We all know someone who's disappointed with AI.

"It just gives generic advice."

"It sounds like every other marketing article."

"It doesn't really understand our business."

And they could totally be right, but not because the model's bad. Without context it doesn't know your campaigns, your segments, or what's actually happening inside your email program.

This morning, we're breaking down the missing link between your AI and better email marketing.

Also inside:
→ Hey [First Name] was never personalization.
→ What you recommend after checkout matters just as much as what got added to the cart.
→ You and me both know why you haven't switched yet...
→ Convenience is making or breaking conversions, legacy brands are chasing the women's market, and NRF wants Congress to tackle swipe fees.

Hey [First Name] was never personalization.

For twenty years we've called it that. A first name in the subject line. A "we picked these for you" block that's the same six products for everyone. 

But none of that actually changed the email.

Meanwhile Meta ads have been matching a different creative to every person who scrolls past, quietly learning who responds to what. 

That's where Allan comes in. 

Upload your headlines, creative, offers, and copy. Allan assembles the combination most likely to convert for each subscriber. 

Real personalization was never about knowing their name. It was about knowing what makes them buy.

Start a 30-day free trial, creative included → getallan.com

*Sponsored

The quick take: What you recommend after checkout matters just as much as what got added to the cart.

The missing link between your AI and better email marketing

Everyone's using AI for email now. It spins up subject line variations, digs through competitor research, and hands you the skeleton of a welcome series so you're not staring at a blank page. That part's solved.

But ask your AI "what should I do with my email program this week" and you get the same top-10 list everyone else gets. The one that fits a $2M skincare brand and a $50M supplement brand exactly the same, because it fits nobody.

That's not AI being bad at its job. It's AI with no context. It doesn't know your campaigns, your segments, or your revenue, so it gives you advice built for a business that doesn't exist.

There's a fix for that context gap, and it comes down to one connection most marketers haven't set up yet. Here's how it works.

The real split: toy vs tool

Used one way, AI is a toy. You type a prompt, get a generic block of copy, and paste it in. That works for warming up a blank page, but it's useless when you need to decide what to actually do next.

Used the other way, it's a tool. Same AI, completely different job. Instead of "write me a welcome email," you're asking "look at my last seven days and tell me what to fix." The first is guessing. The second is reading what your data is telling you.

The difference isn't the model, it's whether the model can see your data.

Where AI actually saves you time right now

Reporting, by a mile. Pulling numbers, comparing weeks, summarizing what happened. That used to be a two-hour Monday ritual. With the right setup it's two minutes, and your team stops dreading the spreadsheet grind and gets to be creative again.

The bigger unlock is the follow-up question. Pulling the report is easy. Knowing what to ask next is the hard part, and this is where a connected AI earns its spot. "Open rate dropped 4% last week. Want me to check which segment drove that?" Now you've got a junior analyst that never gets tired.

A few other places it's legit today:

• Briefs. Voice-note what you want, get back a structured brief for your designer in 60 seconds.

• Audits. It won't do the audit for you, but it'll clunk through screenshots and flows and surface patterns you'd have to sit and stare at.

• Subject lines. Write three you like, then ask for 15 more in a few different voices. Most will be trash, but one or two will beat what you started with.

The ceiling every AI conversation hits

Reporting, briefs, audits, subject lines: every one of those wins runs into the same wall. The AI doesn't know your numbers.

It doesn't know what your campaigns did last week, which segments are converting, or what your customers have bought before. So you copy, paste, screenshot, and describe what you're looking at, hoping for something useful, and what you get back is usually generic.

That's the whole problem, and it's exactly what MCP solves.

What MCP actually is

Skip the acronym. MCP is a secure bridge between your AI tool and your email platform. That's it.

Without it, you're the bridge. You're exporting reports, filtering by date, downloading a table, pasting it into ChatGPT, going back for more. Each trip is three or five minutes. It adds up until analyzing your own data becomes the thing you keep putting off.

With it, your AI goes straight to the source. "Pull my last 90 days of campaign performance and tell me my best and worst sends" runs in under a minute against real, live data, with no screenshots and no hallucinated numbers.

Setup is not the technical nightmare the name suggests. In ChatGPT, find Omnisend in the app directory and click connect. In Claude, add a custom connector with one URL. No API keys, no config.

One real tip: use a thinking or reasoning model, not the free instant one. The fast models will connect to your data and then skim right past half of it, then hand you a confident wrong answer. The reasoning models actually read what comes back.

Three prompts to start with

Once your data's connected, start here.

The weekly snapshot. "Show me the last seven days: revenue, orders, email and SMS performance, top campaigns. Then give me the five things I need to know." This is also your trust test. Check the numbers against your dashboard so you know the connection is solid.

The upsell finder. "Analyze my recent buyers and engagement data, then recommend three upsell segments I should create. Tell me why each one, and give me the best campaign angle for it." This is the one that finds revenue you already earned and forgot about.

The repeat-winner play. "Look at my top campaigns from the last 90 days. What subject lines, offers, and segments should I run again next month?" Stop guessing what worked from memory. 

Let it find the pattern.

For any of these, ask for five when you only need two. Then you pick. That's the whole game.

Where humans stay in the loop

Here's the simple rule: if it's reversible, let AI take a swing, and if it's irreversible, humans only.

Drafting a campaign is reversible, and so is pulling a report. Sending to your whole list is not, and neither is suppressing 20% of your file. Once those go out, getting them back is hard or impossible.

Add one layer to that: high-stakes reversible is still a human call. A flagship launch, anything touching pricing, anything legal or compliance. Technically you can undo it. The cost of getting it wrong is too high to hand off.

And brand voice stays yours, always. AI has a default register that's slightly cheerful, pretty safe, and forgettable, and your customers can smell it from a mile away. If you're running 5 or 50 brands, it doesn't natively know that brand A and brand B sound nothing alike. You have to teach it, and you still review everything it writes, because it drifts.

If you're handing every send to an autonomous agent with no human at the gate, you're playing with fire. One bad send can torch your deliverability and your revenue in an afternoon.

The missing link between AI and better email marketing

The point was never to have AI write your emails. The real value is that it helps you think faster about what's happening inside your program, and those are two very different jobs.

In this setup you're the one in charge, not the AI. It preps the work, finds the patterns, and hands you the options, and you're still the one who approves, manages, and decides. Give it your context and it makes a good marketer faster and a great one dangerous.

Three quick wins to steal:

• Connect your AI to your real data before asking it anything strategic

• Run the weekly snapshot prompt every Monday instead of building the report by hand

• Keep a human on every send, suppression, and launch, no exceptions

You and me both know why you haven't switched yet...

There's a reason you're still logging into a platform you've complained about for a year, and it has nothing to do with price or features. It's the years of flows, segments, and forms you'd have to rebuild from scratch to leave, so you stay, quietly resentful, clicking renew.

But Omnisend just took that off your plate.

Their team rebuilds all of it for you, moving your flows, segments, and forms across so switching stops being a project and becomes a handoff. And right now, making the move comes with 35% off.

👉 Make your move

*Sponsored

🍦 DTC Scoop:

Shoppers are punishing brands that get convenience wrong

Two new surveys put numbers on something DTC brands already feel: 70% of online shoppers have abandoned a cart over shipping costs, and 67% say return fees make them less likely to buy at all. 56% say retailers aren't upfront enough about fulfillment. The kicker: marketplaces like Amazon and Walmart are now the top discovery channel for new brands at 44%. Convenience isn't a perk anymore and the brands that are inconsistent or unclear about it are losing.

Levi's, The North Face, and Columbia are all betting big on women

Legacy apparel brands that have historically skewed male are pouring resources into women's. The US women's apparel market is roughly $159 billion vs. $89 billion for men's. VF Corp's CEO thinks The North Face alone could double from $4B to $8B by expanding with women, and Columbia says its womenswear is growing faster than men's. Interesting signal for DTC brands too: if you're skewing heavily toward one gender, you might be leaving real money on the table.

NRF laid out retail's 6 biggest policy priorities for 2026

The one that matters most for DTC brands: credit card swipe fees hit a record $198.25 billion last year, up 80% since the pandemic. NRF is pushing Congress to create more competition among payment networks to bring those costs down, and 75% of voters are behind it. The other big one is data privacy. Right now every state has its own rules, and NRF wants a single federal standard so brands aren't juggling a dozen different compliance frameworks.

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!