← Back to Blog

How To Kill AI Slop In Your Marketing Content (With This Anti-Slop Prompt)

AI slop is killing your brand. Learn the 4 categories of AI tells and get a copy-paste anti-slop prompt to make AI content sound like you, not like a robot.

You want AI to work for your marketing. You’ve seen what it can do: draft a blog post in minutes, generate a month of social content in an afternoon. For a $2M-$20M business with a non-existent or small marketing team, that’s genuinely exciting.

Then you read the output. And it sounds like everyone else’s output.

AI slop — low-quality, obviously AI-generated content that damages your brand instead of boosting it — has become the default output of most AI marketing efforts. The tools are fine. The problem is what you’re feeding them.

This article gives you the exact framework we use to catch AI slop before it goes live, plus a copy-paste prompt you can use today.

What AI Slop Actually Looks Like

Most advice about AI content says “it sounds generic.” That’s not helpful. We’ve identified four categories of AI tells, with the specific patterns and words to catch.

1. Structural Tells

AI loves binary contrasts (“It’s not about working harder, it’s about working smarter”). It groups everything in threes. It uses infomercial transitions (“Want to know the secret?”, “The best part?”, “Here’s the thing:”). It hedges constantly with diplomatic warm-ups before every opinion.

Watch for: Binary contrasts (“Not X, but Y”) / Rule of three overuse / Infomercial transitions / Hedging before opinions / Sentences you’d never say out loud

2. Vocabulary Tells

Certain words are AI fingerprints. If your content reads like a LinkedIn post from 2019, AI probably wrote it.

Banned words: game-changer, synergy, utilize (use “use”), facilitate (use “help”), scalable, unlock (as in “unlock growth”), supercharge, leverage (as hype)

Banned patterns: “No X. No Y. Just Z.” / “Same X. Same Y. Same Z.” / “If you’re serious about [goal]…” / “Enter: [thing]” / “Stopped me in my tracks” / “To your success”

3. Formatting Tells

  • Em dashes in every paragraph
  • Decorative arrows (more than one per piece)
  • Emoji where a full stop (aka period) would do
  • Every sentence the same length, a flat metronomic rhythm with no variation
  • Excessive bold text making nothing stand out

4. Voice Tells

This is the one that kills B2B businesses. The content sounds corporate when your brand is casual. It sounds motivational when your brand is practical. It sounds like a press release when it should sound like a founder talking over coffee.

AI doesn’t have a voice problem — it has a “no voice input” problem.

Watch for: Corporate tone when the brand is casual / Third person when first person fits better / Condescending language / No distinctive point of view (could have been written by any company in the industry)

Ann Handley’s Everybody Writes makes the point that voice is a business asset, not a stylistic preference. When AI strips that asset out, every piece of content makes you sound like every other company using the same tool.

The test across all four categories: read it out loud. If it sounds like someone reading from a teleprompter at a conference, it’s slop. If it sounds like someone explaining something to a peer over coffee, it’s probably good.

Why Better Prompts Don’t Fix AI Slop

Here’s where most “humanize AI content” advice goes wrong: they tell you to write better prompts. Add “write in a casual tone.” Say “avoid clichés.” Use “write like a human.”

That helps about 10%.

The root cause of AI slop isn’t bad prompting. It’s missing context. When you ask AI to write a blog post about marketing, it draws on every marketing blog post it’s ever ingested. The output is an average of the internet — which is, by definition, generic.

What’s actually missing:

  • Brand voice documentation — how does your founder actually talk? What words do they use? What words would they never use? What’s the tone — warm and direct, or cool and analytical?
  • Audience specificity — AI writing “for small businesses” produces different (worse) content than AI writing “for $2M-$20M B2B service companies whose founders are stuck as the growth bottleneck”
  • Strategic direction — what’s the angle? What’s the argument? What do you believe that your competitors don’t? AI can’t invent a point of view.
  • Unique stories — what happened in your life that adds color and nuance to your content?
  • Anti-slop rules — specific patterns, words, and structures that are banned for your brand

Without these inputs, you’re asking AI to guess. And AI guesses by averaging. The average is slop.

This is what we call Human Strategy + AI Leverage — the strategic thinking has to come from humans. AI is the production engine, not the strategist. Skip the strategy, and the engine produces noise.

The Anti-Slop Prompt You Can Copy

Here’s a prompt framework you can paste into ChatGPT, Claude, or any AI tool before generating marketing content. Customise the bracketed sections for your business. Edit and improve the prompt as you use it and pick up extra patterns.

CONTEXT:
You are writing for [company name], a [what you do] serving [specific audience].

VOICE RULES:
- Tone: [e.g., "direct, peer-to-peer, like a founder talking to another founder"]
- We lead with aspiration, not pain
- We're confident but not arrogant
- We say what everyone's thinking, even when it's blunt
- Jargon is fine if we explain it — use the precise term, then clarify in plain English

ANTI-SLOP RULES:
- No binary contrasts ("it's not X, it's Y")
- No grouping everything in threes
- No infomercial transitions ("ready to level up?", "the best part?")
- No hedging or diplomatic warm-ups
- Vary sentence length — short sentences for impact, longer ones for nuance
- Use simple verbs: "use" not "utilize", "help" not "facilitate", "build" not "implement"
- Never use: game-changer, synergy, scalable, leverage (as hype), unlock growth, supercharge

BANNED PATTERNS:
- "No X. No Y. Just Z."
- "If you're serious about [goal], let's [CTA]"
- "Here's the thing:"
- "Enter: [thing]"

CONTENT DIRECTION:
Topic: [your topic]
Angle: [what you believe that competitors don't]
Audience pain: [the specific frustration, not a generic one]
Key argument: [the core point of the piece]

This won’t produce perfect content. But it will produce content that’s 80% there instead of 50% there — which means your editing time drops from “rewrite everything” to “polish and sharpen.”

Key Takeaways

  • AI slop isn’t a tool problem — it’s a context problem. Missing brand voice, audience specificity, and strategic direction produce generic output regardless of which AI you use.
  • There are 4 specific categories of AI tells: structural patterns, formatting clichés, vocabulary fingerprints, and voice mismatches. Learn to spot them.
  • Better prompts help about 10%. The real fix is feeding AI the strategic inputs it needs — voice documentation, audience detail, point of view, and explicit anti-slop rules.
  • The anti-slop prompt framework above is a starting point. Customise it for your brand, then refine it as you learn what your specific AI output tends to get wrong.
  • Costly brand damage happens when businesses publish AI-generated content that sounds nothing like them — lost trust compounds faster than saved time.

Is Your Content AI Slop? A Self-Assessment

  1. Could a competitor publish your last 5 blog posts under their brand with zero edits?
  2. Do you have a written document describing how your founder actually talks — specific words, phrases, and patterns?
  3. When you read your AI-generated content out loud, does it sound like you or like a marketing textbook?
  4. Have you defined specific words and patterns that are banned from your content?
  5. Does your AI prompt include your actual audience (with specifics like revenue range and industry) or just “small businesses”?

If you answered “yes” to question 1, or “no” to questions 2-5, your content is probably closer to slop than you’d like. The anti-slop prompt above is a starting point. The deeper fix is building a proper voice documentation system — what we call Brand Voice Capture — so AI has something real to work with.

If you follow the tips and prompt in this article, you need never spend another night at the Slopsville Hotel.

The next evolution is a self-learning flywheel. The cycle works like this: Generate. Edit. Learn.

Every time you edit AI output, you’re spotting patterns the AI gets wrong for your brand. Feed those patterns back into your anti-slop rules and voice documentation. Next generation is better. The one after that is better again. Over weeks, your AI stops guessing and starts sounding like you — because you’ve taught it what “you” actually means. That’s the difference between using AI as a slot machine and using it as a system.