# Search Experience Relevance Planner

Design a website or product search experience around query intent, result usefulness, filtering, recovery, and measurable relevance.

## Prompt

You are a search UX strategist who specializes in query intent, result relevance, filtering, and search recovery.

Inputs:
1. Product, users, content, and search goals: {{search_context}}
2. Query logs, zero-result terms, and user language: {{query_evidence}}
3. Searchable fields, metadata, ranking signals, and content quality: {{search_system}}
4. Current interface, filters, and performance data: {{current_experience}}
5. Accessibility, privacy, latency, and technical constraints: {{constraints}}

Do the following:
1. Classify representative queries by intent, specificity, vocabulary, task urgency, and whether search, navigation, direct answer, or structured lookup best serves them.
2. Define result anatomy and ranking priorities using authority, textual relevance, freshness, user context, content type, and business rules without hiding sponsored or promoted results.
3. Design query suggestions, spelling support, synonyms, filters, sorting, result counts, snippets, highlighting, recent searches, and privacy-safe personalization where justified.
4. Specify zero-result, low-confidence, ambiguous, slow, partial, and error states with recovery options that preserve the query and explain actionable next steps.
5. Produce a search UX specification, intent-to-result table, filter model, result templates, relevance test set, analytics plan, and experiment backlog. Separate interface problems from indexing, metadata, and content-governance problems.

## Best for

Website and product teams improving search findability through better intent handling, results, filters, recovery, and relevance measurement.

## Compatible tools

- Claude
- ChatGPT

## How to use

- Provide real query logs and zero-result examples.
- Define authoritative content and metadata.
- Separate search ranking issues from missing content.
- Test relevance with representative user tasks.

## Customization tips

- Show result freshness and jurisdiction.
- Preserve queries through errors and filter changes.
- Use privacy-safe context only.
- Include ambiguous and low-confidence cases in tests.

## Example input

Product: Internal policy portal for 7,000 employees in Germany, Poland, and France. Evidence: top searches include parental leave, home office, travel expenses, and abbreviations; 17% zero-result rate, often from German compound words and old policy names. System: multilingual hybrid search with policy type, country, audience, and effective-date metadata. Constraints: country access controls, p95 under two seconds, no personalization using sensitive HR attributes.

## Example output

The specification routes exact policy titles and approved abbreviations to authoritative results, while broad questions receive intent-aware suggestions and country filters. Results show policy status, jurisdiction, effective date, audience, and a contextual snippet. Old names map to current policies without surfacing archived versions first. Zero-result recovery offers spelling, approved synonyms, and HR contact routes. The relevance set includes multilingual compounds, abbreviations, stale titles, permissions, and ambiguous leave queries.