Search Configuration Settings

Search configuration controls how results are ranked, weighted, and displayed across your platform. By tuning these settings, you can make your most important assets easier to find — surfacing trusted, frequently used data at the top of results while reducing noise from less relevant matches. You can adjust search behavior at two levels:

Global level — rules that apply to every search across all asset types on the platform. See Global Settings.
Asset type level — rules that apply to one specific type, such as Tables, Dashboards, or APIs. See Asset Type Level Search Configuration.

Start with global settings to establish a consistent baseline, then use asset-type settings to fine-tune search for your most critical asset categories.

How to Access

To access and configure these settings, navigate to Settings > Preferences > Search.

Global Settings

Global settings control how search behaves across your entire platform. Changes here affect every asset type, every search query, and every user.

Configurable Options

These settings control platform-wide performance limits and access rules for search. In most cases, the defaults work well for smaller deployments. Increase the limits if your platform contains a large number of assets and users report incomplete or truncated results.

Setting	What it does	Default
Enable Roles & Policies in Search	When turned on, each user’s search results are filtered to show only assets they have permission to access. Turn this on if your platform has data governance or access control requirements.	Off
Max Aggregate Size	Sets the maximum number of grouped results returned in a single search. If users search across many assets and results appear cut off, try increasing this value.	10,000
Max Result Hits	Sets the maximum number of individual results returned in a single search. Raise this limit if searches are returning fewer results than expected.	10,000
Max Analyzed Offset	Controls how much of a text field’s content is scanned during search. If important keywords in long descriptions aren’t appearing in results, try increasing this value.	1,000

Term Boost (Global)

Term boost lets you promote assets with a specific tag so they appear higher across all search results — regardless of asset type. Select a tag from the dropdown, then use the slider to set how strongly it should be promoted. A higher boost value pushes those assets further up the results list. Example:

Tag	Boost value
Tier.Tier1	0.05
Tier.Tier2	0.03
Tier.Tier3	0.01

In this setup, Tier 1 assets appear above Tier 2 and Tier 3 assets when all other relevance signals are equal. This is useful when you want to steer users toward trusted, production-ready data.

Field Value Boost (Global)

Field value boost promotes assets based on a numeric field — such as usage count or votes — so that frequently used or highly rated assets naturally rise in search results. It works the same way as the asset-type field value boost but applies platform-wide, across every asset type. To configure it, open Settings > Preferences > Search and use the Add, Edit, and Delete options in the Field Value Boost section. Changes here apply to every asset type. Any field value boost set at the asset-type level is applied in addition to the global one — both contribute to an asset’s final ranking. The following defaults are seeded by the system and apply globally unless changed:

Field	Impact	Modifier	Missing value
`usageSummary.weeklyStats.count`	0.002	log1p	1
`totalVotes`	0.005	log1p	0
`usageSummary.weeklyStats.percentileRank`	0.0005	none	0

Asset Type Level Search Configuration

Asset type settings let you fine-tune how search behaves for a specific kind of asset — for example, Tables, Dashboards, or Pipelines — without affecting how other types behave. Each section has an Add, Edit, and Delete option so you can build out your configuration incrementally. If something isn’t working the way you expected, select Restore Defaults to reset the entire search configuration — this replaces the full searchConfig with system defaults across all sections, not just the current one.

Search Preview

The asset-type settings screen is split into two panels. The left panel contains your configuration sections — Matching Fields, Term Boost, and others. The right panel is a live Preview pane that shows real search results based on your current (unsaved) settings. Type a query into the Search for Data Assets bar in the preview pane to see which assets would be returned. Each result shows the asset’s name, breadcrumb path, description, domain, owners, tier, and a SCORE value so you can see exactly how your configuration is influencing relevance ranking. The preview updates as you adjust matching fields, weights, or boost settings — without affecting live search results for other users. Once the results look right, click Save to apply your changes. If you want to undo all changes and return to the system defaults, click Restore Defaults in the preview pane header.

Matching Fields

Matching fields define which parts of an asset’s metadata are considered when your search query runs. For example, if description is included as a matching field, searching for a keyword will also look inside asset descriptions — not just names.

To add a new matching field, click Add.
To configure the settings in an existing field, select the respective matching fields card.

For more information about various matching fields available, see Available Matching Fields.

Matching Fields Settings

Each field has four settings:

Setting	What it does
Field name and description	The piece of metadata to search against — for example, `name.keyword` searches the exact asset name, while `description` searches the asset’s description text.
Match type	Controls how strictly the search query must match the field’s value. For match type definitions, see Match Types.
Weight	Determines how much influence this field has on the overall relevance score. A field with a weight of 20 contributes far more to ranking than one with a weight of 2.
Highlight fields	When turned on, matching text in this field is highlighted in search results, making it easier for users to see why an asset appeared.

Match Types

Match type controls how flexible or strict the comparison between your search query and a field’s value is. Choosing the right match type for each field helps you return accurate results without overwhelming users with noise.

Match type	What it does
Exact match	Returns results only where the field value matches your search term character-for-character. Best for fields like `name.keyword` where users know exactly what they’re looking for and partial matches would be misleading.
Phrase match	Returns results where all your search words appear together in the same order. Use this when the full phrase matters — for example, “customer orders” should rank higher than assets where “customer” and “orders” appear separately in different places.
Fuzzy match	Returns results even when the search query contains minor typos or misspellings. Useful for name fields where users may not recall the exact spelling. Use this sparingly — applied to too many fields, it can return a large number of irrelevant results.
Standard match	A flexible general-purpose match that finds close variations of the search term. Works well for descriptions, free-text fields, and any field where the exact wording is less predictable.

Available Matching Fields

The table below lists the fields you can configure, along with their default weights and match types. Table is used as an example — other asset types follow the same pattern with fields suited to their structure.

Field	What it searches	Default weight	Default match type
`displayName.keyword`	The display name — exact spelling only	20	Exact
`name.keyword`	The asset name — exact spelling only	20	Exact
`name`	The asset name — flexible text search	10	Phrase
`name.ngram`	The asset name — matches partial words (for example, “cust” matches “customer”)	1	Fuzzy
`name.compound`	The asset name — matches joined or compound words	8	Standard
`displayName`	The display name — flexible text search	10	Phrase
`displayName.ngram`	The display name — partial word matching	1	Fuzzy
`displayName.compound`	The display name — joined or compound words	8	Standard
`description`	The asset’s description text	2	Standard
`fullyQualifiedName`	The full path name of the asset (for example, `service.database.schema.table`)	5	Standard
`fqnParts`	Individual segments of the full path name, searched separately	5	Standard
`columns.name.keyword`	Column names — exact spelling only	2	Exact
`columns.displayName.keyword`	Column display names — exact spelling only	2	Exact
`columnNamesFuzzy`	Column names — fuzzy matching for partial or misspelled column name queries	1.5	Standard

These fields are available across most asset types and can be added from the field picker:

Field	What it searches
`tags.tagFQN.text`	Tag names — useful for surfacing assets tagged with labels like “PII” or “Sensitive”
`tier.tagFQN.text`	Tier classification names
`certification.tagLabel.tagFQN.text`	Certification label names
`domains.displayName.keyword`	The name of the domain the asset belongs to
`dataProducts.displayName.keyword`	The name of the data product the asset belongs to

Asset-Type-Specific Fields

Some asset types have additional fields that reflect their unique structure. These are only available when configuring search for that specific asset type. Dashboard

charts.name — Names of charts included in the dashboard
charts.description — Descriptions of charts included in the dashboard
charts.tags.tagFQN.text — Tags applied to individual charts within the dashboard

Topic

messageSchema.schemaFields.name.keyword — Exact name of a field in the topic’s message schema
messageSchema.schemaFields.description — Description of a field in the topic’s message schema

Pipeline

tasks.name — Names of tasks within the pipeline
tasks.description — Descriptions of tasks within the pipeline

Container

dataModel.columns.name — Column names in the container’s data model
dataModel.columns.name.keyword — Column names — exact match only

ML Model

mlFeatures.name — Names of features the model uses as inputs
mlFeatures.description — Descriptions of features the model uses as inputs

API Endpoint

responseSchema.schemaFields.name.keyword — Exact name of a field in the API response schema
requestSchema.schemaFields.name.keyword — Exact name of a field in the API request schema

Query

queryText — The full SQL text of the query
queryText.ngram — Partial word matching within the SQL text

Glossary Term

synonyms — Alternative names or synonyms for the glossary term
synonyms.ngram — Partial word matching on synonym text
glossary.name — The name of the glossary the term belongs to

Score Mode and Boost Mode

When multiple boost rules are active at the same time, score mode and boost mode control how they’re combined and applied to the final result order.

Setting	Options	What it does
Score mode	`Avg`, `First`, `Max`, `Min`, `Multiply`, `Sum`	Determines how multiple active boosts are combined into a single boost value. Sum adds all active boosts together. Max uses only the highest single boost. Avg averages them.
Boost mode	`Avg`, `Max`, `Min`, `Multiply`, `Replace`, `Sum`	Determines how the combined boost value is applied on top of the base relevance score. Multiply scales the base score by the boost — the default, and usually the most effective for promoting popular assets. Replace ignores the base score entirely and uses only the boost value.

Defaults: Score mode = Sum, Boost mode = Multiply

Term Boost

Term boost for an asset type works the same way as global term boost, but only affects results for that specific asset type. Select a tag from the dropdown and use the slider to set the boost value. Assets with that tag will appear higher in results when users search within that asset type.

Field Value Boost

Field value boost lets you rank assets higher based on a numeric metric — for example, how many times an asset was used last week, or how many upvotes it has received. This helps surface active, well-regarded assets without requiring users to filter manually. To add a field value boost, click Add and configure the following options:

Configuration options:

Option	What it does
Field	The numeric field to use for ranking — for example, `usageSummary.weeklyStats.count` boosts assets based on how often they were accessed in the past week.
Impact	Controls how much the field value influences the final score. A higher impact value means usage or votes will push results up more aggressively.
Modifier	A mathematical adjustment applied to the field value before it’s used in scoring. `log1p` is recommended for usage counts because it reduces the outsized effect of very large values — an asset used 10,000 times shouldn’t rank infinitely higher than one used 5,000 times. Other options include `none`, `log`, `log2p`, `ln`, `ln1p`, `ln2p`, `square`, `sqrt`, and `reciprocal`.
Missing value	The score to assign when the field is empty or not present on an asset. Typically `0`, meaning assets without that field won’t receive a boost — though some seeded defaults use `1` (for example, the usage count boost).
Range condition	Optionally limits the boost to assets where the field value falls within a specific range. For example, you can boost only assets with more than 100 weekly uses. Options include `Greater than`, `≥`, `Less than`, and `≤`.

Restore Defaults

Selecting Restore Defaults resets the entire search configuration for your platform by replacing the full searchConfig with the original system defaults. This resets all sections at once — matching fields, term boost, score mode, boost mode, and field value boost.

Best Practices

These recommendations help you get the most consistent and useful results from search configuration.

Use Exact match on .keyword fields like name.keyword when you want only precise name matches to count. This is useful when your assets have highly similar names and partial matches create confusion.
Use Phrase match on name and displayName when word order matters. For example, “customer orders” should rank higher than an asset where the same two words appear in different fields.
Use Fuzzy match sparingly. It helps users find assets despite typos, but applying it to too many fields increases the chance of irrelevant results appearing.
Use Standard match for description fields and other free-text content where users’ exact phrasing is unpredictable.
Use Term boost to promote assets with important tags — such as a tier level or certification — so they consistently appear near the top of results for all users.
Use Field value boost to ensure frequently used or highly rated assets surface first, especially in large catalogs where relevance alone may not distinguish quality.
Avoid assigning very high weights to many fields at once. When too many fields have high weights, each individual boost contributes less, and the overall ranking becomes harder to reason about.

Overview

Admins

Search Configuration Settings for Relevance, Ranking, and Field Boosting

Search Configuration Settings

How to Access