Skip to main content

Choosing the right Tag Search Strategy for your tag category

Describes the tag search strategies available to users and when to choose them.

Saul Bard avatar
Written by Saul Bard
Updated this week

Tag Search Strategy Guide

Overview

Tag search strategies help you control how strictly the response engine matches content to your tags. By setting the right strategy for each tag category, you can ensure the response engine uses the most relevant content while maintaining appropriate boundaries between different content areas.

Understanding Tag Categories

Tag categories are the different ways you organize your content (e.g., Product, Region, Department, Content Type, etc.). Each category can have its own match strategy, allowing you to be strict about some categories while flexible with others.

Available Search Strategies

1. Filter by tags - Only matching tags

When to use: When content separation is critical and cross-pollination must be avoided.

How it works: Only shows content that exactly matches or is a parent match to the tags selected on the project. No partial matches allowed.

Example use case: Financial institutions separating fund information, where mixing content between funds could cause compliance issues.

2. Filter by tags - At least one matching tag

When to use: When you want flexibility but still need at least some tag overlap.

How it works: Shows content that has at least one matching tag from this category. This is similar to traditional tag matching behavior.

Example use case: Product categories where related products can share some relevant information.

3. Prefer matches - High Importance

When to use: When exact matches are strongly preferred but expanding scope is acceptable.

How it works: All content is considered, but exact matches get significantly higher priority. Content with poor matches is heavily deprioritized.

Example use case: RFP responses where the specific product combination matters most, but general company information can still be useful.

4. Prefer matches - Medium Importance (Recommended Default)

When to use: For most tag categories where you want balanced behavior.

How it works: All content is considered with moderate preference for better matches. Provides a good balance between relevance and coverage.

Example use case: Department tags, general product categories, or most standard use cases.

5. Prefer matches - Low Importance

When to use: For metadata or supplementary categories that shouldn't heavily influence results.

How it works: All content is considered with minimal impact on prioritization based on tag matches.

Example use case: Content type tags (like "FAQ" or "Policy"), regions (in global organizations), or other "nice-to-have" categorizations.

How Strategies Work Together

The response engine evaluates content against all your tag categories:

  • For "Filter by tag" strategies, content must meet the matching requirements for each category to be included

  • For "Prefer matches" strategies, the quality of match affects how highly the content ranks

Best Practices

  1. Start with defaults: Begin with "Prefer matches - Medium Importance" for most categories and adjust based on results

  2. Review regularly: As your content grows, reassess whether your strategies still meet your needs

Quick Reference

Strategy

Strictness

Use When

Filter by tags - Only matching tags

Very High

Content mixing is prohibited

Filter by tags - At least one matching tag

High

Some overlap is okay

Prefer matches - High importance

Medium

Strong preference for matches

Prefer matches - Medium importance

Balanced

Standard categorization

Prefer matches - Low importance

Low

Supplementary metadata

Questions to Ask Yourself

For each tag category, consider:

  • How important is it that content exactly matches this category?

  • What happens if the engine uses content from adjacent tags?

By thoughtfully selecting tag search strategies, you can create a response engine that delivers relevant, appropriate content while respecting your organization's information boundaries.

Did this answer your question?