Investigate incorrect product results in shop search
Check product number, term mapping and destination product using the same search cases. A worked test list shows which change should fix which error.
When shop search shows the wrong product, follow a specific query through to the item that opens. More results help only if the suitable variant is among them and can be selected correctly. Start with a small test list rather than changing the whole search indiscriminately.
Record the search case and expected product
The fictional filter shop sells insert F-200 and separate housing H-200. For this catalogue, the product owner has confirmed that “filter insert” and “filter cartridge” mean the same insert. This is a decision within the example, not a general equivalence between all filter products.
Search query | Expectation for the planned test | Potential error to check |
|---|---|---|
| The correct insert is unambiguously findable. | Product number in the searched dataset and link to the destination product |
“filter cartridge” | The same insert as for “filter insert” is findable. | Confirmed term mapping and a search rule that is actually supported |
“filter housing” | Separate housing | Overly broad term rule or inappropriate promotion |
| Clear empty-result message; no invented item | No-results display and the next step offered |
The table contains requirements for a trial, not measured results. During the test, also record the actual result and the product displayed after opening it.
Check the dataset before the term rule
Shopify documents searchable product fields and product numbers. Applicable rules also depend on the search application being used. First establish which service answers the query and which data it searches.
If your search uses an index, a dataset prepared for searching, check the current item there. A correct product in the administration system may still be missing from an outdated search copy. Also check language, visible assortment and the mapping from a result card to the actual variant.
Change term mapping selectively
Elasticsearch describes equivalent and directional search-term rules. This is an implementation example, not a promise that every shop has the function. Check your system's capabilities before planning a rule.
In the filter example, the confirmed mapping may make the insert easier to find. It must not define the housing as the identical item. After a change, run both the desired match and the counterexample. Preferential placement does not fix an incorrect product link.
Repeat the entire selection journey
Test suggestions during typing separately from the later results page. Open the result, select the variant and check the cart. For the example shop, the change is appropriate only if F-200 remains the same insert throughout and H-200 does not silently replace it.
Sources
Read more
All articles
E-Commerce with ASES
Online shops, marketplaces and B2B portals