Faceted Search & Aggregations

Also Known As

TL;DR

Explanation

Faceted search computes the count of matching documents per category value as part of the search query, using the same filters applied to the main results. Elasticsearch aggregations, Meilisearch facets, and Typesense facets handle this natively. The SQL equivalent (multiple GROUP BY queries per facet dimension) is too slow. Facets enable exploratory search — users narrow results by clicking filter values whose counts show relevance.

Diagram

flowchart LR
    RESULTS[Search results] --> FACETS[Facets - filter counts]
    subgraph Facet Types
        TERMS[Term facets<br/>category: PHP 42<br/>category: Security 38]
        RANGE[Range facets<br/>difficulty: beginner 150<br/>difficulty: advanced 80]
        DATE[Date histogram<br/>updated: this month 45]
    end
    USER[User selects PHP] --> FILTER[Filter: category=php]
    FILTER --> REFINED[Refined results + updated counts]
    INFO[Counts update dynamically<br/>based on current selection<br/>Elasticsearch aggregations]
style FACETS fill:#1f6feb,color:#fff
style REFINED fill:#238636,color:#fff
style INFO fill:#d29922,color:#fff

Common Misconception

Why It Matters

Common Mistakes

• Computing facet counts with separate SQL GROUP BY queries — O(n) queries for n facets; use a search engine.
• Not filtering facet counts by other selected filters — the facets should reflect the current filtered result set.
• Facet values not sorted by count — users expect most common values first.
• No 'show more' for facets with many values — showing all 500 brand values overwhelms users.

Code Examples

// Multiple SQL GROUP BY — O(n) queries, slow:
$results = $db->query('SELECT * FROM products WHERE category = ?', [$category]);
// Separate query per facet dimension:
$brandCounts = $db->query('SELECT brand, COUNT(*) FROM products WHERE category = ? GROUP BY brand', [$category]);
$priceCounts = $db->query('SELECT price_range, COUNT(*) FROM products ...');
// 5 facets = 6 queries — gets slower as facets are added

// Elasticsearch — one query returns results + all facet counts:
$query = [
    'query' => ['term' => ['category' => $category]],
    'aggs' => [
        'brands'       => ['terms' => ['field' => 'brand.keyword', 'size' => 10]],
        'price_ranges' => ['range' => ['field' => 'price', 'ranges' => [
            ['to' => 50], ['from' => 50, 'to' => 100], ['from' => 100]
        ]]],
    ]
];
// Single query: 1000 results + brand counts + price range counts

References

• ↗ https://www.elastic.co/guide/en/elasticsearch/reference/current/search-aggregations.html

Tags

Related Terms