Search is how most readers navigate documentation. Every Doxbrix site includes fast, full-text search out of the box — no setup, no separate service to configure...
Search is how most readers navigate documentation. Every Doxbrix site includes fast, full-text search out of the box — no setup, no separate service to configure...
Search is how most readers navigate documentation. Every Doxbrix site includes fast, full-text search out of the box — no setup, no separate service to configure. Readers find what they need in keystrokes, and you get insight into what they're looking for.
How readers search
Open search from anywhere on the site:
- Press
⌘ + K(Mac) /Ctrl + K(Windows/Linux), or - Press
/, or - Click the search box in the top bar.
A search dialog opens. As the reader types, results appear instantly — ranked by relevance and grouped by space. Readers navigate results with the arrow keys and open one with Enter.
What gets indexed
Search indexes the full content of every published page:
- Page titles and descriptions.
- Headings (weighted higher).
- Body text, including text inside blocks.
- Code blocks.
Drafts, unpublished pages, and content a reader isn't authorized to see are never returned.
Ranking and relevance
Results are ranked by a combination of signals:
- Title and heading matches rank above body matches.
- Exact phrase matches rank above scattered word matches.
- Popular pages (by analytics) get a relevance boost.
- The reader's current space is favored, with other spaces still searchable.
Matched terms are highlighted in the result snippets so readers can see why a page matched.
Search vs. Ask AI
Doxbrix gives readers two complementary ways to find answers:
| Col 1 | Search | [Ask AI](/reader/site-assistant) |
|---|---|---|
| **Input** | Keywords | A natural-language question |
| **Output** | A ranked list of pages | A direct, cited answer |
| **Best for** | "Take me to the page about X" | "How do I do X?" |
The search dialog offers both: readers can pick a result or hand their query to the site assistant for a synthesized answer.
Keyboard-first
Search is built for power users:
| Key | Action |
|---|---|
| `⌘K` / `/` | Open search |
| `↑` `↓` | Move between results |
| `Enter` | Open the selected result |
| `⌘ + Enter` | Ask AI with the current query |
| `Esc` | Close search |
Filters and scoping
Readers can narrow a search:
- By space — limit results to Documentation, API Reference, etc.
- By group — within a space, focus on a section.
Multi-language search
On multi-language sites, search respects the reader's active locale — it searches and returns results in their language, falling back to the default locale where a translation doesn't exist.
Search analytics
Every search feeds analytics, giving you a window into reader intent:
- Top queries — what readers search for most.
- Zero-result queries — searches that returned nothing, signaling missing content or vocabulary mismatches.
- Click-through — which results readers actually open.
Performance
Search runs against a pre-built index served from the CDN, so results are effectively instant and work even on large sites. There's nothing for you to provision or scale.