From CQ 2.0 to EDS: 25 Years on the Adobe Stack

by Josh Oransky

I started shipping CQ (Communique, what AEM was called before Adobe bought Day Software) in 2000. Twenty-five years of the same content stack, watched from inside a series of customer engagements. Several things have changed wildly. A small number of things have stayed exactly the same. Both lists matter.

What changed wildly

The size of "good performance." In 2003, a fast page was under five seconds. In 2013, it was under two. In 2026, the floor is "felt instant": Lighthouse 100 under field conditions on a mid-tier Android. The platform that wins is the one that hits that floor without the developer writing performance tricks.

The deployment model. CQ 2.0 ran on Sling on Tomcat on a JVM you tuned by hand. AEM Cloud Service abstracted the infrastructure. Edge Delivery removed most of the infrastructure from the equation entirely. The content is a Git repo. The platform is a CDN. You stopped capacity-planning twenty years of compute.

Who edits. CQ-era authoring required an admin user with browser plugins. AEM Sites required a trained user on a desktop. EDS via DA or Google Drive requires anyone who can write a document. Authoring expanded from a fifty-person team to a five-hundred-person team in the same organisation. The platform stopped being a bottleneck for content velocity.

What stayed the same

Three things have not changed in twenty-five years, and probably never will.

The author experience is the product. CMS platforms that prioritised author experience over architectural elegance won every engagement I worked on. The ones that didn't, including a couple of very expensive headless rebuilds, lost the author team and then the executive sponsor.

The shape of the content predicts the shape of the project. If you cannot model your content in a way the author understands, your authors will not maintain the site. They will get marketing to ship one-off PDFs and bypass the CMS. Every twenty-year-old CMS install I've seen has a graveyard of pages no one is willing to touch.

The org chart eats the architecture. Conway's law applies. If the marketing team and the engineering team don't talk, no CMS choice will fix that. If they do, even the worst CMS choice will work. The Adobe stack didn't change this. The technology never does.

What I'm watching now

The interesting front edge of the stack is the intersection of EDS and applied AI: not chatbots bolted on the side, but agentic authoring loops that turn structured content into useful publishing automation. The platform finally has the affordances. The next twenty-five years are about figuring out what to do with them.

I have a couple of opinions on that. Maybe I'll write them down.