AI gets blamed for a lot of bad marketing copy.

But if we're being honest, the problem usually starts before the AI ever generates a word.

We give AI almost no context, then expect it to read our minds. And when it doesn't, it's easy to assume the problem is the tool.

Because when you look at the prompts most people are using, it's usually something like: "Write me an email for our sale." Then they're surprised when the output sounds generic.

The marketers getting real value from AI aren't using secret tools or magic prompts. They're giving the model enough context to do its job.

Today we're breaking down the prompt framework that's helping email and SMS marketers get better outputs, stronger ideas, and a lot more leverage from the tools they're already paying for.

Also inside:

→ ChatGPT just got hired as your Omnisend analyst.
→ The quick take: Your list tells you it's tired before inbox providers do
→ Somewhere right now, a shopper is getting four versions of "did you forget something?"
→ How brands are capitalizing on the World Cup, Amazon expands their trucking network, + rethinking loyalty in the world of AI

ChatGPT just got hired as your Omnisend analyst.

Omnisend's new MCP connection lets ChatGPT (and soon Claude) talk directly to your account, so the "how did Tuesday's campaign actually do" rabbit hole turns into one sentence in the chat you already have open. Performance, deliverability, segments, automation revenue, all answered conversationally.

• 30-second account snapshot, no dashboard required

• Top campaigns ranked by revenue in a single prompt

• Bounce and complaint flags before they tank your sender rep

• Subject line patterns pulled from your own historical sends

• Monthly exec summaries with strategy notes baked in

You'll actually want to check your numbers now. Cause let's face it, the reporting tab was never the fun part.

→ See it in action

*Sponsored

👌 The quick take: Your list tells you it's tired before inbox providers do

The AI prompt framework that actually improves email & SMS performance

The marketers who aren't getting much value from AI are usually running into the same problem.

The issue isn't ChatGPT. It isn't Claude. And it isn't that AI "can't write good copy."

Most of the time, the problem is the prompt.

We've seen marketers spend hours tweaking outputs when the real issue started before they ever hit enter. The brands getting the most value from AI aren't relying on secret tools or hacks. They're simply setting the AI up for success with better inputs and clearer direction.

The reality is that AI isn't magically generating winning campaigns from thin air. It's responding to the information you provide. If your prompt lacks context, direction, or constraints, the output will too.

That's why the marketers seeing the biggest gains from AI aren't treating it like a shortcut. They're treating it like another part of their workflow.

Here's how to do the same.

1. Treat the prompt like a campaign brief

A lot of marketers approach AI like a vending machine. They put in a request, get an answer back, and then decide whether they like it.

That's not how the best teams are using it.

They're treating prompts the same way they'd treat a campaign brief. The more context, direction, and structure they provide upfront, the better the output becomes. In many cases, the quality of the prompt matters more than the model you're using.

The strongest prompts typically include four components:

• Context

• Task

• Constraints

• Output format

Miss one of those pieces and you'll usually end up with copy that's generic, off-brand, or difficult to use.

2. Build every prompt using four layers

Before you hit enter, run through a simple checklist.

Once you start using all four layers consistently, you'll notice a dramatic difference in output quality.

3. Generate subject lines that teach you something

One mistake we see often is marketers asking AI for a single subject line and then evaluating whether it's good or bad.

That's backwards.

The value isn't in getting the perfect answer on the first try. The value is generating enough quality options that you can start spotting patterns.

A strong subject line prompt should include:

• Brand context

• Audience segment

• Campaign objective

• Character limits

• Psychological triggers

• Past winners for reference

For example:

The labels are important.

Once you have ten subject lines categorized by trigger, you start learning which angles consistently show up in your winners. Over time, that becomes far more valuable than any individual subject line.

AI becomes your research assistant, not just a copy generator.

4. Use AI to break through campaign planning bottlenecks

When marketers hit a campaign planning wall, the problem usually isn't a lack of ideas.

It's that they're too close to the brand. After months of building campaigns around the same products, audiences, and offers, it's easy to fall back on familiar angles without realizing it.

When you're deep inside a category every day, it's easy to default to the same angles, offers, and campaign structures over and over again.

This is where AI can be incredibly useful.

Instead of asking for generic campaign ideas, feed it real inputs:

• Products

• Customer segments

• Seasonal moments

• Recent campaigns

• Current business goals

A useful prompt might look like:

The output isn't your final calendar; t's a starting point that helps you explore angles you may not have considered on your own.

And sometimes that's all you need to get moving.

5. Let AI improve your copy instead of writing it

This is probably the highest-leverage use case for most marketers.

The strongest email copy usually comes from someone who understands the audience deeply. AI can help improve that copy, but it's rarely the best source of the first draft.

Instead of asking AI to write the email, try asking it to improve the email.

Some useful editing prompts include:

• Strengthen the opening hook

• Remove unnecessary filler

• Improve flow between sections

• Tighten the CTA

• Match the tone of these previous emails

• Identify areas that feel repetitive

We've found this approach produces far better results than starting from scratch.

The highest-leverage use case for AI isn't replacing the writer. It's helping the writer move faster.

If maintaining brand voice is important, compile a collection of your strongest-performing emails and ask AI to identify patterns across them:

• Sentence structure

• Opening styles

• CTA language

• Common transitions

• Tone characteristics

Once you have that voice profile, you can reference it in future prompts and dramatically improve consistency.

6. Prompt for SMS differently than email

One reason AI-generated SMS often falls flat is because marketers prompt for it the same way they prompt for email.

Email can afford a slower build. SMS can't.

The channel demands clarity almost immediately, which means the prompt needs to account for that.

A strong SMS prompt should include:

• Character limits

• Audience segment

• Campaign objective

• Desired angle

• CTA structure

For example:

We also recommend asking the AI to label each variation.

That makes testing significantly easier and helps you understand which messaging angles resonate most with your audience over time.

7. Build a prompt library and iterate relentlessly

The marketers seeing the biggest gains from AI tend to have clear processes for how they use it.

They've built systems that make AI more useful, repeatable, and easier to scale across their workflow.

A few habits worth adopting:

Stay in the same thread

If the output misses the mark, explain what's wrong and continue refining.

Most of the improvement happens in the second and third round, not the first.

Save winning prompts

When a prompt consistently produces useful outputs, save it.

Build a library organized by:

• Subject lines

• Campaign ideation

• SMS

• Product launches

• Lifecycle emails

Over time, this becomes one of the most valuable assets on your team.

Never ship the first draft

AI-generated content should be a starting point.

The strongest marketers still add:

• Brand nuance

• Customer insight

• Product context

• Strategic judgment

That's the part AI can't replace.

Test against human-written copy

Don't assume AI is automatically better.

Run tests.

Compare AI-generated subject lines against your own. Compare SMS variations against existing controls.

You'll quickly discover where AI excels and where human judgment still has the edge.

Better prompts matter more than better AI tools

The teams seeing the biggest gains are using AI to generate ideas faster, explore more angles, and spend less time on repetitive production work.

The prompt is what determines whether that happens.

The better your instructions, the better your output. And like most things in retention marketing, the advantage rarely comes from the tool itself. It comes from how thoughtfully you use it.

That's why the marketers getting the most out of AI aren't chasing the latest model release. They're getting better at giving the model direction.

Somewhere right now, a shopper is getting four versions of "did you forget something?"

We're all familiar with the usual.

Email hits you with "forgetting something?" SMS sends a "your cart misses you." Push sends a reminder. Maybe WhatsApp hits them one last time.

Four channels. Four templates. None of them know what the others said.

It's the same feeling you get when a cold emailer hits you with "Hey, [First Name]." Technically personalized. Actually clueless.

Outcraft runs all four channels from the same brain.

→ If the voicemail went unanswered, the SMS references it
→ If the shopper said "I'll think about it" on the call, the follow-up email knows
→ If they asked about shipping on SMS, the next touchpoint doesn't ask again

One conversation across every channel, not four separate sequences pretending the others don't exist.

Your retention stack has the channels. It's just missing the memory.

See it run → outcraft.ai

*Sponsored

🍦 DTC Scoop:

The World Cup is retail's Super Bowl moment this summer

FIFA 2026 is turning malls into fan zones, and brands are scrambling for a piece. Goodwill just opened a 70K-square-foot activation space, and retail media networks are repositioning their ad inventory around the tournament. If your brand touches sports, outdoor, or travel, now's the time to build a World Cup flow.

Consumers are telling AI to filter your brand out

A new study says 56% of consumers are comfortable letting AI decide which brand messages make it through, and a third have already told their AI to prioritize certain brands over others. Meanwhile, 61% of 25-34 year olds ditched a brand for a better loyalty program last year. The takeaway: first-party data and community aren't nice-to-haves anymore. They're the only way to survive the AI filter.

Amazon just opened its trucking network to everyone

Amazon launched less-than-truckload shipping for any business in the U.S. through its Supply Chain Services program, and freight stocks immediately cratered. FedEx Freight dropped 7%, Old Dominion fell 5%. For DTC brands, this could mean cheaper, faster logistics if you're willing to ride Amazon's rails. Worth watching how pricing shakes out before locking in Q3 shipping contracts.

Annnnd that’s a wrap for this edition! 

Thanks for hanging with Chase and me. Always a pleasure to have you here.

If you found this newsletter helpful (or even just a little fun), don’t keep it to yourself! Share ecomemailmarketer.com with your favorite DTC marketer. Let’s get them on board so they don’t miss next week’s drops.

Remember: Do shit you love.

🤘 Jimmy Kim & Chase Dimond

PS - Your next best customer might be reading this right now. Want in? Email Jimmy to sponsor this newsletter and more.

Love this newsletter but want to receive it less frequently? Let us know by clicking here!