Zero results pages in site search: turning a dead end into a useful merchandising moment

Updated October 24, 2026 · DeftMerch answers

Short answer: A zero results page appears when site search finds nothing for a query. Some zero results are data problems you can fix: synonyms, spelling, attributes, or products described in different words from those shoppers use. The rest need a good fallback: close alternatives, popular categories, and a way to ask for help. Reviewing top zero results queries weekly is one of the fastest merchandising wins available, because the demand has already arrived and told you what it wants.

Read the queries before redesigning anything

Export the top zero results queries and sort them into groups. Misspellings and spacing variants of products you sell are a search configuration problem. Synonyms, such as trainers for sneakers or couch for sofa, are a vocabulary problem solved with synonym rules. Queries for products you used to sell, or sell in another channel, are a range or redirect decision. Queries for things you have never sold are demand intelligence.

Do not let the loudest query set the agenda. A single odd phrase searched once is noise. The value is in clusters: dozens of phrasings around a product, attribute, or use case you nearly satisfy. Those clusters often reveal that products exist but are missing the attribute shoppers search by, such as a size, material, or compatibility term that never made it into the product data.

Fix the product data, not just the search tool

Search can only find words that exist somewhere in your data. If shoppers search by a problem, a room, an occasion, or a fit, and your products are described only by range names and codes, zero results are guaranteed. Add the language of use to titles, descriptions, and attributes where it is true, and keep a governed synonym list that merchandisers review rather than a pile of one off rules added and forgotten.

Test fixes like a shopper. Search the original failing phrases, on mobile, and judge the first screen of results. A fix that returns the right product at position forty has not fixed anything. Relevance ranking after a data fix matters as much as the match itself, and it should be checked with the actual queries from your log, not invented examples.

Design a fallback that sells without pretending

For queries you cannot satisfy, be honest and useful in one breath. Say there are no matches for the exact phrase, then show the closest category, best selling neighbours, and a clear route to new arrivals or help. Do not silently substitute unrelated products and present them as matches. Shoppers notice, and a search that lies once is never trusted again in that visit.

Offer a capture point where it fits: notify me if this arrives, or an invitation to ask the team. Those requests are a demand list for buying and range planning, gathered for free. Review them with the same seriousness as sales data, because a product requested by searchers who found nothing is a product with proven, unmet intent behind it.

Make zero results a standing report, not a surprise

Add zero results rate and top failing queries to the weekly merchandising pack, with an owner and a fix log. The rate should fall as synonyms and data improve, and spikes should be explainable: a campaign that sent traffic hunting for a featured product, a range that sold out, or a navigation change that pushed people to search.

Celebrate the fixes visibly inside the team. Each resolved query cluster is revenue recovered from visitors you already had, with no advertising spend. That framing keeps attention on this unglamorous page, which quietly decides whether your most motivated visitors, the ones who knew what they wanted, leave with it or leave altogether.

Keep reading: Product badges that help shoppers and badges that just add noise, Clearance sections that sell without training customers to wait for discounts, and Search facets that help and facets that hurt: merchandising filters shoppers understand cover the nearby cases in this series.