Lumenare Search provides instant, real-time search results as users type, with a dropdown showing the most relevant results without page reloads.
Current release: 2.2.01
The dropdown shows search results. Ask Site and results-page answer cards have separate placement controls under Search & Answers.
How It Works
When a user types in the search field:
- After minimum characters (default: 3) and the configured typing delay, search triggers
- AJAX request sends query to server
- Results are processed through the search index
- Most relevant results appear in dropdown
- Results update as user continues typing
Configuration
General Settings
Open Settings → Lumenare Search → General, under Configure search. On smaller screens, choose General from the Settings section menu.
Enable Plugin: Turn live search on/off globally
Search Placeholder: Text shown in search input field
- Example: "Search…", "Find posts…", "Type to search…"
Minimum Characters: Number of characters before search triggers
- Default: 3
- Recommended: 3-4
- Lower = more responsive but more queries
- Higher = fewer queries but less responsive
Live Search Trigger Delay: Delay after typing before live results are requested
- Default: 250ms
- Recommended: 250-500ms for most sites
- Lower = more immediate feedback
- Higher = fewer AJAX requests on busy sites
Display Settings
Open Settings → Lumenare Search → Display.
Dropdown Results: Number of results shown in live dropdown
- Default: 10
- Supported range: 1-10
- Recommended: 5-10 depending on site size
- Fewer results = faster queries
- More results = better discovery
Live Result Images: Show featured-image thumbnails in live dropdown results
- Default: Disabled
- Recommended: Enabled for visual/editorial sites
- Results without featured images still display normally
- Image-enabled and text-only responses use separate cache keys
Enable Search Term Highlighting: Highlight matching terms in results
- Recommended: Enabled
- Improves user experience
- Shows why results match
Search Understanding and Meaning Reranking
Open Search & Answers → Search behavior and review the Live dropdown row. Keyword search and Understand questions use local processing. Question understanding removes filler and expands approved site vocabulary without contacting an AI service.
Live typing stays local when it merely inherits a global meaning-reranking preference. To permit embedding-provider requests while visitors type, explicitly choose Meaning reranking (needs AI) for Live dropdown and save. Check the assigned profile and Actual search behavior column for overrides or fallback reasons.
Meaning reranking needs an approved embedding service, saved model and compatible current public passages. Use Search & Answers → Advanced → AI service and model for content analysis to choose and save the service/model, then run Analyze site content. That preparation sends missing or changed public passages to the provider and stores vectors locally; keep the tab open, watch progress, and use Pause or Resume analysis if needed. Adding an API key or saving these choices does not enable reranking or start analysis.
An opted-in typing request considers at most 20 existing hits and may send an uncached query to the provider for an embedding. Query embeddings use a bounded five-minute cache and rate limits; provider charges and latency may apply. This reorders existing hits rather than discovering articles outside the keyword candidate set. Missing compatible vectors or a provider failure retains local search results. Visitor requests do not prepare content or repair stale vectors.
After content edits or changes to the analysis service, model, size, passage settings or enabled sources, check pending preparation and run analysis again for the saved configuration. Stored analysis totals may include earlier configurations.
Integration
Automatic Integration
Lumenare Search automatically replaces:
- Default WordPress search forms
- Gutenberg Search blocks
- Theme search forms
No additional configuration needed!
Shortcode
Add search form anywhere:
[lumenare_search]
Widget
Add "Lumenare Search" widget:
- Go to Appearance → Widgets
- Add "Lumenare Search" widget
- Configure widget settings
- Place in desired widget area
Theme Integration
Use standard WordPress function:
get_search_form();
The plugin automatically enhances this with live search.
Customization
CSS Styling
Target these classes for customization:
.lumenare-search-form– Search form container.lumenare-search-input– Search input field.lumenare-search-dropdown– Dropdown results container.lumenare-result-item– Individual result item.lumenare-highlight– Highlighted search terms
JavaScript Hooks
The 2.2.00 frontend bundle does not emit lumenareSearch:before, lumenareSearch:after or lumenareSearch:select events. For custom behavior, attach your own listeners to the search form and verify them after plugin updates; do not rely on those event names as a public API.
Troubleshooting
Dropdown Not Appearing
Check:
- JavaScript errors in browser console
- jQuery is loaded (required dependency)
- AJAX is working (
admin-ajax.phpaccessible) - Minimum characters setting
- Plugin is enabled
Solutions:
- Fix JavaScript errors
- Ensure jQuery is loaded
- Check AJAX URL
- Adjust minimum characters
- Verify plugin is enabled
Results Not Updating
Check:
- Browser cache
- WordPress cache
- JavaScript errors
- AJAX requests completing
Solutions:
- Clear browser cache
- Clear WordPress cache
- Fix JavaScript errors
- Check network tab for AJAX requests
Styling Issues
Check:
- Theme CSS conflicts
- Z-index issues
- Positioning problems
Solutions:
- Add custom CSS to override theme styles
- Adjust z-index if needed
- Check positioning in browser inspector
Meaning Search Stays Local
Check Search & Answers → Search behavior → Live dropdown for an explicit meaning-search preference and the effective behavior reported for its profile. A global setting inherited by typing does not authorize provider calls.
If the status reports no compatible public passages, save the intended service/model and run Analyze site content in Advanced. Wait for preparation to complete and check pending work for that configuration. A text-generation connector used for answers may have no embedding model for reranking.
If analysis is disabled, follow the message beside the controls: save changes, verify credentials and connector approval, then use Refresh status. A database-read error needs separate investigation; 2.2.00 fixes the mixed-collation comparison that could block older builds. See Troubleshooting.
No Answer Card in the Dropdown
This is expected. The dropdown continues to show results. Enable answer cards separately under Search & Answers → Placement to show a supported excerpt, AI-written summary or no-answer notice on the first full results page. The selected answer style does not turn the dropdown into an answer-writing surface.
Best Practices
Performance
- Keep dropdown results reasonable (5-10)
- Increase live trigger delay on high-traffic sites
- Enable live result images only when thumbnails help users choose quickly
- Use stop words to reduce query complexity
- Monitor query performance
- Implement caching if needed
- Keep typing local unless meaning reranking improves your site's queries enough to justify provider latency and usage
- Check pending analysis after content or model changes rather than relying on stored vector totals
User Experience
- Set appropriate minimum characters (3-4)
- Use featured-image thumbnails when visual context matters
- Enable search term highlighting
- Ensure dropdown is visible and accessible
- Test on mobile devices
Accessibility
- Ensure keyboard navigation works
- Test with screen readers
- Verify focus management
- Check color contrast