34 patterns. Every check, explained.

SlopFix reviews these 34 patterns across five categories. Each check describes the evidence needed and a suggested change. A familiar font, a single gradient, or one common component is not enough to prove a generic design, and no pattern proves AI authorship.

Color

Indigo-to-violet gradient

The same default indigo-to-violet gradient (around #6366f1 to #a855f7) is used as a template accent on prominent buttons, glows, or hero backgrounds. Do not count a gradient with a different, clearly chosen palette.

Try this instead

Pick one accent with a reason behind it and use it flat.

Gradient headline text

Visible heading text itself is filled with a gradient. Do not count gradients elsewhere on the page.

Try this instead

Set headings in one solid color and build hierarchy with size and space.

Glow-lit premium dark mode

A near-black blue page uses decorative spotlights, colored glows, or gradient card surfaces to signal generic premium dark mode. A dark theme by itself is not this tell.

Try this instead

Hold one flat background and separate surfaces with a hairline.

Stock warm palette

The page repeatedly uses an amber, orange, or stone wash on a soft beige base as a generic cozy palette. One restrained warm brand accent is not enough.

Try this instead

Use a neutral base and one restrained warm accent.

Default rainbow of tinted boxes

Several unrelated blue, amber, green, or red tinted boxes appear together like a default semantic palette. Do not count a restrained, coherent brand palette.

Try this instead

Grow any extra colors out of your one palette, or drop them.

Type

One word in serif italic

A sans-serif headline switches one emphasized word to an oversized or display serif italic for stock dramatic contrast. A consistent editorial type system is not this tell.

Try this instead

Keep one voice per headline; emphasise with weight or a line break.

Display serif as the 'premium' look

A software, developer, or SaaS product uses a high-contrast display serif as a generic premium signal in its hero or UI. Do not count a serif that fits a clearly editorial product and voice.

Try this instead

Choose a face you can defend for this product, not the default of premium.

A kicker above every heading

Tiny uppercase, letter-spaced labels repeat above many section headings and mostly restate their heading. One label that adds a category, date, or other useful context is not this tell.

Try this instead

Delete kickers that repeat their heading; let scale introduce sections.

A whole sentence at display size

A long marketing sentence, around ten or more words, is set at display size and spans multiple lines in the hero. A short headline followed by a normal-size subline is not this tell.

Try this instead

Make the big line a few words and put the rest in a normal-size subline.

The five default typefaces

The type system relies on generic, ubiquitous UI or display fonts as an unconsidered default, with no context-specific typographic choice. A popular font alone, or a purposeful combination, is not sufficient evidence.

Try this instead

Try a few faces with your own copy and keep the one you can explain.

The tasteful terminal

A terminal or code window is used as decorative tech styling even though it shows no real product output or useful code. A genuine, relevant code example is not this tell.

Try this instead

Keep monospace for code and use terminal styling only where it serves the product.

Copy

The AI copywriting voice

Copy repeatedly uses stock symmetrical contrasts such as 'not X, but Y', formulaic three-part punchy lists, or excited claims without concrete nouns, numbers, or consequences. One em dash alone is not enough.

Try this instead

Write one specific sentence with a real number, noun or consequence.

Slop vocabulary

Generic hype words such as unlock, elevate, seamless, effortless, supercharge, or powerful recur without saying what the product actually does. One ordinary use of a word is not enough.

Try this instead

Replace each hype verb with what the product actually does.

A headline that fits any product

The main headline is portable marketing language that could advertise many unrelated products and names no audience, concrete job, or specific outcome. A short but product-specific headline is not this tell.

Try this instead

Say who it is for and what changes for them, in their words.

Em dash habit

Em dashes recur as a default rhythm crutch, especially in short marketing copy. Do not count one or two dashes used where they clearly improve a longer sentence.

Try this instead

Use a period or a comma. In short copy, use no dashes.

Emoji on headings and bullets

Decorative emoji repeatedly lead headings, buttons, or bullets in place of a clear visual or verbal choice. A single meaningful status symbol is not this tell.

Try this instead

Cut decorative emoji; keep one only where it carries information.

Faux-insight setups

Copy repeatedly tees up an ordinary claim with phrases like 'what nobody tells you', 'the part everyone misses', or 'here is the uncomfortable truth'. A real sourced finding is not this tell.

Try this instead

Cut the setup and make the claim on its own.

The invented stat row

A prominent row presents contextless or implausible headline metrics as proof without a source, method, time period, or explanation of what was measured. Do not assume a number is invented merely because its source is offscreen.

Try this instead

Show one number you measured and say where it came from.

Stock calls to action

Repeated prominent calls to action use generic labels such as 'Get started', 'Learn more', or 'Try for free' without naming what the visitor gets. One conventional button is not enough.

Try this instead

Name the first thing the visitor gets when they click.

Components

The announcement pill

Decorative status pills such as 'New', 'Beta', or 'Popular' recur without a real, current status they communicate. One factual release or availability badge is not this tell.

Try this instead

Show a badge only for real, current status.

The glowing status dot

A bright dot has a large pale halo or pulse and sits beside generic 'online', 'ready', or 'live' marketing copy without a meaningful changing state. A small flat dot for a real status is not this tell.

Try this instead

A flat dot plus a word, and only when something is actually live.

Rounded-square icon tiles

Repeated features each get a generic line icon inside an identical rounded-square tile, used mainly to fill a grid rather than explain the feature. One meaningful icon is not enough.

Try this instead

An icon must carry meaning or go. A label and one sentence is enough.

Icons in a tint of themselves

Many icons sit inside low-opacity backgrounds tinted to match each icon's own hue, making a grid of soft colored squares. A coherent opaque brand surface is not this tell.

Try this instead

Let the icon inherit the text color, with no container.

Pills and frosted glass

The interface repeatedly uses maximum-radius capsule shapes and/or translucent cards with visible backdrop blur as a decorative style preset. Normal small-radius buttons or cards alone do not count.

Try this instead

Hold one small radius site-wide and use solid surfaces.

The oversized soft shadow

Small cards or icons carry a very wide, diffuse, low-opacity shadow that greatly exceeds their size, often alongside a border or colored glow. A restrained shadow matching a real elevation is not this tell.

Try this instead

Use a tight, colorless shadow, or only a hairline.

Cards with a colored left border

A thick colored stripe on a rounded pale card is repeated across feature lists, rows, or notes as decoration. One real warning or note callout is not this tell.

Try this instead

Let a list be a list; save callouts for one or two real asides.

AI-drawn logo or mascot

The visible mark looks like crude placeholder vector art, such as a rounded square with a face or a mascot made from primitive shapes. Do not infer that a logo was AI-made from style alone.

Try this instead

Get a real mark from a designer, or refine one until it is crisp.

Layout

The template hero

The opening follows a generic landing-page recipe, such as an eyebrow or announcement pill, centered oversized headline, generic subline, paired buttons, and decorative gradient, without product-specific proof or interface. One centered hero element is not enough.

Try this instead

Start with what only your product can show above the fold.

The three-card feature grid

Three equal feature cards use generic icon, heading, and short-description content, giving unrelated features identical emphasis instead of showing the product doing something specific. Any ordinary three-column layout is not enough.

Try this instead

Pick the most important thing and show it fully instead of equal boxes.

01 / 02 / 03 section markers

Several sections are decorated with leading 01, 02, 03 markers even though they do not describe real sequential steps. Actual numbered instructions are not this tell.

Try this instead

Number only what really happens in order.

Cards inside cards

The visual design repeatedly nests rounded, bordered, or filled card surfaces inside other cards so multiple containers compete. A semantic wrapper or one nested example is not enough.

Try this instead

One surface per region; group inside it with spacing.

Bento grid of tinted tiles

A mosaic of unequal feature tiles is washed in separate accent tints and used to tease many features without showing a real workflow or product state. An informative dashboard grid is not this tell.

Try this instead

Show the product doing one real thing instead of a mosaic of teasers.

Builder template left as shipped

The page visibly retains a recognizable site-builder starter structure, placeholder copy, and stock sections with little product-specific editing. Detecting Webflow, Framer, or another builder in source code alone is not evidence.

Try this instead

Replace the template's structure with one decided for your product.

How is the score calculated?

The score combines 60% of a whole-page pattern estimate with 40% of the severity-weighted mean of the eight strongest check estimates. The result is rounded to a number from 0 to 100. Lower means fewer matched patterns. A match estimate is a model judgment rather than a calibrated authorship probability.

The public reference describes the formula and bands, but does not publish each severity coefficient. SlopFix uses its own explicit coefficients while following that structure. It is an independent implementation based on Made with Jev’s public checklist, with checks adapted from kill-ai-slop and no-ai-slop.

What should I do with a match?

Confirm it on the page, decide whether the choice helps the product, and apply the suggestion only when it solves a real problem. The tool captures the first three desktop screenfuls; it does not test complete user journeys, mobile interactions, accessibility conformance, or conversion.

Read how to remove AI slop for an editing workflow, or the UI and UX checklist for a broader review.

Review my website ↗

Copy manually

Your browser blocked clipboard access. Select the text below and copy it.