Search in Elevate is powered by advanced logic that combines product data, customer intent, and business rules. Because of this, search issues are rarely caused by a single factor. This guide helps you understand how to investigate and resolve common search challenges in a structured way. By following these steps, you can identify whether the issue is related to product data, search logic, ranking, or manual configurations.
How it works
Before troubleshooting, it’s important to understand that Elevate search is built on three core components:
- Product intent, based on how your products are classified
- Customer intent, based on how a search query is interpreted
- Business intent, based on ranking logic and manual controls
Each search goes through several steps:
- Your product data is classified and mapped to our knowledge graph.
- The search query is broken down into concepts such as product type, brand, color, or other attributes.
- Relevance is calculated based on how well products match the interpreted query.
- Product ranking is applied, including sales performance, trend sensitivity, lifecycle logic, availability, and any manual boosts or pinning.
Because multiple layers interact, troubleshooting requires checking each layer methodically.
How to investigate a search issue step by step
Confirm the search query
Start by identifying the exact search phrase used. Small differences in spelling, language, or structure can impact results.
Consider:
- Is the search phrase spelled correctly?
- Is it a mix of languages?
- Is it very broad or very specific?
If misspellings or variations are common, Elevate may already compensate. However, heavy corrections can slightly lower relevance scores. If the misspelling is too severe, Elevate will not be able to recognize it. Normally elevate has autocorrect that should be implemented on site.
Check product data quality
Many search issues originate from the product data feed.
Review the affected products and verify:
- Correct product type
- Accurate brand
- Proper attributes such as color, material, or size
- Valid images
- Correct stock status
- Accurate list price and selling price
If incorrect data is sent to Elevate, the system will classify and rank based on that data. Updating the product feed is often the first and most important correction.
Review classification and attribute mapping
Elevate classifies products using a knowledge graph and classifier logic. In some cases:
- A product may be misclassified
- An attribute may be missing or incorrectly inferred
- A concept may be too general or too specific
For example, if a product is classified as "clothing" instead of "tank top," it may rank lower in a specific search. If you suspect classification issues, contact Support so we can evaluate whether adjustments are needed in the knowledge graph.
Evaluate search intent interpretation
Search queries are broken down into structured concepts such as:
- Product type
- Brand
- Color
- Shape or fit
- Occasion
- Free text
If results seem incorrect, ask:
- Did Elevate correctly identify the product type?
- Is the brand correctly interpreted?
- Is a specific attribute overly restrictive?
In some cases, search relaxation may apply. If no exact match exists, Elevate removes less important attributes and returns secondary matches. This helps avoid zero-result searches, but may cause results that are not perfect matches.
Check for "No hits" in the search report
The Search Report is one of your most powerful troubleshooting tools.
Focus on three areas:
- Top searches
- Trending searches
- No hits
If a term is not searched for so often, it may not be worth investigating the reasons. It is always best to focus on frequently searched terms.
If a frequently searched term returns no results:
- Confirm that relevant products exist in the catalog
- Check whether the terminology matches your product data
- Consider whether a synonym or keyword is appropriate
Establish a regular cadence for reviewing search data, such as monthly or quarterly. Avoid acting on one-off searches. Look for repeated behavior before making changes.
How to use synonyms correctly
In Elevate 4, true synonyms are managed within our dictionaries. You normally do not need to add them manually.
Use synonyms when:
- A search phrase does not return expected results
- The term is business-specific and not a true linguistic synonym
- You need a temporary workaround while Support reviews the case
For example: If customers search for "CK" but your products use "Calvin Klein," you may add a synonym so that searching for "CK" also searches for "Calvin Klein."
Important:
- Synonyms expand searches
- They are applied site-wide per language
- They should not override core product concepts
Do not use synonyms to replace product types, core attributes, or standard definitions. If Elevate does not understand a product type or concept, contact Support instead of forcing it with a synonym.
How to use keywords carefully
Keywords in the primary list enrich products directly and act as strong signals.
Use keywords when:
- Running campaigns such as Black Friday
- Promoting a limited collection
- Making a temporary concept searchable
Keywords are added to a Primary List, not to the page itself. They apply to all products in that list.
Important considerations:
- Keywords are exact matches
- Variations must be added individually
- They override parts of the relevance logic
Avoid using keywords for:
- Core product types
- Standard attributes like colors
- Concepts that should already exist in the knowledge graph
Overusing keywords can distort search relevance and lead to undesirable results.
How ranking impacts search results
Even when products match a search query, their position depends on ranking.
Ranking is influenced by:
- Clicks (if in-session personalization is activated)
- Add to cart actions (if in-session personalization is activated)
- Purchases
- Trend sensitivity
- Newness settings
- Price changes
- Stock status of generally available stock
- Manual boosts or pinning
- Out of stock setting (bury or show out of stock products)
If a product appears but is lower than expected:
- Check whether it is out of stock or has broken variants
- Verify whether it has recent engagement
- Confirm whether another product has been boosted
Boosting increases visibility relative to other products but still respects ranking logic. Pinning forces fixed positions and overrides ranking entirely.
How to identify front-end vs search issues
Sometimes the issue is not in Elevate but in the front-end implementation.
If possible:
- Validate the raw search response
- Compare results in different environments
- Confirm filters and sorting are not altering the expected outcome
In some cases, custom front-end logic may conflict with Elevate’s intended behavior.
When to contact Support
Contact Support when:
- You suspect classification errors
- A product type or concept is missing in the knowledge graph
- True synonyms are not covered
- Search results contradict expected logic after data validation
Provide:
- The exact search phrase
- Example product IDs
- Description of expected behavior
- Screenshot if relevant
This allows us to evaluate whether the issue is data-related, logic-related, or knowledge graph-related.
Search troubleshooting examples
- When searching for sweaters, why do I get cardigans as well in the search result?
Cardigans are connected to sweaters in Elevate's ontology. Most likely the pure sweaters are higher up in listings than cardigans due to the match score. - When searching for pants we get an unexpected result of winter ski pants - why?
Seasonal trending products, for example winter ski pants, are related to pants and are trending at the moment. - Why is product X at the top of listings?
This would need investigation. It could be pinning, boosting, or ontological match vs text match. - Why isn't product Y at the top of ranking when it is our bestseller over time.
Elevate is trend sensitive and will promote products that are currently selling well and taking into account other factors impacting ranking. Purchases are just one factor impacting the entity score of the product ranking. - Why don't I see the boosted products in the search result?
Probably because boost tracking is activated on the boost. Remove tracking to see 100% session impact of the boost.
Summary
Troubleshooting search in Elevate requires a structured approach:
- Verify the search query
- Validate product data
- Review classification
- Check search intent interpretation
- Analyze ranking
- Use synonyms or keywords carefully
- Review Search Reports regularly
Because Elevate combines classification, intent analysis, and ranking logic, quick fixes are rare. A methodical review ensures that changes improve search performance without negatively impacting overall relevance.
Article last updated