Get found by search engines and AI
Arc CMS writes the technical part for you on every publish: structured data, a sitemap, robots.txt, llms.txt, feeds and a Markdown copy of each page. Your part is three settings screens and the way you write. The full reference is SEO and discoverability.
Fill in who you are
Open Settings, then About. Set Site Name and Production URL, the address people will use, such as
https://www.example.com.Under Identity for search engines and AI assistants, choose the Publisher type, and fill in Logo URL (absolute, at least 112 by 112 pixels), a one or two sentence Description, your Social and profile links (one per line), and the public contact email and phone if you want them shown. These go on every page as schema.org data, and your site can print them too (see Let editors change your home page). Press Save Settings.
Open Settings, then Discoverability. Every AI crawler is allowed by default. Block the Training bots if you do not want your pages used to train models; leave the Search and answer bots allowed, or assistants cannot cite you. Leave Publish /llms.txt and /llms-full.txt and Notify search engines on publish (IndexNow) on.
Press Save and apply to site.
Canonical links and the sitemap are built from the Production URL. Left empty, they point at the Firebase Hosting address, and search engines index that instead of your domain. Set it before launch. Saving it republishes the home page; to write every other page again, use Republish website (Content, Content types) or deploy the website with npm run deploy.
Say what each content type is
Pages publish as an Article by default, which fits most writing. For a type that is something else, edit it under Content types and use the Structured data section: choose Product, Service, Event, Person, Recipe or How-to guide (or Blog post, News article, or Web page for a page that is not an article, as the standard pages use), then map each property to one of your custom fields. Unmapped properties are left out.
What is written for you
| File | When |
|---|---|
/sitemap.xml and one /{type}/feed.xml per content type | Every publish |
/robots.txt, with your crawler choices and the sitemap address | Every publish, and Save and apply to site |
/llms.txt and /llms-full.txt | The same, unless switched off |
A Markdown copy of each item at its URL plus .md | When the item is published |
| Structured data (organization, page type, author, breadcrumbs, FAQ and other blocks, and the rows of an FAQ field) | In every published content page. Static pages (/p/…) carry none: only the head tags you write in them. |
| Your home page's head tags: canonical address, language links, Open Graph and Twitter tags, the site's structured data (see Build your home page) | Every website deploy, Republish website, when content it shows changes, and when you save a setting it uses |
Check them from outside with Google Search Console and Bing Webmaster Tools; the Discoverability screen links to both.
Write each page to be found
In the editor, the SEO tab holds the SEO Title, the Meta Description, the Sources the page draws on, Mark as updated today, and the Canonical URL. Leave the canonical empty: the page then points at its own address on the Production URL. Fill it in only when the original of the piece is on another page, such as a post first published elsewhere. The Checks tab scores the draft from 0 to 100. It is advice and never blocks publishing; 80 or more means the shape is right. It checks:
| Check | Weight |
|---|---|
| Opening paragraph is a direct answer (15 to 70 words) | High |
| At least one heading is a question | High |
| Has an FAQ or Key takeaways block | High |
| Title of 70 characters or fewer | Medium |
| Meta description 50 to 160 characters | Medium |
| Has a list or a table | Medium |
| Links to at least 3 pages on this site | Medium |
| Has an author | Medium |
| Cites at least one source | Medium |
| At least 300 words | Low |
| Has a cover image | Low |
| Has at least 2 tags | Low |
| Published or updated within 12 months | Low |
Each failing check says the one thing to fix. The tab also suggests related pages on your site to link to, and inserts a link in one click.
Habits that get pages quoted
Assistants quote passages, not pages, and Google rewards the same things:
- Answer first. Open the page, and each section, with a complete answer in a few sentences; detail comes after.
- Headings are questions people ask, in their words: "How much does it cost?" rather than "Pricing".
- Facts, not adjectives. Numbers, dates, names and comparisons are quotable; praise is not.
- Give it structure. Lists, tables, and the FAQ, Key takeaways, How-to steps and Definition blocks from the editor's grid icon, which also become structured data.
- Say who and when. Name the author, and press Mark as updated today only after a real revision, never for a typo.
- Cite and link. Add your sources, and link to at least three of your own pages.
Do not stuff keywords, rewrite what is already on the web, or write FAQs nobody asks. What only you can say (your data, your cases, your named opinion) is what gets cited.
Next
If you publish in several languages: Publish in more than one language. Otherwise, go on to Add search.