Product recommendation emails can easily become generic. A customer bought one product, so you show them a few more or you fill an email with best sellers and call it personalized.
That is not enough.
A product recommendation campaign starts by identifying a customer group whose behavior gives you a reason to recommend something next.
Then you work out which product is most relevant, why it fits, and when the recommendation should arrive.
Find the behavior that earns a recommendation
Before deciding what to recommend, first identify which customers are showing a pattern worth acting on.
That could come from:
- what they bought before
- what they have been browsing recently
- which products people like them usually buy next
- whether their category interest is changing
- whether they are approaching a normal reorder window
- whether they have responded to previous cross-sell campaigns
For example, say the people who buy one skincare product tend to come back for a second one five or six weeks later. That pattern is a stronger starting point than a decision to promote the second product this week.
The behavior gives you a reason to build the campaign.
Decide what kind of recommendation you are making
Product recommendations can come from very different customer situations.
| Recommendation type | Example |
|---|---|
| Complementary | Bought shoes, recommend relevant socks or apparel |
| Next purchase | First purchase is often followed by another category |
| Category expansion | Customer is showing interest in a category they have not bought from before |
| Similar alternative | Customer repeatedly browses a product but has not purchased |
| Replenishment-related | Customer is nearing a repeat purchase and may also need a related product |
The kind of recommendation narrows which products are worth considering.
- A complementary recommendation should help complete or extend the current or original purchase.
- A next-purchase recommendation should reflect what similar customers tend to buy after the first order.
- A category-expansion recommendation should be based on the customer showing real interest in something new.
A recommendation lands better when it fits what the customer has already browsed or bought than when it simply shows more products.
Use more than one signal to choose the product
Most catalogues offer several plausible next products for the same customer. One signal rarely tells you which one belongs in the email.
Suppose a customer bought running shoes six months ago. That tells you something about their interests, but not necessarily what they want now.
If that same customer has recently started browsing running shorts, and customers with similar purchase patterns often buy apparel after footwear, the recommendation becomes much stronger.
You can combine signals such as:
- previous purchases
- recent browsing
- products similar customers buy next
- category preference
- price range
- past response to recommendations
- current product availability
The more evidence points toward the same product or category, the more confident you can be about the send.
Make sure there is a reason to send it now
A good product recommendation can still fail if it arrives at the wrong time.
The timing should come from:
- a recent purchase
- repeated browsing
- a change in category interest
- a normal next-purchase window
- a new product that fits previous behavior
- renewed activity after a quiet period
For example, three days after a purchase the customer has barely used the product, so a recommendation has little to attach to. The same email lands differently once they come back and browse the category themselves. The behavior, not the calendar, is what makes the send make sense.
The recommendation and the timing should be decided together.
Make the recommendation feel relevant
The email needs to make it clear why the product is being shown, and that does not mean over-explaining the recommendation.
A simple line can be enough:
- “Pairs well with what you bought”
- “Customers who bought this often come back for…”
- “Based on what you’ve been looking at…”
- “A useful addition to your routine”
- “New in a category you’ve been exploring”
The wording will depend on the brand and the campaign.
What matters is that the recommendation does not feel like a random product grid. The customer should be able to understand why these products are relevant to them.
Relevance first, incentive only if it helps
A strong recommendation may not need an incentive.
If the customer is already showing interest, or the product is a natural next purchase, a discount may simply reduce margin without changing the decision.
Start with relevance. Use an offer when the customer behavior suggests they need more incentive, or when the economics of the campaign justify it.
- For some customers, no offer may be necessary.
- For others, a bundle, free shipping threshold, or small discount may make sense.
The recommendation should come first; the incentive should only be added if it helps.
Change the recommendation when the customer behavior changes
Product recommendation campaigns cannot run on autopilot.
- If the customer buys the recommended product, the campaign should stop.
- If they begin showing stronger interest in another category, the next recommendation should change.
- If they add a different product to cart, that behavior may become more important than the recommendation campaign.
This is especially important when recommendation campaigns sit alongside flows such as browse abandonment, cart abandonment, or replenishment. The latest customer behavior should influence which message gets priority.
Measure whether the recommendation changed what customers bought
Opens and clicks can help you understand whether the email got attention, but the better question is whether the recommendation actually changed what the customer bought.
Look at:
- conversion on recommended products
- revenue per recipient
- cross-sell rate
- second-purchase rate
- average order value where relevant
- whether customers went on to buy from the recommended category
It is also useful to compare recommendation methods.
For example:
- Does a behavior-based recommendation outperform simply showing best sellers?
- Does a recommendation based on recent browsing work better than one based only on past purchases?
That tells you whether the recommendation logic itself is improving.
A simple product recommendation campaign workflow
| Step | Question |
|---|---|
| Find the opportunity | Which customer behavior suggests another product may be relevant? |
| Select the audience | Which customers show that behavior strongly enough? |
| Choose the product | What is the most relevant next product or category? |
| Check timing | Why should this recommendation be sent now? |
| Build the message | Why does this product make sense for this customer? |
| Decide on incentive | Is relevance enough, or is an offer actually needed? |
| Measure | Did customers buy the recommended product or category? |
A recommendation campaign is a sequence of decisions rather than a product selection: which behavior justifies contact, which kind of recommendation that behavior implies, which product fits it, and when it should arrive. The first decision constrains every one after it.
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 →