Online retailers: site search is the lever you are not measuring
About half of your visitors search instead of browsing. When search comes up empty, they leave without a trace in your sales reports. It is the only conversion lever that can be measured in one week and that costs no advertising budget.
- The Baymard Institute observes that about half of the participants in its tests use search as their preferred strategy for finding a product, while the other half browse.
- Its 2026 study finds that 56% of sites fail to adequately support users' search needs, across a benchmark of more than 170 major ecommerce sites and apps.
- A failed search produces no commercial signal. The visitor leaves, and nothing in your reports tells you why.
- Your site search failure rate can be measured in one week with the tools you already have.
- On 40,000 searches a year with an 18% failure rate and $4.20 of value per successful search, the loss approaches $30,000 a year.
On this page
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 differs from the zero-results rate, which only counts searches that return nothing at all. A search that returns four hundred irrelevant products is as much a failure as a search that returns none, and that is what makes the measurement useful.
Why the failure is invisible
What follows is written for ecommerce managers and owners whose catalogue runs past a few hundred SKUs: shoes, furniture, hardware, home furnishings, books, parts and distribution. If you sell twelve products, the question does not arise for you.
Site search occupies a place that most management teams underestimate. The Baymard Institute observes that about half of the participants in its tests choose search rather than navigation to find a product.
A visitor who searches has a much more precise intent than one browsing a category. They know what they want. It is your most qualified traffic, and it is also the traffic you lose most quietly.
When an advertising campaign underperforms, the report shows it. When a site search fails, nothing happens: no product page viewed, no cart, no cart abandonment. The visitor closes the tab and your dashboard records one more visit.
A company that does not measure its site search will conclude that its problem is traffic, and will buy more media. It will then pay to send more visitors into a system that already loses half of them. It is the most common and most avoidable expense in ecommerce.
Measure your failure rate in one week
The measurement is done with the tools you already have, without buying anything.
Three numbers are enough: the total number of searches over the period, the number of searches followed by a product page view, and the number of searches followed by an add to cart.
The first report gives you your failure rate: the share of searches that lead to no product page. The second gives you the value of a successful search, by dividing the revenue from sessions with search by the number of searches.
One point of method changes everything: do not settle for the zero-results rate. A search that returns four hundred irrelevant products never shows up in that count, yet it is a complete failure from the buyer's point of view. It is the difference between measuring your engine and measuring your customer.
Then pull the list of your hundred most frequent queries and read them one by one. Half a day of reading will teach you more about your buyers than six months of reports.
The six query types that break a large catalogue
The part or model number. A professional buyer searches by reference. If your engine does not index it, or chokes on an extra space or hyphen, the sale is lost.
A competitor's reference. The most profitable case and the least addressed. Every equivalence you enter turns a lost search into a sale, at a marginal cost, that is, what it costs to produce one more sale, close to zero once the process is in place.
The query by symptom or by use. "Oil leak", "creaky floor", "chair for back pain". The Baymard Institute notes that these searches often return irrelevant results at the top of long lists.
The query with attributes. "Men's winter boot 11 wide". If your attributes are not structured, the engine treats those words as free text and fails.
The customer's vocabulary rather than yours. They search for "space heater", your product page says "portable radiator". A synonym dictionary settles this case in a few hours.
Typos and accents. In French, and your Quebec customers do search in French, missing accent tolerance produces massive and perfectly silent failures.
Cases 5 and 6 can be fixed in a day and often account for half of the gain. Case 2 takes long, steady work and builds the most durable asset, because your competitors will not do it.
What failure really costs
The calculation uses your own numbers and fits in three lines.
| Line | What you take | Scenario |
|---|---|---|
| Annual searches | Total number of site searches over twelve months | 40,000 |
| Failure rate | Share that leads to no product page | 18%, or 7,200 |
| Value of a successful search | Revenue from sessions with search divided by the number of searches | $4.20 |
The annual loss approaches $30,240. That amount appears in no report, is the subject of no meeting, and compares directly with an advertising budget.
Two caveats. The value of a successful search is an approximation, since some of those visitors would have bought through navigation. Use it as an order of magnitude for trade-offs, not as a forecast of recoverable revenue. And redo the calculation with your real volumes: on a distribution catalogue, annual searches often run well past 40,000.
Fix it without changing tools
The reflex is to buy a search engine. It is rarely the first step and it is almost always the most expensive one.
Start with synonyms. Take your hundred most frequent failing queries, and create the equivalences to your vocabulary. A few hours, no development, immediate effect on sales.
Continue with tolerance for accents and typos: a simple configuration setting on most platforms, and a decisive lever in French.
Then deal with attributes. An engine can only filter what is structured. This brings us back to product data and its cost, covered in our article on product data maturity. This is where the real ceiling sits: no tool compensates for missing data.
Consider a new engine only once those three steps are done and the residual failure rate still justifies it. Many companies buy the tool first and discover the problem was in their catalogue.
Searches that are not about a product
Some of your visitors are not looking for an item, but for a piece of information that decides the purchase.
The Baymard Institute observes that 34% of users attempt to search for non-product content on an ecommerce site. It also notes that 15% of sites do not support basic queries of this type, such as a search for the return policy.
It is a doubly foolish loss. The information exists on your site, often in the footer, and the visitor cannot find it at the precise moment it conditions their decision.
The fix takes an hour, at close to no cost: make the return policy, delivery times, payment methods, warranty and contact details searchable. On mobile, where secondary navigation is harder, the gain is even clearer.
What to establish before raising the advertising budget
These five numbers can be pulled in one week with your current tools. Without them, additional media budget feeds a system that loses part of what it receives.
- How many site searches does your site receive per year, and what share leads to no product page?
- What is the average value of a successful search, in revenue?
- Among your hundred most frequent queries, how many fail over vocabulary or accents?
- Are your product attributes structured, or does the engine treat them as free text?
- If the failure rate does not drop after fixing synonyms and accents, who decides the next step, and on what budget?
The useful answer gives a failure rate as a percentage and a value in dollars. A weak answer says the engine works fine. A vendor who recommends a new search engine without having measured your failure rate or read your top hundred queries is selling a tool, not a fix.
Measuring that rate and reading your real queries is part of what we deliver in a paid audit.
What never gets delegated: your customers' real vocabulary, the equivalences with competitor references that only your sales team knows, and the decision to invest in data rather than in a tool. What gets delegated: measuring the failure rate, reading the queries, building the synonym dictionary, configuring accent tolerance and the monthly follow-up. A company that reads its top hundred queries finds, in half a day, fixes that pay back more than a quarter of advertising. A company that never reads them will buy a more expensive engine to index data that still is not there.
Reading the intent behind those queries is developed in search intent.
The overall view of conversion is in what conversion rate optimization really fixes.
Turning a catalogue into a tool that sells, rather than a storefront to maintain, is at the heart of the Improve your site's conversion goal.
Your catalogue is heavy and the problem plays out across tens of thousands of SKUs? See our work in ecommerce.
Every second of load time also costs sales: the math is in our article on the slowness of a large catalogue.
Frequently asked questions about site search
Is the zero-results rate enough?
No, and it is the most common measurement mistake. A search that returns four hundred irrelevant products never appears in that count even though it is a complete failure from the buyer's point of view. Measure instead the share of searches that lead to no product page view: that is the customer's behaviour, not the engine's response.
Should you buy a specialized search engine?
Rarely first. Synonyms and tolerance for accents and typos can be fixed in a day without development and often account for half of the gain. An engine can only filter structured attributes: if your product data is incomplete, the tool will not change that and you will have paid to find out.
How long before you see an effect?
Vocabulary and accent fixes produce an effect the following week, because they touch queries that repeat every day. The work on attributes and competitor equivalences spreads over several months and is the durable gain. Track the failure rate monthly rather than traffic.
Are competitor reference equivalences worth the effort?
It is the longest job and the most durable asset. Every equivalence you enter turns a lost search into a sale, and your competitors probably will not do it. The difficulty is not technical: that knowledge lives in your sales reps' heads and leaves with them, which makes structuring it doubly useful.
Does this also affect SEO?
Indirectly, and usefully. Your internal queries reveal your buyers' real vocabulary, which often differs from the wording on your product pages. It is a source of real words, at no cost and specific to your customer base, that most companies never use for their content.
What about searches that are not about a product?
They count, and they are often neglected. Some visitors look for the return policy, delivery times or the warranty at the precise moment that information decides the purchase. Making that content searchable takes an hour, for a sales gain that is especially clear on mobile, where secondary navigation is harder.
- Baymard Institute, Ecommerce Search UX Best Practices, 2026 study whose site search benchmark covers more than 170 major ecommerce sites and apps, American and European. Source for the finding that 56% of sites fail to adequately support users' search needs, for the split between search and navigation, and for the behaviour of symptom-based queries.
- Baymard Institute, E-Commerce Sites Must Support Users' Non-Product Search, accessed July 2026. Source for the 34% of users who attempt to search for non-product content and the 15% of sites that do not support it.
- Falia working framework, arithmetic of the annual loss tied to the failure rate. Amounts are explicit scenarios, to be redone with your own volumes.
Geneviève puts the strategy for your engagement into action. She leads all our web development projects: Shopify, WordPress and the new ways of building a site with AI. She manages our team of developers and translates your business needs into technical language. She runs your organic search (SEO), your visibility in AI answers (GEO) and your site's conversion rate optimization (CRO). Her work is at the heart of three goals: Attract customers with SEO and AI, Improve your site's conversion, and Strengthen your visibility in AI answers. With Gabriel, she also builds the landing pages for your advertising campaigns. She writes mainly about SEO, AI visibility and web design.
About Falia →
SEO
Paid advertising
Conversion
AI visibility
Web design