Is your AEM site readable by LLMs so they can cite the content?

[CedricHuesler]

The AEM product team has been building something new: Adobe LLM Optimizer

In the next few weeks I’m going to share actions for you folks to do, to improve your AEM sites to be more visible in LLMs.

My first ask: Check that your pages main text is in the initial HTML (ok to ignore header, nav, footer). LLMs like ChatGPT and Perplexity don’t run JS when they fetch page during the reasoning phase.

If the content isn’t there, you won’t get cited at the moment they try link back to your site.

🧰 Test your pages with this Chrome/Comet browser extension → https://llmo.page

Questions? - let’s talk here

[Himanshu_Jain]

Hi @CedricHuesler ,

Definitely like to get more insights and make this as a OOTB feature in the product itself which allow content authors to enable pages for LLM optimization.

Thanks

[CedricHuesler]

First version of LLM Optimizer is leveraging the data and content we have of your instance to come up with recommendations - connected to how the brand is currently being cited in the LLMs. We are planning - the same way also Site Optimizer - is bringing in predictive recommendation for pages you are currently working on - that are not live yet. Further we are going to connect LLM Optimizer to Microsoft Copilot - so you can bring the insights into other tools connected with Copilot. More on this in Dec/Jan.. 

[kautuk_sahni]

@CedricHuesler 

Thank you for sharing this, Cedric. Very timely and relevant in the age of LLMs. Curious to learn more:

1. For ensuring pages are visible to LLMs, is the recommendation primarily to have all main content in the initial HTML, or are there best practices for dynamic content as well?

2. Does Adobe LLM Optimizer provide actionable insights on which pages or sections are currently being overlooked by LLMs, and how to prioritize optimizations?

3. Will the tool offer ongoing monitoring or reporting so we can track improvements over time as we apply these changes?

Really looking forward to understanding how this can help improve our content’s discoverability by AI agents.

[CedricHuesler]

Hard and fast rule, content that is relevant needs to be in the initial HTML. LLM Optimizer comes with a capability to edge-side include the dynamic parts on the CDN - for requests from chatbots only. So regular users still have the page working as is - but agents will get a version which includes the dynamic parts included.

Yes, LLM Optimizer can tell you areas of your site that don't get much citation. There can be various reasons, and we are currently supporting some of them as opportunities to change the outcome (as in we detect the situation and allow you to fix it). We will keep adding more opportunities as we discover them.

Yes - the "agents" work behind the scenes continuously - and future version of LLM Optimizer will track mid-term impact of changes made.

[manjunathdj]

@CedricHuesler LLM Optimizer can be used with both AEM Cloud Service and Edge Delivery Service ? 

Yeah looking forward for understanding more insight and usability features.

Thanks,

Manjunath DJ

[CedricHuesler]

Yes, LLM Optimizer is for any CMS - but has built-in support for AEM 6.5 on AMS, AEM as a Cloud Service and Edge Delivery Services. 

[Shashi_Mulugu]

As LLMs become gatekeepers for web content, wouldn’t it be game-changing for brands to communicate their “content credentials” to these systems?
Imagine AEM enabling authors to annotate pages with explicit signals of subject matter authority, recency, regulatory compliance, or audience suitability—helping LLMs reliably choose the “best source,” not just the first one.
This moves us from “can LLMs see us?” to “do LLMs value and cite us for why we matter?” That’s the edge AI-powered discovery needs.

My Two questions for the product roadmap:

1. Authority & Brand Signal Annotation:
   Can LLM Optimizer help organizations surface authority signals (such as expertise claims, certifications, or audience trust scores) within HTML or metadata, and track how these affect LLM citation frequency and visibility over time?

2. Competitive Benchmarking for LLM Citation:
   Could future releases offer dashboards that compare our pages’ LLM discoverability and citation rates against key competitors or industry benchmarks—enabling targeted, data-driven improvements for standing out as the preferred source in agentic search?

[Wilson-Faure]

Would LLM Optimizer work for headless (i.e: using React Framework)? 

[EstebanBustamante]

@CedricHuesl How does the tool handle personalization solutions like Adobe Target, which often use anti-flicker snippets? Is it simply flagging hidden text or content as a penalty?