Search Tuning — Per-Index Relevancy Controls
Search Relevancy
Search Tuning — Fine-Tune How Your Search Ranks Results
Every index is different. A news site needs freshness. A product catalog needs exact matches. A knowledge base needs semantic understanding. Search Tuning gives you visual controls to shape relevancy per index — no code, no config files, instant effect.
DEFAULT SETTINGS
wireless headphones review
Sony WH-1000XM5 Review#1
Posted yesterday — comprehensive review...
Best Wireless Headphones 2024#2
Old roundup from 2 years ago...
Headphone Buying Guide#3
→
AFTER TUNING
wireless headphones review
Sony WH-1000XM5 ReviewFRESH
Posted yesterday — boosted by freshness...
Headphone Buying Guide#2
Semantically relevant — matched by meaning...
Best Wireless Headphones 2024#3
Older content ranked lower...
Freshness boost + semantic balance pushed the new review up and demoted stale content.
Where to Find It
Search Tuning lives inside Index Settings in your Opensolr dashboard. Open your index, click the gear icon or Index Settings, and expand the Search Tuning section. Every change saves automatically — move a slider, and your very next search uses the new settings.
The Six Controls
Field Weights
Control how much each field contributes to relevancy ranking. The four searchable fields are Title, Description, URI, and Text (full body). Use the master slider to quickly shift between title-focused ranking (great for navigational queries) and text-focused ranking (great for deep content search).
5.0
Title
4.0
Description
0.5
URI
0.01
Text
Default: Title 5.0, Description 4.0, URI 0.5, Text 0.01 — title-heavy. Drag the master slider right to give body text more influence, or type exact values into each field.
Freshness Boost
How much newer content is preferred over older content. Higher values push recently published or updated pages toward the top. Uses the document's
creation_date field with a time-decay curve — recent documents get the biggest boost, which fades over days and weeks.Range: 10 (barely noticeable) to 1000 (aggressively fresh). Default: 100. Only applies when search mode is set to "Fresh" — standard search ignores this setting.
Minimum Match
How many of the user's search words must appear in a document for it to be considered a match. Three presets:
Flexible
Show more results
Some words can be missing
Some words can be missing
Balanced
Most words must match
Good middle ground
Good middle ground
Strict
All words must match
Fewest but most precise
Fewest but most precise
Default: System-managed (adapts automatically for vector indexes). Choose a preset to override.
Semantic vs Keyword Balance
Controls how much weight goes to semantic (vector) understanding versus exact keyword matching. Only available on vector-enabled indexes (those with
embeddings in the schema). Move left for keyword-heavy results, right for semantic-heavy.
More Keyword
More Semantic
Range: 0.0 (pure keyword) to 3.0 (heavily semantic). Default: 1.5 — balanced. The system also adapts dynamically based on query length (longer queries get more semantic weight), but your override takes priority.
Result Quality Threshold
The minimum relevance score a document must reach to appear in results. Raise it to filter out weak matches and show only highly relevant results. Lower it to be more inclusive and show everything that has some match.
Range: 0.0 (show everything) to 1.0 (only near-perfect matches). Default: 0.60 — filters out low-relevance noise while keeping useful results.
Results Per Page
How many search results are returned in each page. Applies to both the Opensolr Search UI and API responses. Higher values show more results but increase response size.
Range: 10 to 200. Default: 50. Adjust based on your UI layout — grid layouts work well with 20-30, list layouts with 50+.
How It Works — Under the Hood
1
You move a slider
Change any control in the Search Tuning panel. The value saves automatically after a 400ms debounce — no Save button needed.
2
Stored per index
Your custom value is saved to your index configuration. A
NULL value means "use system defaults" — so resetting a control removes the override entirely.3
Applied on next search
When a search request comes in, the engine loads your custom values and applies them as overrides on top of the system defaults. No reindexing, no restart. The very next query uses your tuning.
Reset Behavior
Every control has its own Reset button that restores it to the system default. There's also a Reset All to Defaults button at the bottom of the panel that clears all customizations at once.
Reset Individual Control
Click the Reset button next to any control. The value goes back to system default and the override is removed from your index. System defaults include adaptive behavior — for example, vector indexes automatically adjust semantic weight based on query length.
Reset All to Defaults
Clears every custom value at once. Your index goes back to behaving exactly like it did before you opened Search Tuning. All adaptive behaviors are restored.
Quick Recipes
News Site
Prioritize fresh articles
Set Freshness Boost to 500-800. Set Minimum Match to Flexible. Leave field weights at defaults — titles already have the highest weight, and news articles have strong titles.
Knowledge Base
Semantic understanding first
Set Semantic vs Keyword to 2.0-2.5 (more semantic). Set Minimum Match to Flexible. Set Field Weights — increase Text weight to 1.0+ so body content has more influence. Freshness doesn't matter for evergreen docs, keep it low (10-30).
E-Commerce
Exact product matches
Set Minimum Match to Strict — users searching for "blue wireless headphones" should see results with all three words. Keep Semantic at 1.0-1.5 so typos still work. Set Result Quality Threshold to 0.70+ to cut weak matches. Results Per Page at 20-30 for grid layouts.
Blog / Content Site
Deep content discovery
Increase Text field weight to 0.5-1.0 (use the master slider toward "Text-focused"). Set Freshness at 100-200 for moderate recency bias. Minimum Match on Balanced. Semantic at 2.0 for natural-language queries that blog readers tend to use.
Defaults at a Glance
| Control | Default Value | Range |
|---|---|---|
| Title Weight | 5.0 | 0 – 20 |
| Description Weight | 4.0 | 0 – 20 |
| URI Weight | 0.5 | 0 – 20 |
| Text Weight | 0.01 | 0 – 20 |
| Freshness Boost | 100 | 10 – 1,000 |
| Minimum Match | System-managed | Flexible / Balanced / Strict |
| Semantic vs Keyword | 1.5 | 0.0 – 3.0 |
| Result Quality Threshold | 0.60 | 0.0 – 1.0 |
| Results Per Page | 50 | 10 – 200 |
FAQ
Do I need to reindex after changing tuning settings?
No. Search Tuning controls are applied at query time, not index time. Your changes take effect on the very next search request.
What happens if I don't customize anything?
Everything stays at system defaults. The search engine uses battle-tested defaults that work well for most use cases, including adaptive behavior for vector indexes that adjusts parameters based on query length.
Does Semantic vs Keyword show up for all indexes?
No. It only appears on vector-enabled indexes — those using the
embeddings field for semantic search. Non-vector indexes use pure keyword search, so the control isn't shown.Does Freshness Boost always apply?
Only when the user searches with Fresh mode enabled (the "Fresh" toggle on the search UI, or
fresh=yes in the API). Standard search does not apply freshness boosting regardless of this setting.Can I set different tuning for different indexes?
Yes — that's the whole point. Every index has its own Search Tuning settings. A news index can have high freshness and flexible matching, while a product index on the same account has strict matching and low freshness. Each index is tuned independently.
Ready to Tune Your Search?
Open Index Settings in your dashboard and expand Search Tuning. Changes take effect on the very next search.