image: AI-generated editorial artwork, not a photograph of a real BuzzBoost office, client or result.
