Query Filters Search Settings

Navigate to Groundworx → Query Filters in the WordPress admin. This is where you configure how the search engine behind Global Search and autocomplete works.

You don’t need to touch these settings to use filters. The Filter Terms and Filter Order By blocks work purely through WordPress’s built-in query system. These settings control the search side of the plugin — Global Search, Suggested Searches, Suggested Links, and Filter Text’s autocomplete.

The Search Index

At the heart of the search features is a custom search index. When you activate the plugin and enable search, it scans all your posts and builds an index — a database table of individual words and phrases, each tagged with where they came from (title, content, excerpt, taxonomy) and how important they are.

When someone searches for “chocolate cake,” the plugin doesn’t run a slow SQL LIKE query against your entire posts table. It looks up “chocolate” and “cake” in the index, finds every post that contains those words, scores each post based on where the words appear and how close together they are, and returns the results ranked by that score.

This is what makes the search fast and the results relevant.

Relevance Mode

The Relevance Mode toggle controls whether this index is used:

• Off — The plugin doesn’t touch search behavior. WordPress’s default search runs as normal (which is a basic LIKE query with no ranking).
• Full — All search goes through the Groundworx index. Results are ranked by relevance score.

Leave this on “Full” unless you have a specific reason to use WordPress’s default search (like a conflict with another search plugin).

Content Weights

Not all content is equally important for search. A match in a post’s title is almost always more relevant than a match buried in paragraph 47 of the content. The weight settings control this.

Each content source has a weight — a number that controls how much a match in that source contributes to the post’s relevance score:

• Title (default: 5) — The most important. If someone searches “chocolate cake” and a post is literally titled “Chocolate Cake,” that should be the top result.
• Excerpt (default: 4) — The hand-written summary. Usually a strong signal of what the post is about.
• Taxonomy (default: 3) — Category, tag, and custom taxonomy term names. If a post is tagged “chocolate,” that’s a clear topical match.
• Content (default: 1) — The full body text. Matches here are common but less focused — the word might appear once in a 2,000-word post.

These defaults work well for most sites. You’d adjust them if your content has unusual characteristics. For example:

• A recipe site where titles are highly descriptive → keep title weight high.
• A documentation site where the body content is the main substance and titles are generic (“Getting Started,” “Configuration”) → lower title weight, raise content weight.
• A site that relies heavily on taxonomy organization → raise taxonomy weight.

Compound Term Boost (default: 3) gives extra weight to multi-word phrases you’ve defined in the Dictionary (covered below). If “New York” is a compound term, the combined token “new york” gets this boost when matched, ranking posts about New York above posts that just happen to contain the words “new” and “york” separately.

Per-Post Type Overrides

You can set different weights for different post types. Click “Add Override” and select a post type to give it its own weight configuration.

This is useful when one post type has very different content characteristics than another. Products might need high title weight because product names are specific. Blog posts might benefit from more excerpt weight because they have carefully written summaries. Pages might need lower content weight because they’re often short and every word is equally important.

Proximity Settings

When someone searches for multiple words — “chocolate cake recipe” — how close those words appear to each other in a post matters. A post where “chocolate,” “cake,” and “recipe” appear in the same sentence is a better match than one where they appear in three different paragraphs.

Proximity settings control this scoring:

Exact Phrase Boost (default: enabled, boost 8) — If the exact phrase “chocolate cake recipe” appears in a post (all three words, in that order, adjacent), the post gets a big relevance bonus. This is the strongest signal that the post is what the visitor is looking for.

Nearby Words Boost (default: enabled, boost 4, distance 5) — If the search words appear within 5 words of each other but not as an exact phrase (“chocolate cake” in one sentence and “recipe” three words later), the post gets a moderate bonus. Close proximity usually means the words are related in context.

Distant Word Filter (default: disabled, distance 50) — When enabled, this penalizes or excludes posts where the search words appear far apart — more than 50 words between them. On a long blog post, “chocolate” in paragraph 1 and “cake” in paragraph 12 probably aren’t related. Enabling this tightens results but might exclude some legitimate matches on content-heavy sites.

For most sites, the defaults work well. You might tighten them (lower distances, higher boosts) on sites with short, focused content like products or directory listings. You might loosen them (higher distances, lower boosts, disable the distant filter) on sites with long-form content like articles or documentation.

Like content weights, proximity settings support per-post type overrides.

Excluding Post Types and Taxonomies

The transfer list on the settings page lets you exclude entire post types or taxonomies from the search index. Excluded content won’t appear in Global Search, autocomplete suggestions, or any search-powered feature.

Common exclusions:

• Utility post types (navigation menus, reusable blocks, form submissions)
• Private taxonomies (internal categorization that shouldn’t be public)
• Post types handled by another search system (WooCommerce products if you’re using their built-in search)

---

Dictionary

The Dictionary lives under the search settings and manages three kinds of entries that shape how content is indexed and searched. Think of it as teaching the search engine about your specific content’s language.

Compound Terms

By default, the search engine treats every word independently. Search for “ice cream” and it finds every post containing “ice” and every post containing “cream” — including posts about “ice skating” and “cream cheese” that have nothing to do with ice cream.

Compound terms fix this. When you add “ice cream” as a compound term, the indexer treats it as a single token. Now a search for “ice cream” specifically matches posts about ice cream.

Add compound terms for: – Multi-word names: “New York,” “San Francisco,” “Apple TV” – Domain-specific phrases: “machine learning,” “real estate,” “ice cream” – Brand names and product names relevant to your content

Synonyms

Synonyms map related terms so that searching for one also finds the other.

Bidirectional (↔) — “couch ↔ sofa” means searching “couch” also finds posts containing “sofa,” and searching “sofa” also finds posts containing “couch.” Use this when the terms are true equivalents.

Unidirectional (→) — “car → automobile” means searching “car” also finds “automobile,” but searching “automobile” does not find “car.” Use this when one term is a broader version of another, or when you want to expand a common search term without the reverse.

Practical uses: – Regional variations: “apartment ↔ flat” – Abbreviations: “NYC → New York City” – Common misspellings or alternate names for things your visitors search for

Stopwords

Common words that add no meaning to a search — “the,” “and,” “is,” “at,” “which.” The plugin ships with a standard English stopword list. These words are stripped from the index so they don’t bloat it or dilute relevance scores.

You can add domain-specific stopwords if your content has common words that aren’t useful for search. You can also reset to the defaults if you’ve made changes you want to undo.

A note on index rebuilds: Changes to compound terms and stopwords change how content is tokenized, which means the index needs to be rebuilt. The rebuild runs automatically in the background after you save changes. On large sites (thousands of posts), this can take a few minutes. Synonym changes don’t require a rebuild — they’re applied at search time.

---

Per-Post Search Settings

Sometimes you need to fine-tune search at the individual post level. Every post, page, and custom post type entry has two search-specific settings.

Excluding a Post from Search

Check Exclude from Global Search and the post is removed from the search index entirely. It won’t appear in Global Search results, autocomplete suggestions, or any search-powered feature.

Use this for: – Utility pages that visitors shouldn’t find via search (privacy policy, terms of service, thank-you pages) – Duplicate or placeholder content – Draft-quality pages that are published but not ready to be promoted

Boosting (or Burying) a Post

The Search Boost multiplier adjusts how a post ranks in search results. The default is 1 (normal ranking). Set it higher to promote a post, lower to demote it.

• Boost of 2 — The post’s relevance score is doubled. It shows up higher in results than it normally would.
• Boost of 0.5 — The score is halved. The post still appears in results but ranks lower.
• Boost of 5 or higher — The post dominates results for any matching query. Use sparingly.

Real-world uses: – Boost your product pages or key landing pages so they outrank blog posts that mention the same keywords. – Demote old or less important content that’s still relevant enough to keep published. – Temporarily boost seasonal content (holiday promotions, event pages) and reset it afterward.

Both settings are accessible from: – The Global Search column in the post list table (a quick visual indicator) – Quick Edit on any post (fast individual changes) – Bulk Edit (update many posts at once)

---

Accessibility

The plugin is built for keyboard and screen reader users:

• Every filter works with keyboard only. Dropdowns respond to Arrow, Enter, Escape, and Tab keys.
• Custom dropdowns use proper ARIA attributes — role="listbox", aria-expanded, aria-selected, aria-activedescendant — so screen readers announce the state and options correctly.
• Filter labels are always exposed to assistive technology, even when visually hidden with the “Hidden” label position.
• Focus is managed carefully: it’s contained within open dropdowns and returned to the trigger when the dropdown closes.
• Autocomplete inputs use the ARIA combobox pattern with live region announcements so screen readers report how many suggestions are available.

You don’t need to configure anything for this — it’s built into every block.

---

URL Parameters

Filters work through URL parameters. When a visitor selects a filter, the URL updates — ?category_name=news or ?s=chocolate. This means filtered states are bookmarkable and shareable. A visitor can copy the URL and send it to someone, and they’ll see the same filtered results.

Parameter Reference

Multiple selections are comma-separated: ?category_name=news,sports,tech

Multiple Query Loops on One Page

When a Query Loop does not inherit from the template, its filter parameters are namespaced with the query block’s ID: ?query-1-category_name=news&query-2-tag=featured. This prevents conflicts when two or more filtered Query Loops appear on the same page.

---

Tips

Limit search columns if you don’t need full-content search. Title-only search is significantly faster and usually more relevant.

Use contextual terms selectively. It runs an extra query per filter to figure out which terms still have results. On pages with many taxonomy filters, this adds up. Enable it where it genuinely helps the user and skip it where the taxonomy is small enough that dead-end selections aren’t a problem.

Place Filter Active at the top of your filter layout so visitors immediately see what’s active. It’s easy to lose track when there are multiple filters, especially on mobile.

Use the Global Search Dropdown in headers and the expanded variant on search pages, can be very useful in a mega menu panel. Quick Links make the dropdown useful even before someone starts typing — it becomes a navigation shortcut, not just a search box.

Don’t over-tune relevance settings. The defaults are sensible for most sites. Adjust weights if you notice specific problems — like blog posts outranking product pages for product-name searches — rather than preemptively tweaking numbers.