Internet Is Fun 
Open ten sites for ten unrelated products and you will see roughly the same page. A large centred heading, a short sentence under it, a rounded button, three columns of icons, some testimonials, a footer with four link columns.
This is not laziness, or not only laziness. Several forces pushed in the same direction at once.
Design systems made one layout very cheap
Component libraries mean a competent team can assemble a professional-looking page in a day. The components are good, well-tested and accessible. They are also the same components everyone else is using, so the output converges.
The alternative β designing something specific β costs more and can easily come out worse. Most teams make the rational choice.
Mobile removed most of the options
A layout has to work at 375 pixels wide. At that width almost every arrangement collapses into a single column of stacked blocks. Once the mobile version is one column, the desktop version tends to become the same thing with more padding.
Multi-column, dense, print-influenced layouts are much harder to make responsive, so they mostly went away.
A/B testing optimises towards the familiar
Testing rewards whatever performs slightly better this quarter, and the familiar reliably performs well β people know where to look. Run enough tests and you converge on the layout everyone already recognises.
It is a local maximum. Testing will never find a genuinely different design, because a genuinely different design tests badly at first.
What it costs
Convergence is not automatically bad. Familiar layouts are easy to use, and a great deal of βcreativeβ web design was actively hostile.
The cost is that a site stops telling you anything about who made it. When everything is the same template, the only difference is the copy β and a lot of that is converging too.
Where it has not happened
Personal sites, obviously, which is most of why people have started making them again. Also anything built by one person for no money, which has no A/B test to satisfy and no brand guidelines to follow.
That is the actual appeal of the weird web: not that the sites are better, but that they are clearly the work of somebody specific.
