# How should product search handle shoppers who describe what they want instead of naming it? | KilterMuse

Source: https://kiltermuse.com/blog/2026-10-18-search-shoppers-describe-not-name/

KilterMuse answers


October 18, 2026

**Short answer:** A shopper types the problem. The catalog answers in product language. When search cannot translate between the two, the shopper concludes the store does not sell the thing it absolutely sells, and leaves to describe the same need to a competitor.

## Descriptive search is a large, quiet share of queries

First time visitors, gift buyers, and anyone new to a category describe instead of naming: warm waterproof boots for winter dog walking, a dress for a wedding that is not white, storage for a tiny kitchen. These queries are longer, more specific, and full of purchase intent, and keyword matching handles them worst of all.

Zero result and poor result reports usually hide this segment, because the shopper did search, got irrelevant matches for one word in the phrase, and left. The failure looks like browsing in analytics. It was a translation failure.

## Attributes are the bridge, if the data exists

Translation works when products carry rich structured attributes: use, material, fit, occasion, compatibility, warmth, capacity, and the rest of the properties descriptive queries name. A search engine cannot match a need to a product whose data never recorded that the product meets it.

Attribute cleanup is unglamorous and compounding. Every product enriched with honest attributes becomes findable by every descriptive query that names them, forever. It is some of the highest returning catalog work a store can do, and it improves filters and recommendations at the same time.

## Synonyms and language layers do the rest

Shoppers say trainers where the catalog says sneakers, settee where it says sofa, and a hundred trade specific variations. A maintained synonym layer, grown from real query logs, teaches search both vocabularies. Query logs are the curriculum: every descriptive search that failed is a lesson written by a customer for free.

Modern search tools add semantic matching that handles description far better than keywords alone. They still depend on your data quality, and they still need review: semantic systems can be confidently wrong in ways a shopper spots instantly, matching the vibe of a query while missing its hard constraint.

## Results pages should confirm the interpretation

When search interprets a descriptive query, show the interpretation: filters it applied, attributes it matched, phrased as the shopper can correct. A line that says showing waterproof boots in your size builds trust and gives the shopper a handle to refine, instead of a mystery list they must audit item by item.

Offer the vocabulary back. Suggesting the category name and related terms as the shopper browses teaches them how your store talks, so their second search is sharper. Good search is partly a translation service and partly a tutor.

## Measure descriptive search as its own segment

Separate long, descriptive queries in reporting and watch their conversion and exit rates against name based searches. The gap between the two is the size of the translation problem, and it usually survives as the largest recoverable loss in the whole search funnel.

Stores that close that gap do not just sell more. They learn how customers describe needs, and that language improves product pages, guides, and ads everywhere else it is reused.

Keep reading: [How do recommendation engines create filter bubbles that shrink what shoppers discover?](https://kiltermuse.com/blog/2026-10-18-recommendation-filter-bubbles-discovery), [What should recommendations do when the product they would suggest goes out of stock?](https://kiltermuse.com/blog/2026-10-18-recommendations-out-of-stock-product), and [How do you personalize email product blocks without rebuilding every send?](https://kiltermuse.com/blog/2026-10-18-personalize-email-product-blocks-sends) cover the nearby cases in this series.

## Frequently asked questions

### Why do descriptive searches fail on many stores?

Because the shopper describes a need in ordinary language while the catalog uses product vocabulary, and keyword search cannot translate between them. The shopper got irrelevant matches and left, even though the right product was in stock.

### What data makes descriptive search work?

Rich product attributes: use, material, fit, occasion, compatibility, and capacities, recorded honestly per product. Search can only match a need to a product whose data says the product meets that need.

### How do stores learn shopper vocabulary?

From query logs. Every failed or descriptive search shows the words real shoppers use, which builds the synonym layer between shopper language and catalog language. Suppliers and trade terms add the other direction.

### How should results show that search understood the query?

By displaying the interpretation: the filters and attributes matched, in terms the shopper can correct. That confirmation builds trust and teaches the shopper the store vocabulary for the next search.