What AI slop actually looks like

Generated sites converge. Not because generation is bad, but because every generator reaches for the same defaults, and the defaults are recognisable once you know them.

The tells are specific, not a vibe

Slopmeter scores how closely a site matches the defaults generated sites ship with. It reads the page rather than judging it, so every signal is something you can go and look at.

• A hero line straight from the template, and copy written in clipped slogans.
• The vocabulary chatbots reach for, and the it-is-not-just-X-it-is-Y construction.
• Stiff connectors: moreover, furthermore. Filler phrases that say nothing.
• A bento grid, cards inside cards, and a trusted-by logo row with nothing behind it.
• Coloured glows instead of shadows, text filled with a gradient, graph-paper grids.
• Rounded icon tiles stacked above headings, and tiny all-caps labels above them.
• The default icon set untouched, stock component-library theme values, one of the handful of fonts everything uses.
• Body text set too tight or too small to read comfortably.

Why it is worth knowing

A templated page is not ugly. It is unspecific. Every choice it did not make is a thing it does not say about you, and a visitor who has seen forty pages like it has no reason to remember the forty-first.

The same convergence has a second cost. When a page is assembled from the same defaults as everything else, there is less on it that an answer engine could lift and attribute to you rather than to the category.

The score describes the output, never the author

Band names here are deliberate. They describe what came out, not who made it or how hard they worked, and a low score is not an accusation about the person who shipped it. Plenty of deliberate sites lean on a default or two that the crowd also uses, and the ladder has a band that says exactly that.