Search Synonyms That Backfire: When Mapping Goes Wrong

A synonym group is supposed to help shoppers find products. A careless one sends them to products they never wanted, and the search gets worse, not better.

Synonyms backfire when the mapping is too broad, one-directional when it should be two-way, or built from a guess instead of the search log. Map the words shoppers actually use, test the mapping, and remove the ones that pollute results.

Too broad is worse than none

The classic failure is mapping shoe to everything with a sole: boots, sandals, sneakers, slippers. A shopper who searches for shoe now gets the whole footwear catalog, and the search is useless. The mapping was meant to catch the shopper who types a generic word, but it caught everyone, and the result list is so broad it might as well be the homepage.

The rule is to map narrow to narrow. Trainers maps to sneakers, because the two words mean the same product. Shoe maps to nothing, because it means too many products. A synonym group should make the result list more accurate, not longer. If the mapping doubles the results, it is too broad, and the shoppers will tell you by bouncing.

Related reading on this site: RelevanceRail and Synonym Mapping: When Shoppers Use the Wrong Words for Your Products cover the neighbouring problems in more depth.

One direction is not enough

The second failure is direction. A store maps trainers to sneakers, so a shopper who types trainers finds sneakers. But a shopper who types sneakers does not find the products listed as trainers, because the mapping only runs one way. The fix is to map both directions, or better, to map both words to a shared group, so either word finds either product.

Shopify's synonym groups handle this when they are set up as groups rather than one-way redirects, but the setup is the part most stores skip. They add the mapping that fixes today's zero-result query and never check the reverse. The weekly test is simple: search for each word in the group and confirm the results match. If they do not, the group is half built.

Test the mapping before you trust it

Every synonym group deserves a test. Search for the word, read the top ten results, and ask whether a shopper who typed that word would want those products. If the answer is no for more than two of them, the mapping is wrong. The test takes a minute per group, and it catches the mappings that look right in the admin and fail on the storefront.

The other test is the click rate. A synonym group that sends shoppers to products they never click is polluting results, and the search log will show it as a query with results and no clicks. A query with results and no clicks is a synonym problem just as surely as a zero-result query is. Remove the mapping, and the clicks come back.

Frequently asked questions

When do search synonyms backfire?

When the mapping is too broad, one-directional, or built from a guess instead of the search log. A bad synonym sends shoppers to products they never wanted.

Should synonym mappings run both ways?

Yes. Map both words to a shared group, so a shopper who types either word finds either product. A one-way mapping fixes one query and leaves the reverse broken.

How do I test a synonym group?

Search for the word, read the top ten results, and ask whether a shopper who typed it would want those products. If more than two are wrong, the mapping is wrong.

What does a query with results but no clicks mean?

Usually a synonym problem. The mapping is sending shoppers to products they do not want, so they see results and click nothing. Remove the mapping and the clicks come back.