Index schema

Understand the schema Algolia uses to index your Shopify data.

Algolia transforms and stores your Shopify data in a format that’s optimal for your search experience.

Products and variants

Algolia stores variants, not products, in an index PREFIX_products. The default prefix is shopify. Variants are stored as independent records.

For products without variants, the product attributes are first transformed into a schema that parallels the structure of the variant’s schema and then uploaded.

The Algolia AI Search & Discovery app only supports indexing products in the “Online Store” sales channel.

Product variant schema

Algolia stores your product variants in the following format:

jsonc

{
  objectID,                       // Variant id (unique identifier also used by Algolia)
  sku,
  barcode,
  id,                             // Product id
  title,                          // Name of the product
  variant_title,                  // Name of the variant
  product_type,
  vendor,
  handle,                         // URL handle for the product
  tags,
  named_tags: {                   // Key value association from tags like this: "key:value" (See Named tags section)
    <TAG_KEY>
  },
  body_html_safe,                 // Description (without HTML, truncated to 2,000 characters)
  compare_at_price,               // Initial price
  price,                          // Actual (optionally discounted) price
  price_ratio,
  price_range,
  variants_count,                 // Total count of the variants of the related product
  variants_min_price,             // Min price of all the variants of the related product (only relevant with distinct enabled)
  variants_max_price,             // Max price of all the variants of the related product (only relevant with distinct enabled)
  variants_compare_at_price_min   // Min compare_at_price of all the variants of the related product
  variants_compare_at_price_max   // Max compare_at_price of all the variants of the related product
  option1,                        // Variant option 1
  option2,                        // Variant option 2
  option3,                        // Variant option 3
  options: {                      // Hash containing the option `names: value`
    <OPTION_NAME>
  },
  option_names,                   // Array containing the option names, in order
  requires_shipping,
  taxable,
  grams,
  weight,                         // Weight with its original unit
  image,
  product_image,
  inventory_quantity,             // Number of available items in the inventory
  position,                       // Position of the variant in the list of products
  template_suffix,
  meta: {                         // Metafields (See Metafields section)
    namespace1: {
      key1,
      key2
    },
  },
  created_at,
  updated_at,
  published_at,
  recently_ordered_count,          // How much have recently been ordered : useful for ranking
  collections: [
    'collection-1-handle',         // Array of product collection handles
    // ...
  ],
  collection_ids: [
    'collection-1-id',             // Array of product collection IDs
    // ...
  ],
  locations_inventory: {
    key1: value1,                  // Object with location_id:quantity - location_id is auto-generated
    key2: value2
  }
}

The image attribute uses the variant’s image if it exists, otherwise it uses the first image of the associated product.

Recently ordered count

The recently ordered count attribute represents all orders since the last full reindex. Initially, it doesn’t exist when the index is first created. It gets added when you sell products. For example, if a customer buys two blue t-shirts in size S, the value of this attribute increases by two.

The recently ordered count attribute is used for the Most Popular sort order.

The recently_ordered_count attribute is reset during a full reindex. Therefore, it’s a count of all orders since the last reindex, not since the beginning. The recently_ordered_count attribute is preserved during real-time indexing.

Searchable attributes for products

By default, these attributes are searchable. They’re ranked in order of relevance—for example, results with a match in the vendor attribute rank before results with a match in the title attribute:

1. unordered(vendor)
2. unordered(title)
3. unordered(variant_title)
4. product_type
5. unordered(tags)
6. unordered(handle)
7. sku
8. barcode
9. unordered(body_html_safe)

Show products instead of variants

To only show the most relevant variant instead of all variants in your search results:

1. Go to the Shopify admin and go to the Algolia AI Search & Discovery app.
2. On the Search options tab, in the Variants section, select Show products instead of variants.

Remove products from search results

Products are indexed if they:

• Have the “Active” status and are published to a sales channel
• Don’t have a publishing date in the future
• Don’t include algolia-ignore in their list of tags
• Don’t include [hidden] in their title

To hide a product from the search results—for example, a product that’s not released yet, you can make one of these changes:

• Add algolia-ignore to the product’s list of tags
• Add [hidden] to its title
• Set the status to “Draft”
• Archive the product

Remove variants from search results

To remove a variant from the search results add [hidden] to the variant’s name. The position of the string [hidden] in the title doesn’t matter.

For example, if you want to hide a variant “blue” from the search results, change its title to blue [hidden] or [hidden] blue.

Collections

If you enabled Collection pages, Algolia stores your collections in a dedicated index PREFIX_collections. The default prefix is shopify. The relationship between products and collections is stored in the product records, in the collection_ids and collections attributes.

The Algolia AI Search & Discovery app only supports indexing collections in the “Online Store” sales channel.

Collections schema

Algolia stores your collections in the following format:

jsonc

{
  objectID,           // Collection id (in Shopify and Algolia)
  title,
  handle,             // URL handle
  body_html,          // Description (with HTML, truncated to 2,000 characters)
  body_html_safe,     // Description (without HTML, truncated to 2,000 characters)
  image,
  products_count,     // Product count in the collection (not updated in real time)
  template_suffix,
  meta: {             // Metafields (See Metafields section)
    namespace1: {
      key1,
      key2
    },
  },
  updated_at
}

Searchable attributes for collections

The following attributes are searchable. They’re ranked in order of relevance—for example, results with a match in the title attribute rank before results with a match in the handle attribute:

1. title
2. handle
3. unordered(body_html_safe)

Remove collections from search results

Collections are indexed if they:

• Are published to a sales channel
• Don’t have a publishing date in the future
• Have at least one product
• Don’t include [hidden] in their title

To remove a collection from the search results, you can:

• Exclude it from the “Online Store” sales channel
• Add [hidden] to the collection title

Hiding a collection from search results won’t hide the products within the collection.

Blogs

If you select Index blog posts (on the Indexing tab of the Algolia AI Search & Discovery app), Algolia stores your Shopify blog posts in a dedicated index PREFIX_articles. The default prefix is shopify.

Blogs schema

Algolia stores your blogs in the following format:

jsonc

{
  _tags: [
    blog_id,
  ],
  author: {
    name,
    objectID,
  },
  blog: {
    commentable,
    created_at,
    feedburner,
    feedburner_location,
    handle,
    objectID,
    title,
    updated_at
  },
  body_html,
  body_html_safe,
  created_at,
  handle,
  image,
  meta,
  named_tags: {},
  objectID,
  published_at,
  tags: [],
  title,
  updated_at,
}

Searchable attributes for blogs

By default, these attributes are searchable. They’re ranked in order of relevance—for example, results with a match in the title attribute rank before results with a match in the handle attribute:

1. title
2. handle
3. tags
4. blog.title
5. blog.handle
6. unordered(author.name)
7. unordered(body_html_safe)

Remove blog posts from search results

To remove a blog post from the search results:

• Add algolia-ignore to its list of tags
• Add [hidden] to its title

Remove entire blogs from search results

Blog posts belong to blogs. To remove all articles from a blog from the search results, add [hidden] to the blog name.

Pages

If you select Index pages (on the Indexing tab of the Algolia AI Search & Discovery app), Algolia stores your Shopify blog posts in a dedicated index PREFIX_pages. The default prefix is shopify.

Pages schema

Algolia stores your pages in the following format:

jsonc

{
  author,
  body_html,
  body_html_safe,
  created_at,
  handle,
  meta,
  objectID,
  published_at,
  shop_id,
  template_suffix,
  title,
  updated_at,
}

Searchable attributes for pages

By default, these attributes are searchable. They’re ranked in order of relevance—for example, results with a match in the author attribute rank before results with a match in the body_html attribute:

1. author
2. body_html
3. title

Remove pages from search results

To remove a page from the search results, add [hidden] to its title.

To hide the [hidden] phrase on your page, replace {{ page.title }} with {{ page.title | replace: "[hidden]", "" }}. For more information, see replace