Skip to content
Start a scan

AI slop checker

Does your website look AI-generated?

Paste a URL. Honello checks the live page for the nine patterns that make AI-built sites look the same, scores it from 10 to 100, and gives you a prompt to fix every one.

  • Results in about a minute
  • 1 credit ($12 or less)
  • Refunded if a check fails

What is AI slop?

AI slop is the generic, interchangeable look of a website shipped straight from an AI builder or template without real design decisions: the same default font, an unmodified framework color palette, a purple-to-blue hero gradient, frosted-glass cards, emoji used as icons, three identical feature cards, and headlines that could describe any company.

Builders like Lovable, Bolt, v0, Replit, and Cursor make it fast to ship a site, and they tend to ship the same one. Any single default is harmless. Five or six together is what visitors recognize, and a site that looks like a template is harder to trust.

What is the AI Slop Score™?

The AI Slop Score™ is a 10 to 100 rating of how many of those template tells a live web page shows, and how strongly. Higher means more slop. It is deterministic: Honello reads the rendered HTML and CSS, runs nine weighted checks, and the same page always gets the same score.

Each check returns a severity from 0 (clean) to 1 (textbook tell). The score is the weighted average of those severities, mapped onto 10 to 100. Lower is better, which is the opposite of most website scores.

How to read your score

Four bands, straight from the scanner. The verdict on your result uses the same wording.

AI Slop Score™ bands and verdicts
ScoreVerdict
10–34Reads as custom-builtFew or no template defaults. The page looks designed on purpose.
35–59A few tells, mostly customSome defaults left in. Usually two or three quick swaps fix it.
60–79Heavy template tellsVisitors who have seen a few AI-built sites will recognize this one.
80–100Textbook AI slopMost of the nine patterns at full strength. It looks like the builder’s demo.

The nine checks, and what triggers them

Measured on the rendered page, not guessed from a screenshot. Weight is how much each check counts toward the score.

  • Purple/blue gradient2× weight

    Flagged when: Two or more elements with a linear-gradient through purple, violet, indigo, magenta, or fuchsia. One is flagged as borderline.

    Replace it with: A solid brand color, a real photo, or a gradient outside the purple-blue-pink range.

  • Marketing copy clichés2× weight

    Flagged when: Stock phrases in the H1–H3 headings (list below). Three or more is a strong tell; one or two is mild.

    Replace it with: Name a specific outcome, number, or customer instead of a generic verb.

  • Typography1.5× weight

    Flagged when: Headings, body, buttons, or links rendered in one of 11 default AI-builder fonts, or no custom webfont at all.

    Replace it with: A distinctive display face for headings, paired with a plainer font for body text.

  • Color palette1.5× weight

    Flagged when: More than 15% of sampled text and background colors are unmodified Tailwind defaults (indigo, violet, purple, blue, pink at 500/600). Over 5% is partial.

    Replace it with: Two or three brand colors chosen on purpose, starting from the logo.

  • Repeated icon-card grids1.5× weight

    Flagged when: Two or more groups of 3+ identical containers, each holding an icon, a heading, and a paragraph.

    Replace it with: Vary the layout: a list, a table, a screenshot with callouts, or one strong example.

  • Stock imagery1.5× weight

    Flagged when: More than 30% of images served straight from Unsplash, Pexels, Pixabay, unDraw, Storyset, or Freepik. Any at all is mild.

    Replace it with: Real screenshots, your team, your customers, or your product in use.

  • Gradient text and glassmorphism1× weight

    Flagged when: Three or more gradient-clipped headlines or backdrop-blur “glass” panels combined.

    Replace it with: Keep the effect on one focal element at most, or use flat color.

  • Border radius1× weight

    Flagged when: More than half of rounded elements use a 16px+ radius, so every card, button, and image is the same soft pill.

    Replace it with: Small radius on buttons and inputs; large radius only where it means something.

  • Emoji as icons0.75× weight

    Flagged when: Three or more emoji in headings, used as icons or bullet markers (🚀 ✨ ⚡). One or two is mild.

    Replace it with: A consistent icon set, or no icons.

Fonts flagged as AI-builder defaults

Good fonts, all of them. The problem is that every builder reaches for the same short list.

  • Inter
  • Poppins
  • Manrope
  • Plus Jakarta Sans
  • DM Sans
  • Space Grotesk
  • Sora
  • Outfit
  • Urbanist
  • General Sans
  • Satoshi

Phrases flagged in headings

Matched in H1, H2, and H3 text, including variants (“revolutionizing,” “supercharged”).

  • revolutionize
  • elevate your
  • seamless
  • unlock the power
  • supercharge
  • game-changing
  • cutting-edge
  • empower your
  • next-generation
  • effortless
  • in today’s fast-paced world
  • harness the power
  • unleash
  • transform the way
  • one-stop
  • in today’s digital…
  • take your … to the next level

AI didn’t make your site generic. The defaults did, and defaults are the easiest thing to change.

How the AI slop check works

  1. 01

    Paste a live URL

    Any public page from any builder or template. Nothing to install, no code access.

  2. 02

    Honello renders the page

    A real browser loads it at 1440×900, reads the computed fonts, colors, gradients, effects, layout, headings, and image sources, then captures desktop and 390px mobile screenshots.

  3. 03

    Fix, then check again

    Paste the fix prompt into your builder, redeploy, and run the check again. Because the score is deterministic, a lower number means the change worked.

What you get back

  • An AI Slop Score™ from 10 to 100, with a verdict
  • All nine checks marked Strong tell, Mild tell, or Clean, with a note on what was found
  • A specific fix for every tell
  • A design review of the desktop and mobile screenshots, by category, with quick wins
  • A fix prompt to paste into Lovable, Bolt, v0, Replit, Cursor, or Claude Code

What it doesn’t measure

Whether AI built the site
It measures how much the page looks like default AI output. A hand-coded site can score high; an AI-built site with real brand choices can score low.
Whether the text was written by AI
It is not an AI text detector. It flags a fixed list of stock phrases in headings, not authorship.
Whether the page converts
A low slop score doesn’t mean the page sells. Conversion, clarity, mobile UX, and SEO are covered by a full Honello audit.

What it costs

One credit per page check: $12 on its own, less in packs, and about $5 a page on Pro. Credits never expire, and a failed check refunds its credit automatically. Already ran a Honello audit? You can run the check from that report too.

See pricing

AI slop checker FAQs

Straight answers to what people ask before they run it.

What is AI slop on a website?

AI slop is the generic, interchangeable look of a website shipped straight from an AI builder or template without real design decisions: the same default font, an unmodified framework color palette, a purple-to-blue hero gradient, frosted-glass cards, emoji used as icons, three identical feature cards, and headlines that could describe any company.

How do I check if my website looks AI-generated?

Paste the page URL into Honello’s AI slop checker. It loads the live page in a real browser and runs nine checks on the rendered HTML and CSS: typography, color palette, purple/blue gradients, gradient text and glassmorphism, border radius, marketing copy clichés, emoji icons, repeated icon-card grids, and stock imagery. You get a score from 10 to 100 and a fix for each tell in about a minute.

Is a high AI Slop Score™ good or bad?

Bad. The AI Slop Score™ runs the opposite way from most website scores: higher means more template tells. Under 35 reads as custom-built, 35 to 59 has a few tells, 60 to 79 has heavy template tells, and 80 or above is textbook AI slop.

How do I make my Lovable, Bolt, or v0 site look less AI-generated?

Replace the defaults the builder chose for you. Swap the default font (Inter, Poppins, and similar) for a distinctive heading font, replace the indigo and violet utility colors with your brand colors, remove the purple-to-blue hero gradient, rewrite headlines without words like “seamless” or “supercharge,” replace emoji icons and stock images with real product screenshots, and break up grids of identical feature cards. Honello’s checker generates a fix prompt you can paste into the builder to do this.

Is the AI slop checker free?

No. Each check costs one Honello credit, which is $12 on its own and as little as about $5 on the Pro plan. If a check fails, the credit is returned automatically. Honello’s free website scan covers the UX, SEO, and AI discoverability audit, but not the AI Slop Score™.

Does the AI slop checker tell me if AI built my site?

No. It measures how closely the page matches the usual output of AI website builders, not how it was made. A hand-coded page that uses the same defaults will score high, and an AI-built page with deliberate brand choices will score low.

Is this the same as an AI content detector?

No. AI text detectors guess whether prose was written by a language model. The AI Slop Score™ checks visual and structural design defaults, plus a fixed list of stock marketing phrases in headings. It doesn’t estimate authorship.

Does Google penalize websites that look AI-generated?

Google doesn’t penalize a site because AI helped build it. Its guidance focuses on whether content is helpful, original, and made for people. Template sites tend to underperform because they say nothing specific, and visitors trust them less, not because of how they were made.

Will I get the same score if I check the same page twice?

Yes, if the page hasn’t changed. The score comes from fixed rules applied to the rendered page, not from a language model, so it is repeatable and you can use it to confirm a fix worked. The written design review that comes with it is generated separately and doesn’t change the number.

Keep going

Method last updated . Scores are measured at a 1440×900 desktop viewport.

Make it look like you built it on purpose.

See which defaults give your page away, and get the prompt that replaces them.