Most ecommerce brands already segment customers using engagement, recency, frequency, and lifetime value. These signals are useful because they tell you whether someone is still active, how recently they bought, how often they purchase, and how valuable they have been to the business.
But they still leave out a large part of the customer picture.
They do not tell you what a customer is interested in right now, whether they usually buy at full price or wait for discounts, what price range they tend to shop in, whether they are moving into a new category, or which products they are actively considering.
For ecommerce brands, those details can matter more when deciding what campaign to send.
Start with the usual suspects
Engagement, recency, frequency, and LTV are still a good base.
They help you answer broad questions about the customer. Are they active? Have they purchased recently? Do they buy often? Are they valuable to the business?
That gives you a first layer of segmentation. You can separate active customers from inactive ones, frequent buyers from occasional buyers, and high-value customers from everyone else.
The problem that remains is that these segments are still broad from a campaign point of view.
A group of high-LTV customers may include people interested in very different categories. Two highly engaged customers may respond very differently to the same promotion. A frequent buyer may prefer premium products, while another frequent buyer may almost always shop during sales.
If you stop at the usual segmentation signals, you still know relatively little about what each customer should actually receive.
Use more of the customer data your store already has
Ecommerce brands usually have several other signals that can make segmentation much more specific.
One of the strongest is category and product interest.
Purchase history tells you what someone has already bought. Browsing behavior can tell you what they are interested in now. Repeated product views, category visits, and movement into a new part of the catalogue can all help identify what the customer is considering.
A sports retailer, for example, may know that one customer is mainly interested in running products while another spends most of their time looking at tennis equipment. Both customers may be equally engaged and have similar LTV, but they should not automatically receive the same campaign.
Discount sensitivity is another signal.
Some customers regularly buy at full price, while others tend to purchase only when there is a promotion. That difference can help decide whether an offer is needed at all.
A customer who has repeatedly bought without discounts may be a good audience for a new-product or early-access campaign, while one who usually converts only during promotions may need a different message or offer. This also matters for protecting margin. If someone is likely to buy without a discount, there is little reason to train them to wait for one.
You can also look at price preference.
Two customers may both be interested in skincare, but one may usually buy premium products while the other stays within a lower price range. That should influence which products you recommend.
Average order value can help here, but product-level purchase history is often a better guide because it shows what the customer actually chooses.
Current intent can add another layer. Historical purchases tell you what someone liked before. Recent browsing tells you what they may want now.
If a customer who usually buys skincare starts repeatedly browsing haircare products, that may matter more than their old category history for the next campaign. The same applies when someone begins viewing a specific product several times without purchasing.
For some businesses, gender can also matter when it materially affects product relevance. The same applies to geography, subscription status, loyalty status, or other customer attributes. These should not be used just because the fields exist. They are worth using only when they change what the customer should receive.
There are other signals that can matter depending on the business:
- how often someone usually repurchases
- whether they tend to buy replenishable products
- whether they usually buy one product or build larger baskets
- whether they move between related categories
- whether they regularly return products
- whether they tend to buy new launches or established products
- whether they usually buy for themselves or appear to shop around gifting periods
The goal is not to use all of these. It is to understand which ones actually improve the campaign.
Group signals by the decision they help you make
One way to think about segmentation is to group signals by what they help you decide.
| Signal group | Examples | What it helps decide |
|---|---|---|
| Customer value | LTV, AOV, purchase frequency | How important the customer is and how much attention they may deserve |
| Product interest | Categories bought, products viewed, recent browsing | What products or categories to promote |
| Buying behavior | Discount sensitivity, full-price behavior, preferred price range | Whether to use an offer and what products to show |
| Current intent | Recent product views, repeated browsing, cart activity | What the customer may be considering now |
| Purchase patterns | Repurchase timing, replenishment behavior, cross-category movement | When to send and what to recommend next |
| Customer attributes | Gender, geography, loyalty status, subscription status | Whether the campaign needs to change for a specific group |
This gives you a much better way to build segments than simply adding more conditions.
The question is not, "How many customer fields can we combine?"
It is, "What do we need to know to make this campaign more relevant?"
Combine signals when the combination changes the campaign
The strongest segments often combine two or three signals.
A segment based only on running interest tells you what category the customer likes, whereas adding discount sensitivity tells you whether that customer is likely to need an offer. A segment based on skincare interest tells you more when you also know whether the customer usually buys premium products or lower-priced ones.
For example:
| Customer segment | Possible campaign |
|---|---|
| Running interest + low discount sensitivity | New running collection without leading with a discount |
| Running interest + high discount sensitivity | Running promotion with an offer |
| Skincare interest + premium price preference | Premium skincare launch |
| Skincare interest + lower price preference | Product recommendations within a lower price range |
| Tennis interest + repeated racket views | Campaign focused on the rackets they are considering |
| Haircare browsing + existing skincare customer | Cross-category campaign introducing haircare |
| Replenishable product buyer + usual repurchase window approaching | Replenishment campaign |
| Relevant gender segment + category interest | Campaign featuring products that are more relevant to that audience |
You can still add engagement, recency, frequency, or LTV when they improve the segment, but those signals should support the campaign rather than become the entire campaign logic.
Do not make segments more complicated than they need to be
More granular segmentation is worth it only when the extra detail changes what you send.
It is easy to create a segment such as:
high-LTV customers + running interest + low discount sensitivity + recent browsing + purchased in the last 60 days
That may sound sophisticated, but if the last condition does not change the campaign, it adds complexity without adding value.
Every additional condition also makes the audience smaller.
The rule is simple: if removing a condition would not change the product, offer, message, timing, or campaign, you probably do not need that condition. The goal is not to build the most detailed segment possible, it is to build the smallest segment definition that gives you a clear reason to market differently.
Keep segments updated as customer behavior changes
Customer interests do not stay fixed.
- Someone who mainly bought running products last year may start browsing cycling products.
- A customer who used to buy only during promotions may begin purchasing at full price.
- Someone who previously bought lower-priced products may gradually move into a premium range.
Segments should reflect those changes.
This is especially important for product interest and intent. Recent behavior may tell you more about what the customer wants now than purchases they made a year ago. That means segmentation should not only be about creating customer groups once, but also about updating those groups as new behavior comes in.
AI can make deeper segmentation practical
This level of segmentation becomes difficult to manage manually.
A marketer may want to consider category interest, product views, discount sensitivity, price preference, purchase cadence, LTV, and recent behavior at the same time. The number of possible combinations can grow quickly.
AI can help create those segments, keep them updated, and identify which combinations are worth acting on.
It can also help identify campaigns for each group.
For example, AI may find a group of customers who are interested in running products, usually buy at full price, and have recently viewed a new shoe collection. That can become a clear campaign opportunity without the marketer manually creating the segment first.
Another group may have the same category interest but a history of buying only during promotions. The campaign may use the same products but a different offer.
Once the audience and campaign are clear, AI can also use that context to help generate the email.
The marketer still decides whether the segment and campaign make sense before anything is sent.
Engagement, recency, frequency, and LTV are useful ways to understand the basic state of a customer. But ecommerce brands usually know much more than that.
Category and product interest, discount sensitivity, price preference, current intent, purchase patterns, and other relevant customer data can help determine what campaign a customer should receive and how that campaign should change.
That is the real value of going beyond engagement.
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 →