Site search analytics: what zero-result searches tell you

TL;DR

• A zero-result search is a visitor telling you, in their own words, what they wanted and could not find. It is one of the clearest signals your analytics can give you.
• Google Analytics 4 records site searches automatically when the term is in the URL, but not the number of results. Add a results count to your search event, or use your search tool's report.
• Most zero-result searches fall into a few patterns: different words, typos and codes, missing content, missing products or features, navigation and full questions. Each has a different fix.

Why zero-result searches matter

People who use site search are usually looking for something specific, and a dead end is where many of them give up. Nielsen Norman Group warns that when users get no results, "there's a high risk of site abandonment," and that in its research people often hit a no-results page "when the content exists," but the search did not match their wording (NN/g).

That makes zero-result searches doubly useful. Each one is a lost visitor you can win back with a better results page, and a piece of research that shows the gap between how visitors talk and how your site is written. On a B2B site with modest traffic, a few dozen zero-result searches a month can be enough to find the integration nobody mentions by name, the product line you call something else, or the policy page that does not exist.

Tracking site search in GA4

What enhanced measurement gives you

GA4's enhanced measurement includes site search. It sends a view_search_results event "each time a user is presented with a search results page, as indicated by the presence of a URL query parameter." By default it looks for five parameters: q, s, search, query and keyword, and you can add your own in the data stream settings. The term fills the search term dimension (Google Analytics Help).

Two limits matter:

• Detection depends on the URL. If your search shows results without putting the term in the URL, as many JavaScript search widgets do, enhanced measurement will not see it. Send the recommended search event yourself instead; its only parameter, search_term, is required (GA4 events reference).
• There is no result count. GA4 knows a search happened and what was typed, not whether anything was found.

Adding the result count

The simplest fix is to send the number of results with the search event, as a custom parameter:

// Call this when your search results render
gtag("event", "search", {
  search_term: query,
  results_count: results.length
});

Then register results_count as a custom metric in GA4's admin settings, so you can filter and report on it. Google notes that custom metrics are for numeric values from event parameters, are always event scoped, and appear in reports 24 to 48 hours after you create them (Google Analytics Help). A zero-result search is then simply a search event with a results count of zero. If you prefer, send a separate event such as search_no_results with the term.

If your search runs on a hosted search product, check its own analytics first. Most include a "no results" report that is easier to read than anything you build in GA4.

Keep personal data out

Visitors type all sorts of things into search boxes, including email addresses and order numbers. Google's policy is that you must not send data Google could recognize as personally identifiable information, such as email addresses or phone numbers. Page URLs and parameters must be free of it, and GA4 offers data redaction that removes email addresses on a best-effort basis and query parameters you specify (Google Analytics Help). Strip or mask anything that looks like an email or a number before you send a search term.

The numbers worth watching

Do not chase an industry benchmark for the zero-result rate; published numbers mix very different sites. Track your own rate month by month and by section.

Reading the patterns

Export your zero-result terms for the last month, group variants together and sort by count. Most of them will fit one of these patterns:

Two of these deserve special attention.

Missing features are product research. A cluster of searches for something you do not offer is demand you can measure. Share it with product. Meanwhile, a short "we don't offer this, here is what we do instead" answer turns a dead end into a useful reply.

Full questions are a sign of changing habits. Visitors used to chat assistants and AI search increasingly type questions, not keywords. A keyword engine matches words, so "can I change plans mid-year" may find nothing even when your billing page answers it. Search that works from meaning, using embeddings rather than exact words, handles these much better. That is the idea behind AI site search, which answers the question and cites the page instead of returning a list of links. Either way, these searches show search intent in its purest form: the visitor's exact question.

An illustrative month

Here is what one month of zero-result terms might look like for a fictional B2B software company with about 400 site searches:

Five changes, each under an hour of work, cover 41 failed searches, roughly one in ten. The next month's export shows whether they held.

Fix the no-results page too

Some searches will always fail. The page visitors see when that happens decides whether they leave. NN/g gives three guidelines: "Clearly explain that there are no matching results," "Offer starting points for moving forward," and "Don't mock the user." Baymard Institute's research adds that nearly 50% of sites fail to provide effective ways to recover from a search with no results, and recommends five strategies: related categories, alternative searches, personalized recommendations, showing contact options such as phone and chat, and popular items (Baymard).

For a B2B site, a good no-results page:

• Says plainly that nothing matched, and repeats the search term.
• Suggests a corrected or broader search, if one exists.
• Links to the three to five pages visitors most often need: pricing, product overview, docs, contact.
• Offers a person or an assistant: "Ask us directly." Baymard notes that visitors on a no-results page are "in a high-risk group" for leaving.

A monthly review routine

Zero-result searches are one of several sources of unanswered questions. Chat conversations where an assistant could not answer and questions sent to sales are others; our guide to knowledge gaps shows how to work through all of them in one weekly loop, and mining chat logs for content ideas covers the SEO side. Every fix you make to your pages also improves your knowledge base if you run an assistant built on your site.

For ASDK, Search, an agent that replaces the search box with short answers that cite your pages, is coming soon and will be included from the Starter plan when it launches. Until then, the routine above works with any search tool.

Zero-result searches are small, honest messages from visitors: "I looked for this and you didn't have it." Track them properly, read them every month and fix the cause, whether it is a missing synonym, a missing page or a missing feature. Few analytics reports give you so much direction for so little effort.