What a browser actually receives when you publish a page

Before anything is drawn on screen, a browser is handed a description and asked to make sense of it.

It helps to picture the sequence. You publish a page, somebody asks for it, and what travels across the network is not an image of the page. It is a description of the content and its structure, in order, from top to bottom. The browser reads that description and builds an internal model of the document: a tree of parts, each one labelled with what it is and each one sitting inside another part. Only after that model exists does anything visual happen. Appearance is applied to the model, and behaviour is attached to the model. If a part of the model is wrong or missing, every later stage inherits the mistake.

This matters because the model, not the picture, is what other software uses. When a page is read aloud, the software walks the model. When a reading mode extracts an article, it inspects the model for the pieces that look like an article. When a search engine builds an understanding of your page, it works from the model. When somebody presses a key to move to the next section, the software is asking the model where the sections are. None of those readers can see the picture, and several of them would not care about it even if they could.

The practical consequence for a beginner is calming rather than daunting. It means your first job is not visual at all. Your first job is to write down what the content is, in the order a person would sensibly read it, using the plainest possible description for each part. If you do that and nothing else, you already have a page that works: readable, navigable, searchable and usable by somebody who cannot see it. Everything you add afterwards is refinement on top of something sound. Everything you add on top of a bad description is decoration on a broken foundation.