3 Strategies to Kill AI Slop in Your Email Copy | MarTech

Can you guess? Merriam-Webster’s Word of the Year 2025? Yes, it is sloppy: “low-quality digital content usually produced in large quantities using artificial intelligence.”

If you spend any time on LinkedIn, you’ve probably seen it everywhere. And you’ve definitely seen people complaining about it. But AI doldrums aren’t just a problem on social media. It quietly damages inbox trust, engagement, and conversions.

Jay Schwedelson recently shared data suggesting AI-sounding language has negative consequences email engagement rates. That is the trade-off that many teams are currently making. Use LLMs without strong editing and quality assurance systems and risk damaging engagement, sender reputation and brand trust.

And if this feels like one of those “it won’t happen to us” scenarios, read on. Chances are you will find at least one solution that you can implement immediately.

There are three strategies that can help teams eliminate AI slop without sacrificing speed. None of them are flashy, but they work.

1. Create better message overviews

The first step in improving and humanizing your email campaigns is better command. I mean it.

Strong messaging provides strategy, structure, and a compelling sales or marketing argument for every email you send. They help tailor the offer to the audience, the audience’s level of awareness and the business objectives of the campaign.

A solid assignment supports stronger involvement and better results. It also helps to humanize your content. That matters because one of the biggest reasons people hate AI sloppiness is that it feels thoughtless, irrelevant, and relies on vague, generic claims and promises.

A good briefing protects the intent, relevance and content of every email. If you don’t have an email yet, you can use mine.

2. Better QA systems

Editing is a skill. It’s also a process you can and should train your team in, especially if you use LLMs to help produce email campaigns. A simple editing workflow would look like this:

Edit for strategy

This is where a strong command saves time and prevents adverse consequences. A quick scan of the message letter helps confirm whether the copy meets the core objectives:

• Is the correct viewpoint and angle being used?
• Does this address the right problems, solutions, and outcomes based on the buyer journey or segment funnel stage?
• Are the style and content appropriate for the sender and the audience’s lead or deal status?
• Does the call to action support the intended conversion and align with the intended landing page?

Edit for structure

Then confirm that each email uses a structure that supports the intended objective or target metric.

The core principle is simple. The copy structure should match the result you want to achieve with the email. If the audience segment has a contact or lead status, or is at the top of the funnel, emails tend to perform better with an attention-interest-desire-action structure.

For mid-funnel segments, it is often more effective to start with the problem and use a problem-agitation-solution structure. Ideally, the structure guidance will be built directly into your message outline and copy workflow.

I previously discussed the copy cascade of corners, hooks, and copy structure in a MarTech article on event emails for readers who want to delve deeper.

Edit for line

Finally, it’s time to polish and humanize the copy at the individual line level using editing sweeps or editing passes. I get this data from the excellent work of Chris Silvestri And Joanna Wiebe from Copyhackers (she has done work for Intuit and Shopify, among others). The swipe gestures are:

• Rule of one: Is there one idea per sentence?
• You rule: Does the copy use reader-oriented language?
• Brightness: Is the copy easy to read and understand? (Many marketers will talk about reading levels, but really you also want to aim for market sophistication levels.)
• Voice and tone: Is it consistent? Is it brand appropriate for the audience?
• So what?: Does each claim matter to the readers’ goals, beliefs, and motivations?
• Prove it: Is there sufficient evidence to substantiate claims? Was anything left out?
• Specificity: Are there concrete details instead of vague generalities?
• Increased emotion: Is the emotion of the email appropriate for the stakes involved?
• No risk: Have you reduced friction on CTAs?
• Decision maker: Do you appeal to all four decision styles?

In this final step, check the email copy to ensure it passes the checks described above. Yes, it’s a lot of steps. Yes, it is rigorous. But that’s how great content is created. It also doesn’t take as long as you think. Silvestri has a solid guide to it convert this into a GPT if you are interested.

3. Dedicated person in the know

If you can’t be the person to create instructions and make changes and checks, appoint or train someone on your team to do this and build systems around the work. You can also outsource the job or system construction to an expert.

No matter how you approach it, LLM generated content needs a skilled human to review and examine each output before it goes live. When anyone can generate an email campaign in seconds, LLMs alone are not a differentiator. Ensuring the offer, message, and text are well positioned to perform in the inbox.