What Should an Ecommerce Brand Look for in an ESP?

Most ESP comparisons still get settled on the feature list. Can it build campaigns, automate flows, segment customers, and produce a report someone will read?

Every serious platform now does all four. Which means the feature list has stopped separating anything, and the questions that do separate platforms sit underneath it: how much of your store data arrives, how fast it arrives, and what is allowed to act on it once it is there.

We have covered the evaluation process itself in how to choose an ecommerce ESP, and what agent-operability actually requires in what kind of email stack to choose in 2027. This is the scorecard version: the areas worth grading a platform on, and what good looks like in each.

What to look for in an ecommerce ESP

Start with strong ecommerce data

For an ecommerce brand, the first thing to check is not what the platform can do. It is what the platform knows, and when it finds out.

Customer profiles and store events should sync into the ESP in real time:

• customers
• orders and purchases
• products and categories
• inventory
• product views
• cart activity
• checkout activity
• purchase history

Every integration will claim all eight. The difference shows up in the lag.

Say a customer opens the same product page three times on Tuesday, adds it to the cart on Wednesday, and leaves without buying. If those events reach the ESP in minutes, Wednesday evening's browse abandonment email is still about something the customer is actively thinking about. If they arrive in an overnight batch, the brand finds out on Thursday, by which point the customer has bought it elsewhere or moved on. Identical integration on the feature page, different outcome.

So do not stop at asking whether the platform integrates with Shopify. Ask which customer profiles and store events are synced, how quickly they are synced, and whether those events can actually be used when building audiences and automations. Shopify customer data and inventory data are both worth very little to you if they land a day late.

A shallow integration can connect the two systems and still leave out the data you need to run the program.

Segmentation should reflect how customers actually behave

Once that data is in, the platform has to make it usable. You should be able to build audiences from combinations of store behavior, purchase history, lifecycle stage and email engagement:

• customers who viewed a particular product at least twice in the last week and opened the last campaign
• customers whose purchase frequency is starting to decline
• buyers approaching their normal replenishment window
• customers showing strong interest in a particular category
• high-value repeat buyers
• customers who usually purchase only during promotions
• first-time buyers who have not yet made a second purchase

Every one of those is a combination, not a filter, and that is the actual test. Can the platform hold three or four signals at once, or does the marketer end up exporting the data and rebuilding the audience somewhere else? Segmenting beyond engagement is what those combinations look like in practice, and AI segmentation is the part that gets hard to maintain by hand.

Segmentation built on demographics, or on whether someone opened an email, will not get you to any of them.

Campaigns, flows, and reporting all need to be strong

An ecommerce brand needs both campaigns and automated flows. They fail in different ways, so a platform that is strong at one and weak at the other tends to reveal it within the first month.

Campaigns cover product launches, promotions, seasonal events, inventory-led opportunities and category pushes. Flows cover the recurring moments: welcome, cart abandonment, browse abandonment, post-purchase, replenishment, second purchase and win-back. Where the line between the two sits is a question of its own, and we have worked through it in campaigns vs automated flows.

The reporting has to go past opens and clicks. You should be able to answer whether first-time buyers came back, whether customers reordered sooner, whether different groups responded differently, whether a campaign drove repeat purchases, whether an offer lifted conversion but cost you margin, and whether a flow kept performing as customer behavior changed. The metrics that actually matter is the list we would grade a platform against.

The point is to connect the send to what the customer did afterwards, which is exactly what a default dashboard does not do.

The ESP should connect cleanly to the rest of the stack

The ESP rarely holds everything an ecommerce marketer needs. It has to work with the commerce platform (Shopify, in most cases), analytics, support, loyalty, reviews and creative tools.

Those connections should not require constant exports, manual uploads or workarounds. If important data sits in another system, the ESP should be able to reach it reliably and keep it current.

This matters more once AI is involved. An agent cannot make a good decision when customer data, campaign history, store behavior and product information are sitting in separate tools that do not talk to each other, which is what a typical marketing tool stack still looks like in most brands.

Look at whether AI agents can actually operate the ESP

Many ESPs now have AI features. Most include subject-line generation, copy assistance, send-time optimization, product recommendations and predictive segments.

Those are useful, and they are not what you should be grading. The question is whether an agent can reach the data and take real actions inside the system:

• MCP support
• deep APIs
• structured access to customer and campaign data
• ability to create and update segments
• ability to create campaigns and emails
• ability to create or modify flows
• access to performance data

Read access on its own buys you a better report. An agent that can only read will tell you a group of customers is approaching its replenishment window; one that can also write will build the audience, choose the products, draft the email and leave it waiting for approval. MCP use cases for ecommerce marketing covers what that access opens up, and the 2027 email stack post goes deeper on the writable half.

A platform with AI features inside it and a platform an agent can operate are two different products, however similar the pricing pages look.

Make sure agents can operate without giving up control

AI-operable does not mean unrestricted. The brand should still set:

• permissions
• approval requirements
• discount limits
• frequency rules
• excluded products
• customer suppression
• brand rules

The marketer decides what the agent does automatically, what it prepares for review, and what always needs a human. Those controls are what make the access safe to hand over in the first place, and they are the difference between an agent and a flow that runs whatever it was given a year ago.

Scalability belongs in the same conversation. A platform can look cheap and flexible while the list is small, then get difficult as data volume, campaign volume, API usage and flow complexity climb. Check that the pricing, APIs, data model, integrations and automation still work at the size you expect to be in two years, not the size you are now.

The basics are the floor, not the decision

Deliverability, segmentation, campaigns, flows, reporting and commerce integration are still the basics, and a platform weak on any of them is not a candidate. But the basics no longer separate anything, because almost everyone clears them.

What separates platforms now is whether store events arrive fast enough to act on, whether that data is easy to act on, and whether your marketers and an agent can both operate the program inside rules you set.

So the buying question is no longer only:

Can our team use this ESP well today?

It also has to be:

Will this ESP still work when more of the week's work is being done by an agent?

If you already have a platform and are weighing the move, should you switch ESPs and how to evaluate an AI email marketing platform are the two to read next.

Nexie is a retention marketing agent for e-commerce brands. It reads your customer data across engagement, recency, category interest and lifetime value together, surfaces the audiences worth campaigning to, and builds those campaigns for your team to review and approve, inside the guardrails and marketing policies you set. See how it works →