Wrapping arbitrary HTML in an <a href=""> — a pattern often called a block link — is technically legal, and it is a common way to turn a whole card of content into one large click target. That legality doesn't make it a good default, for reasons that show up both in assistive technology and in ordinary pointer use.

Adrian Roselli has argued that wrapping everything in the anchor is close to the worst option available. The problem is the accessible name: it is built from every piece of content inside the anchor, so a screen reader user tabbing through controls hears the heading, the image (undeclared as an image) and the body text read out in full before learning that the thing is a link at all.

Perhaps the worst thing you can do for a block link is to wrap everything in the <a href>. […] for a screen reader user the entire string is read when tabbing through controls. In the following example, the first link contains the heading, image (without declaring it as an image), and block of text, taking about 25 seconds to read before announcing it as a link. When tabbing, you do not always know the control type until the accessible name is complete.

That figure comes from a fairly ordinary card: a header, an image and a paragraph. Twenty-five seconds is the practical consequence of leaving the control's identity until the end of its name.

<a href="https://css-tricks.com/article/"> <!-- display: block; -->
  <div class="card">
    <h2>Card</h2>
    <img src="..." alt="...">
    <p>Content</p>
  </div>
</a>

The narrower anchor, with an expanded hit area

Leaving the anchor where it belongs — on the heading alone — avoids the reading penalty.

<div class="card">
  <h2><a href="https://css-tricks.com/article/">Article</a></h2>
  <img src="..." alt="...">
  <p>Content</p>
</div>

The click target can then be recovered by stretching the link's clickable area across the full card rather than moving the markup:

.card {
  position: relative;
}
.card h2 a::after {
  content: "";
  position: absolute;
  top: 0;
  left: 0;
  width: 100%;
  height: 100%;
}

This pairing handles both concerns: the pointer gets the big target, and screen reader users get a short, predictable link name. It's the minimum any card-as-link implementation should reach.

What neither approach fixes: text selection

Both strategies share a defect that has nothing to do with assistive tech. With a link covering the card, you can't place the cursor inside the text and select it the way you'd expect. On mouseup the click fires the link, provided the pointer is still over it. Text isn't strictly unselectable, but selecting inner content becomes awkward in a way users notice.

Fixing it isn't straightforward. An exotic JavaScript approach could watch for the start of a text selection and suppress the click in that case while still following the link on a plain click, but the author finds something about that arrangement suspicious. Heydon Pickering's "The redundant click event" section of his Cards article is offered as further reading on the subject.

The verdict, then, is that block links are a bad idea, with an open invitation for anyone to demonstrate an implementation that resolves the selection problem alongside the accessibility ones.