How Semrush fits into AEM’s ecosystem
August 2026
With Adobe acquiring Semrush, I wanted to get up to speed on Brand Visibility, the LLMO capability it now powers. I know the AEM Sites and Edge Delivery side well. What I didn't have was a picture of how Semrush, Brand Visibility, Sites Optimizer, and Brand Concierge connect to the architecture I already work with, or where each one gets its data. When I asked Adobe's AI assistant, the answers kept pointing at analytics tools, which didn't help me place anything. Diagrams are how I learn, so I mapped the whole flow myself.
How I think about it
Coming at this as an AEM architect, the easiest way in was to put the site at the center. AEM Sites and Edge Delivery is the part I already know. It produces the content, the pages, and the delivery traffic. Everything new hangs off of it as a layer:
- Brand Visibility is the discovery layer. Semrush provides the AI and search intelligence (prompts, mentions, citations, competitors) and CDN log forwarding shows the AI agents and crawlers hitting the site.
- Sites Optimizer is the site-quality layer, running audits and Operational Telemetry against the pages themselves.
- Brand Concierge is the conversational layer, with its own widget and its own events.
- An analytics tool is the business layer, and it's the only one of the four that needs traditional web analytics. It enters the picture when you want to tie the rest to leads and revenue, and not before.
Four layers, four different data sources. Once I saw it that way, the rest of the products stopped feeling like a bundle of overlapping SEO tools and started looking like an architecture.
The architecture
Here's how I'd draw the whole flow. Content and delivery sit at the top, the measurement products hang off of it, and everything rolls up into one scorecard that feeds decisions back into the site.
flowchart TD
A["Customer goals\nRevenue • qualified leads • self-service • loyalty"]
B["AEM Sites + Edge Delivery Services\nOwned content • pages • structured data • delivery"]
P["Semrush AI/search intelligence\nPrompts • mentions • citations • competitors"]
L["CDN log forwarding\nAI agents • crawlers • referral traffic"]
C1["AEM Operational Telemetry\nSampled page-performance signals"]
C2["Web instrumentation\nPlatform Web SDK • Tags • GA4"]
C3["Brand Concierge\nWidget or SDK • conversation events"]
D["Adobe Brand Visibility\nAI visibility • prompt monitoring • referral traffic"]
E["AEM Sites Optimizer\nAudits • opportunities • recommendations • fixes"]
F["Adobe Analytics / GA4 / CJA\nWeb engagement • conversion • commerce"]
G["Brand Concierge analytics / CJA\nConversation • feedback • intent • handoff"]
H["Discovery outcomes\nBrand Visibility • Share of Voice • mentions • citations"]
I["Experience outcomes\nCore Web Vitals • accessibility • SEO • site health"]
J["Digital outcomes\nVisits • clicks • carts • orders • revenue"]
K["Conversation outcomes\nEngagement • completion • satisfaction • conversion"]
M["Customer Outcomes scorecard"]
N["Prioritize content, technical fixes, and concierge tuning"]
A --> B
P --> D
L --> D
B --> C1
B --> C2
B --> C3
B --> G
C1 --> E
C2 --> F
C3 --> G
D --> H
E --> I
F --> J
G --> K
H --> M
I --> M
J --> M
K --> M
M --> N
N --> B
N --> P
What data each product consumes
Each product pulls from a different source, and only one of them is about human behavior. The table below breaks down what goes in, what comes out, and whether Adobe Analytics is required.
Sources: Adobe Brand Visibility Overview, Sites Optimizer Basics, Brand Concierge Documentation, Brand Concierge FAQ
How data gets into each tool
Semrush into Brand Visibility
My first instinct was that you'd export data from Semrush and load it into AEM somewhere. You don't, and you shouldn't. Brand Visibility is the customer-facing Adobe layer. You define a prompt strategy and monitored prompts, and Semrush provides the AI and search intelligence underneath the Brand Presence experience. The citations, mentions, competitors, and opportunities show up in Brand Visibility without any manual data movement.
flowchart TD
A["Prompt strategy\nMonitored prompts you define"] --> B["Adobe Brand Visibility"]
B --> C["Semrush AI visibility and\nresponse intelligence"]
C --> D["Brand Presence\nCitations • mentions • competitors • opportunities"]
Source: Brand Visibility
CDN logs into Brand Visibility
This is the part I find most interesting. When an AI agent requests one of your pages, it usually doesn't execute your JavaScript, so your browser analytics never sees it. CDN log forwarding is how Brand Visibility observes that invisible traffic: AI agents, crawlers, and the referral clicks that follow. The Referral Traffic dashboard is populated entirely from CDN logs. Its Business Impact view is the one place in this section that needs a connected analytics provider.
So the split is simple: CDN logs catch the agents, and your analytics tool catches the humans who click through afterward.
Sources: Referral Traffic, LLM Optimizer Integration
AEM Sites and EDS into Sites Optimizer
Sites Optimizer works off the site itself: rendered pages, audits, and Operational Telemetry. On AEM as a Cloud Service, Operational Telemetry is automated and privacy-oriented, but it's sampled. It's built for spotting performance regressions (LCP, INP, CLS, performance by device type), not for replacing web analytics. I'd treat it as a smoke detector, not a dashboard.
Sources: Operational Telemetry for AEM as a Cloud Service, Core Web Vitals Opportunity Documentation, Edge Delivery Services Overview
AEM Sites and EDS into Adobe Analytics or CJA
This one is a dependency, not a requirement. It's only in the picture if you want page, CTA, funnel, or conversion reporting. The instrumentation is the standard Adobe pattern I already knew: AEM Client Data Layer, into Tags or the Platform Web SDK, through Platform Edge, and out to Analytics or CJA. Adobe has a tutorial that walks through collecting page-view and CTA-click data with the Web SDK, which is a good starting point.
flowchart TD
A["AEM Client Data Layer"] --> B["Adobe Tags or Platform Web SDK"]
B --> C["Adobe Experience Platform Edge"]
C --> D["Adobe Analytics and/or CJA"]
D --> E["Analysis Workspace and reporting"]
Source: Integrate AEM Sites and Adobe Analytics with Platform Web SDK
Brand Concierge into its analytics layer
Brand Concierge draws on approved content, product catalogs, commerce data, knowledge bases, and Experience Platform profiles. It ships with its own conversational analytics (conversation starts, duration, sentiment, feedback, handoffs, resolution), and CJA can power the deeper journey reporting when you want to see whether conversations actually influenced revenue.
Sources: Brand Concierge Documentation, FAQ, Conversational Experience Blueprint
Measuring the actual outcome
Once the layers made sense, the scorecard view fell out of it. Each row is a business question, the metrics that answer it, and which layer those metrics come from.
The trap to avoid: visibility metrics alone don't prove revenue impact. If you want to make a business-outcome claim, you need a baseline period, a clearly identified change (an optimization, a Concierge launch), instrumentation for the events you care about, a measurement period after the change, and some attribution method. Before-and-after reporting is a start, but an A/B test or control group is much stronger evidence.
Brand Visibility has an Impact Measurement Engine for supported opportunities that follows the same pattern: capture a baseline, deploy, measure over an evaluation period, compare.
Source: Impact Measurement Engine
Three ways to roll it out
Start minimal. AEM Sites or EDS, Brand Visibility with monitored prompts, CDN log forwarding, Sites Optimizer, and Brand Concierge's native analytics. You get AI visibility, agent traffic, site health, and conversation feedback with nothing extra to buy. What you don't get is a revenue or customer-journey view, and that's a fine place to start.
Add Adobe Analytics. If the customer already owns it, add the Web SDK or Tags instrumentation and the commerce or lead events. Now there's a real path from AI visibility to referral to engagement to revenue. One thing that surprised me: Brand Visibility connects to Adobe Analytics with the customer's existing access and report suite. The current docs say no new Developer Console project, technical account, or client secret required. (Adobe Analytics Integration)
Go cross-channel with CJA. When the journey spans web, mobile, offline, CRM, Commerce, and the contact center, CJA plus Experience Platform is the option that can join AI referral, site behavior, Concierge conversations, and offline outcomes into one view. (Customer Journey Analytics Integration)
Where I landed
The mental model I ended up with: the AEM site is the subject. Semrush is the intelligence feed watching how AI experiences talk about the brand, CDN logs are the observation point for traffic my browser JavaScript never sees, Sites Optimizer is the quality feedback loop on the pages, and Brand Concierge is a new front door. Every one of those layers works on its own. The analytics tools the AI assistant kept pointing me at are just the last mile, for when someone wants to tie it all to revenue.
For ramping up on Brand Visibility with LLMO and Semrush, drawing the flow was what made it click. If diagrams help you the way they help me, start with the big one at the top and the rest of this reads itself.