Search with Algolia
How CMS Orbit keeps published pages in Algolia: the index names, the shape of each record, when records change, and how to query them from your website.
Reference3 min read
When the organisation has connected Algolia in Orbit under Connections (Growth or higher), CMS Orbit writes every published page of every space to your Algolia app. Your website searches Algolia directly with its own search-only key; CMS Orbit only writes.
Index names
One index for each space, environment and language:
<prefix>_<space>_<environment>_<language>prefixis the index prefix set in Orbit, such asnorthwilds.spaceis the space's slug,environmentthe environment's key andlanguagethe locale code, such asen-gbornb.
Everything is in lower case, and anything other than letters, digits and hyphens becomes a hyphen. The Northwilds website's English pages in production are in northwilds_website_production_en-gb, and the same pages on staging in northwilds_website_staging_en-gb. Regions that share a language share its index; filter on region to tell them apart. Settings in the space lists the indices.
Records
Each published page is one record, with the page's id as objectID:
{ "objectID": "6a1f0c2b9d4e8f7a6b5c4d3e", "title": "Lofoten ridge", "name": "Lofoten ridge", "description": "Five days along the ridge above the fjord, sleeping in mountain huts.", "path": "/trips/lofoten-ridge", "url": "https://northwilds.com/trips/lofoten-ridge", "locale": "en-GB", "region": "europe", "contentType": "trip", "publishedAt": "2026-11-16T09:00:00.000Z", "publishedAtTimestamp": 1794819600, "text": "Walk the Lofoten ridge & sleep in huts.\nAbove the Arctic Circle"}titleis the page's SEO title, or its name when it has none.descriptionis the SEO description.urlis the full address on the domain for the page's region and language, or the path when the space has no domain or preview address.textholds the words of the page's Text, Long text and Rich text fields, including those in components, with the HTML removed. Other field types are left out.- Records stay under 10 KB, Algolia's limit on its smallest plans; long text is shortened.
Only published pages are written: drafts, global entries, entries without an address and pages marked Hide from search engines are not.
When records change
CMS Orbit updates the index from the same events as webhooks: cms.page.published, cms.page.rolledBack, cms.page.promoted, cms.page.unpublished, cms.page.trashed, cms.page.restored, cms.page.purged, cms.release.published and cms.folder.moved. A page that is no longer published is deleted from the index. If Algolia doesn't answer, the change is tried again later, like any other event.
Rebuild index in the space's Settings writes every index again. It fills a temporary index, <index>_rebuild, with your index's settings, synonyms and rules, and moves it into place when it is ready, so search keeps working throughout.
Configure the index
CMS Orbit never changes an index's settings, so set them in Algolia once:
- Searchable attributes:
title,description,text. - Attributes for faceting:
contentType,region,locale. - Custom ranking:
desc(publishedAtTimestamp)to prefer newer pages.
Query from your website
Use a search-only API key, never the admin key you gave Orbit:
import { liteClient } from 'algoliasearch/lite'const search = liteClient('NW4LGOL1A', '<search-only key>')const { results } = await search.search({ requests: [{ indexName: 'northwilds_website_production_en-gb', query: 'lofoten', hitsPerPage: 10, attributesToSnippet: ['text:30'] }],})Each hit has the page's title, url and a snippet of its text, enough for a results list. Fetch the full page from the delivery API by its path when you need more.
For search by meaning across languages without a third party, see Search by meaning.
Was this page helpful?