Definition
What is the site search failure rate?
The site search failure rate is the share of searches performed on your site that lead to no product page view. It counts searches with no results, but also those that display products nobody opens. It measures the buyer's behaviour, not the engine's response. You can calculate it with the tools you already have, on one week of data.
What the measurement counts as a failure
The rule is simple: a search succeeds when the buyer opens at least one product page after running it. Everything else is a failure, whether the engine displayed zero results or four hundred.
Three cases need a written rule before the first calculation.
- The reworded search. A buyer types "space heater", finds nothing, then types "portable radiator" and opens a product page. Counted per search, that is a failure followed by a success. Counted per session, it is a success. Both rules are defensible: pick one and keep it from one month to the next.
- The information search. A search for the return policy or delivery times succeeds when it leads to the right page, not to a product page. Classify those queries separately, or they inflate the failure rate.
- The search that ends on a category. Some engines send the visitor straight to a category page when the query matches its name. Decide whether an opened category counts as a success or whether only the product page counts.
Calculating it in GA4
GA4's enhanced measurement includes a "Site search" option. When it is on, GA4 records a view_search_results event every time the page address contains a search parameter (q, s, search, query or keyword by default; you can add others). The search term is in the search_term parameter.
A product page view corresponds to the view_item event, one of the ecommerce events Google recommends. A two-step funnel exploration (the search, then the product page) then gives you the share of people who open a product page after searching. That exploration counts users rather than searches: treat the result as an approximation, which is enough to follow the trend.
Two situations block the measurement. A search that displays its results without changing the page address is not captured: you then need an event set up in Google Tag Manager. A site that does not send the view_item event cannot link the search to the product page: that is the first thing to check.
Read the rate rather than aim for one
The failure rate is compared with itself, month after month, rather than with a market average. It depends on the size of the catalogue, your customers' vocabulary and the counting rule you chose: two sites that count differently cannot compare their numbers.
The overall figure is an alert. The work is done on the list of failing queries, sorted by frequency. The most common queries repeat every day: a fix on one of them can move the rate as early as the following week. Also separate mobile from desktop, where the search field and the results list are not displayed the same way.
Site search failure rate = searches that lead to no product page ÷ total searches × 100
Searches that lead to no product page include those with no results and those whose results are never opened. Information searches (returns, delivery, warranty) are removed from the calculation and tracked separately.
Take a fictional business that receives 12,000 searches a month on its site. Its engine displays zero results for 720 of them: its zero-results rate is 6%, a figure that reassures its team.
It then links each search to the product pages opened. For 1,800 other searches, products are displayed but nobody opens a single one. Its failure rate therefore reaches 21%: 2,520 searches out of 12,000. Reading the queries involved mostly reveals poorly recognized part numbers and words its customers use that its product pages do not.
We measure the failure rate before talking about tools or advertising budget. The measurement takes one week with GA4 or the tools you already have. We then read, one by one, the hundred most frequent queries that fail: that reading is what locates the problem (vocabulary, accents, part numbers or attributes missing from the product data).
We fix things from the simplest to the longest: synonyms, tolerance for accents and typos, then the structure of attributes. A new search engine is only justified if the rate stays high once those three steps are done. We track the rate every month, next to the conversion rate of sessions with search, to tie the work to sales.
Not to be confused with
- Zero-results rate
- The zero-results rate only counts searches for which the engine displays nothing. The failure rate adds the searches whose results are never opened. It is therefore always equal or higher.
- Search exit rate
- The search exit rate counts visitors who leave the site right after a search. The failure rate also counts those who stay, browse elsewhere or search again without opening a product page.
- Conversion rate
- The conversion rate measures the share of visits that end in a sale. The site search failure rate is measured one step earlier: it tells you whether the buyer found a product worth looking at.
Related concepts
- Conversion rate
- Conversion rate optimization
- Cart abandonment
- Faceted navigation
- Product data maturity
- GA4
- Core Web Vitals
Further reading
- eCommerce site search: improving results across your catalogue
- Product data: organizing your online catalogue
- eCommerce site speed: prioritizing fixes for a large catalogue
- Search intent: how to answer your customers' questions
- CRO: a guide to conversion rate optimization
- Conversion tracking: connecting your campaigns to actual sales
Related services
Frequently asked questions
Where do I find the site search failure rate in Google Analytics?
GA4 does not show it as such. Turn on site search in enhanced measurement, check that your product pages send the view_item event, then build a funnel exploration that links the search to a product page view. The result counts users rather than searches: it is enough to follow the trend from one month to the next.
Is a search that leads to the returns page a failure?
No. An information search succeeds when it leads to the right page. Classify queries about returns, delivery, warranty or your contact details separately, or they will inflate your failure rate. An information search that finds nothing is still a failure: you fix it by making those pages searchable.
What failure rate is normal?
The figure is mostly compared with itself. It depends on the size of your catalogue, your customers' vocabulary and the counting rule you chose. Track it every month with the same rule: a drop after fixing synonyms and accents shows the work is paying off, while a rate that does not move points instead to the product data.