ParticleSearch ParticleSearch
Skip to article
Search Configuration Jun 17, 2026 20 min read

Synonym Groups for Ecommerce Search: A Governance and Testing Guide

A synonym group is a claim that several terms are interchangeable for a specific catalogue intent. When that claim is correct, shoppers can use their own language. When it is loose, one ambiguous term spreads irrelevant products across every other term in the group.

This guide covers the editorial and testing system behind safe synonym groups. The step-by-step Shopify admin path lives in the setup guide. Shopify’s current synonym requirements and analytics scope were checked July 28, 2026.

The short answer

Create a group only when every term should retrieve the same intended product set. Write the boundary, test every direction, include negative queries, and preserve a control query. A lower zero-result rate is not a win if the group introduces irrelevant results.

Chapter 1 · Understand the graph

Every added term expands the meaning of the whole group

Shopify treats terms in a synonym group as exact substitutes. Think of the group as a fully connected graph: if “sneakers,” “trainers,” and “athletic shoes” share one group, every term is connected to every other term. Adding “boots” does not create one new relationship; it tells search that boots are equivalent to all three existing terms.

Vocabulary graph

One ambiguous term changes every path

Green relationships are intended; the blush term creates unsafe equivalence.

A synonym group as a connected graphSneakers, trainers, and athletic shoes are safely connected. Adding boots creates connections to every term and introduces a category mismatch.sneakerstrainersathleticshoesbootsA four-term group creates six pairwise equivalence claims.
A conceptual model of group semantics, not Shopify’s internal implementation.
Chapter 2 · Choose true substitutes

Good groups preserve one buying intent across different vocabulary

PatternExampleWhy it can workTest
Regional vocabularysneakers · trainersDifferent regional words can name the same product concept in this catalogue.Confirm both terms should return the same product set in every market where the group applies.
Category aliassofa · couchTwo common category names can be interchangeable when the store does not distinguish them.Search each term and inspect categories, product types, and the first useful results.
Acronym and full formPPE · personal protective equipmentA known acronym may represent the same buying category.Check that the acronym does not have another meaning in the catalogue.
Brand formattingray ban · Ray-BanSpacing and punctuation can differ while identity remains the same.Verify the brand owner, product set, and any built-in normalization before adding a group.
Trade and consumer termhex key · Allen keyThe buying job can be identical even when audiences use different terminology.Check regional trademark and catalogue meaning; avoid merging tools that merely look similar.

The example is never the evidence. Your product set decides whether the terms are substitutes. A store that separates “sofas” from “couches” by product type should not copy a group from a store that uses the words interchangeably.

Chapter 3 · Keep other problems out

Related words, misspellings, identifiers, and destinations need other controls

shoe · boot

Related categories, not exact substitutes.

Better owner: Category structure or broader matching

phone · case

Complementary products, not equivalent intent.

Better owner: Recommendations or bundles

red · burgundy

A parent colour and a specific shade may not be interchangeable.

Better owner: Attribute normalization or filter grouping

AB-12 · AB-21

Near identifiers can refer to different sellable products.

Better owner: Exact identifier policy

sneekers · sneakers

One is a misspelling, not alternative vocabulary.

Better owner: Typo tolerance

return policy · returns

The intent may be a page rather than a product category.

Better owner: Navigation or redirect

laptop · electronics

One is narrower than the other.

Better owner: Taxonomy, category browse, or semantic search

If the failure is a misspelling, use the typo-tolerance guide. If the query is an exact SKU or barcode, synonyms cannot repair it on Shopify. If it asks for a policy or help page, route the intent rather than polluting product vocabulary.

Chapter 4 · Write the decision record

A group should be explainable without opening the admin

1

Decision

Problem statement

Record

The raw failed or weak query, surface, market, expected products, and why vocabulary—not visibility or field coverage—is the cause.

Prevents

Creating groups from intuition alone.

2

Decision

Meaning boundary

Record

A sentence defining when every term is interchangeable and when it is not.

Prevents

Polysemy and category overreach.

3

Decision

Positive set

Record

Representative products that every term should retrieve.

Prevents

A group that fixes one product while missing the rest of the intent.

4

Decision

Negative set

Record

Products or categories that must not appear because of the group.

Prevents

False positives hidden by a lower zero-result rate.

5

Decision

Owner and review trigger

Record

Who approved the group and what catalogue, market, or query change requires a new test.

Prevents

Orphaned vocabulary rules that nobody can explain.

6

Decision

Outcome

Record

Before/after result relevance, no-result behavior, clicks, purchase rate where available, and replay date.

Prevents

Calling a group successful because it saved without error.

Chapter 5 · Build the judged query set

Test relevance gained and relevance lost

A useful test does not stop when the failed query returns products. It asks whether each group direction works, whether ambiguous contexts stay clean, and whether unrelated control queries remain stable.

TestPassFail
Every term, both directionsEach term retrieves the intended products.One catalogue term expands to products that only fit another term.
Exact phraseA multi-word synonym behaves as the intended phrase.A word inside the phrase broadens unrelated queries.
Ambiguous contextA second meaning remains outside the result set.“Boot” pulls computer startup products into footwear.
Mixed queryAdded attributes still narrow the intended category.“red trainers” returns every red product because the group is too broad.
Predictive and results pageThe required surfaces behave according to the written expectation.The full page passes while the dropdown still fails or ranks unrelated products.
Control queryAn unrelated working query remains stable.The group changes a result set that was not part of the repair.

Positive query

A term that should recover the intended set.

Negative query

An ambiguous context that must not broaden.

Control query

A working result set that should not change.

Chapter 6 · Publish one group

Make synonym changes like small search releases

  1. 01

    Select

    Choose one verified vocabulary gap where the intended product set is known.

  2. 02

    Define

    Write the meaning boundary and the positive, negative, phrase, mixed, and control queries.

  3. 03

    Configure

    Create the smallest group that expresses the equivalence; do not fill the documented limit.

  4. 04

    Preview

    Run the full judged set on the storefront surfaces that matter.

  5. 05

    Publish

    Save the group with owner, date, source query family, and rollback condition.

  6. 06

    Measure

    Compare relevance and the affected results-page query family, then review predictive behavior separately.

Current Shopify boundaries

Ceilings are not recommendations

The limits below come from Shopify’s synonym requirements checked July 28, 2026. Keep groups as small as the meaning allows.

1–5

words per synonym

20

synonyms per group

1,000

store-wide synonyms

No

reuse across groups

Source: Shopify Search & Discovery settings.

Questions that change the group

Synonym questions, answered by meaning and evidence

Are Shopify synonym groups bidirectional?+

Shopify describes terms in a group as exact substitutes. Test every term as the shopper query because the business meaning can still be asymmetric even when the search rule treats the terms as equivalent.

Source: Shopify documentation

How large can a Shopify synonym group be?+

As of July 28, 2026, Shopify documents one to five words per synonym, up to 20 synonyms in a group, up to 1,000 synonyms store-wide, and no reuse of the same synonym across groups. Treat those as ceilings, not design targets.

Source: Shopify documentation

Do synonyms apply to SKU and barcode searches?+

No. Shopify excludes SKU and barcode matches from synonym processing. Exact identifiers need a field-coverage and normalization policy.

Source: Shopify documentation

Should misspellings go into synonym groups?+

Usually no. Misspellings belong to typo recovery, while synonyms join correctly spelled terms with the same meaning. Keeping them separate makes rules easier to test and prevents large maintenance lists.

Source: Shopify documentation

Do I need synonym groups for singular and plural forms?+

Shopify says common singular, plural, and standard spelling variations generally do not need synonym groups. Verify your catalogue, then reserve groups for genuine vocabulary differences.

Source: Shopify documentation

How do I know a synonym group improved search?+

Replay the judged positive, negative, phrase, mixed, and control queries. Then compare the affected query family using searches-with-no-results, searches-with-no-clicks, click rate, and purchase rate where available. Predictive interactions require a separate check.

Source: Shopify documentation

Use the Shopify setup guide when the group is ready to configure. Start with the zero-result intelligence framework when the candidate vocabulary is not yet verified, and the zero-results diagnostic when product existence or field coverage is unresolved.

When the vocabulary is real but native synonym governance is too narrow for the store’s query set, ParticleSearch is a fit because synonyms, redirects, and ranking decisions live in one merchant workflow. After verification, you no longer need to maintain disconnected language fixes or wonder which rule changed a result. The query-tools guide shows the positive, negative, and control queries that should pass before publishing.