- SitecoreAI CMS
AI discoverability limitations and troubleshooting
Use this topic to troubleshoot missing signals or answer engine optimization (AEO) opportunities, recommendations that can't be applied, and unexpected AI visibility results in SitecoreAI CMS.
Supported capabilities
AI discoverability features currently support only the web channel and don't include personalization capabilities.
Missing signals or AEO opportunities
Missing widgets or tabs
If the Signals widget and the AEO opportunities tab don't appear, check the connection troubleshooting steps.
If only the AEO opportunities widget is missing from a customized Channels dashboard, add the widget manually.
No signals, opportunities, or recommendations
An empty list doesn't necessarily indicate a connection problem between SitecoreAI and Scrunch. If a list is empty or shows fewer results than you expect, check the following:
- Analysis is pending - Scrunch hasn't analyzed your brand, site, or page yet. Check back after its next run. For a specific page, make sure Scrunch can reach its public URL.
- Scrunch returned no signals or recommendations - there are no signals for the selected period or no recommended updates for the page.
- No contributing pages or opportunities - SitecoreAI shows all signals, but some have no contributing pages. A contributing page doesn't necessarily have AEO opportunities or content recommendations. To identify content improvements across your site, view AEO opportunities in Channels.
- Crawler access is blocked - bot protection or
robots.txtrestrictions can prevent ScrunchBot from reading your site's public pages. See Allow ScrunchBot to access your site.
A hostname mismatch can also prevent opportunities from appearing for a site. Make sure the site's hostname or target hostname matches the domain Scrunch tracks for your brand. To perform the check:
- Find your domain in Scrunch in Brand Settings > Configure > Brand Entities > Primary Brand > Edit brand > Brand Website or Alternative Brand Websites.
- Find your domain in SitecoreAI in Channels > your site's Dashboard > Settings > Site hosts > Edit > Hostname or Target hostname.
- Make sure the two values match. Example domain:
www.my-site.com
Signals need about 30 days of data. If you still see no signals or AEO opportunities after 30 days, contact your Sitecore account executive.
Allow ScrunchBot to access your site
ScrunchBot is the crawler Scrunch uses to read the public pages of sites configured in Scrunch and analyze their content. ScrunchBot follows each site's robots.txt rules, and it doesn't crawl the wider internet.
Bot protection services, such as Cloudflare or Akamai, can block automated visitors and bots, including ScrunchBot. When this happens, Scrunch can't fully read your site, so AI discoverability scores and content recommendations can be incomplete. Scrunch displays a warning when bot protection blocks the crawler.
Allowlisting requires changes by both your web or security team and Scrunch. Coordinate the changes so your site's allowlist and Scrunch's allowlisted crawling are enabled together.
To allow ScrunchBot to access your site:
-
Make sure your site's
robots.txtfile allows ScrunchBot to access the public pages you want Scrunch to analyze. -
Allowlist ScrunchBot.
- Ask your web or security team to allowlist ScrunchBot in your site's bot protection tool. ScrunchBot uses the following User-Agent string:
- Contact Scrunch to arrange allowlisted crawling for your brand.
If your security team requires IP addresses, request the current list from Scrunch. ScrunchBot uses a fixed set of IP addresses when allowlisted crawling is enabled. Don't reuse an IP address list from an old email or document.
After Scrunch can read your site, the bot protection warning in Scrunch disappears. If the warning says that your site is blocking ScrunchBot even though allowlisting is enabled, Scrunch has enabled allowlisted crawling for your brand. In this scenario, ask your web or security team to review the site's allowlist rule.
Pages that can't be updated
SitecoreAI uses a page's public URL to locate the site and page in the current CMS environment. To apply a recommendation automatically, it must also match the suggested change to editable content.
In Signals and Channels, a page can still be listed when SitecoreAI can't locate the corresponding site or page in the CMS. A status message explains why the page can't be updated.
Read the status message and check the following:
- Page not found - make sure the page is managed in the current SitecoreAI environment.
- Site or subdomain not found - make sure the public domain or subdomain maps to a site managed in the current SitecoreAI environment. Pages managed outside SitecoreAI can't be updated through this integration.
- Site and page access - to view a page in SitecoreAI, you need access to the site that matches the URL and view access to the page itself. Ask your administrator to check both.
URL matching might fail in the following cases:
- The public URL of a page doesn't match its route in SitecoreAI CMS, for example, because of custom front-end routing.
- The site uses redirects or URL rewriting.
- Locales use custom domains or subdomains, for example,
example.deorde.example.cominstead ofexample.com/de-de. - The URL of the default locale doesn't include the language, for example,
example.com/aboutinstead ofexample.com/en/about.
Ask your site administrator to check how the public URL maps to the page in SitecoreAI CMS.
If SitecoreAI finds the page but can't match the suggested change to an editable field, see Unmatched recommendations.
Unmatched recommendations
Scrunch analyzes the public, rendered page, and SitecoreAI uses the text to match recommendations to CMS content. If it finds the page but can't match a suggested change to an editable field in the current CMS content, it can't apply the recommendation automatically.
Matching can fail when:
- Rendered text combines several fields - for example, a single block of text contains a title, subtitle, and description. SitecoreAI might find the content item but not the individual field to update.
- The content isn't stored in an editable CMS field - it might be hard-coded in the front-end application, for example, in a component file.
- The content is generated dynamically - for example, personalized content or content created by application logic.
- The content comes from an external system - for example, a commerce or product information system, or through an API.
- Different sites or locales use different CMSs - the content might be managed in another CMS.
- The content has changed since analysis - a recommendation reflects the published content Scrunch analyzed, not necessarily the current CMS content. For example, the page might have been edited after Scrunch analyzed it. Check the Last scan date where available.
Review each recommendation before making a manual edit. If you can locate and edit the relevant content in the Page Builder, make the change manually. If the content comes from another source, update it at that source. Dismiss the recommendation if it's no longer relevant or you don't plan to apply it.
The review page in Signals and Channels can also show related states:
- On this page - needs a manual edit - review the suggestion and edit the content manually.
- No longer on this page - the suggestion refers to content that isn't in the current version of the page, so there's nothing to apply to that content.
In the Page Builder, unmatched recommendations appear with the Not found status.
On the review page, a page is marked Completed after you apply all its recommendations. New recommendations for that page appear only after the next scan.
No immediate change in AI visibility
Applying a recommendation creates a new content version; it doesn't publish the change. First, publish the page and related items to make the updated content publicly available.
Publishing content updates doesn't guarantee an immediate change in citation or mention rates. Projected lift is an estimate, not a guarantee. Publication timing, competitor activity, and changes in how AI tools assess and surface content can affect results.
Content recommendations are a starting point for improving AI visibility, not a comprehensive solution.