The engine makes every catalog different. That is only useful if you can tell what the difference did. The way to tell is to hold the strategy still and change the audience, so the same offer goes to different kinds of customer and the results can be compared. Groups are how you do that.

Groups are exclusive by rank

Customer groups are ordered, and a customer belongs to the highest group they match. This is not a detail; it is what makes comparison honest. If "bought this year" sits above "opened an email recently", a recent buyer who also opens emails is counted in the first group only. The second group is then really "opens emails but has not bought", which is a different person and a different result.

Order your groups from most specific to least. Put the small, valuable, clearly defined groups at the top and the broad catch-alls at the bottom. The editor shows how many match each group and how many were taken by a higher one.

Four groups most stores can define today

All from filters in the group editor; nothing has to be imported.

  • Recent buyers: has made a purchase, and days since last order under 60. These people already trust you. They are the group to learn depth from, because they will buy at a shallower discount than anyone else.
  • Quiet customers: has made a purchase, days since last order over 90, and days since last open or click over 90. Re-engagement is a group, not a campaign type. The same strategy sent here tells you what it takes to bring someone back, and the answer is usually not "deeper".
  • High value: lifetime revenue or average order value above a threshold you choose from your own numbers. Small group, top of the order, first to see anything limited.
  • First-timers: total order count equal to one. The second purchase is the one that makes a customer; a catalog built from what they bought is the most relevant offer you can send.

Two more filters are worth knowing. Days until predicted next order lets you reach customers just before they would reorder anyway, which is where a shallow offer does the most. Engagement tier separates people who open from people who do not, so you can send the non-openers something else, or nothing.

Run the same strategy across them

Create one strategy and one template. Run them, in turn or in parallel, to two or three of the groups above. Then open Analytics, switch the scope to each campaign, and compare the same numbers: orders, average order value, the funnel from email opened to checkout, and the Discount dependency chart. The Products tab shows what each group bought and at what depth.

What you learn is not a conversion rate for a discount. It is a map: recent buyers take shallow offers, quiet customers need the reminder more than the depth, first-timers buy the product next to the one they already own. The next strategy per group follows from that, and from then on each group has its own.

Re-engagement is cadence, not a special campaign

A quiet-customer group only works if the run reaches them at a sensible interval. A customer holding a live catalog is not emailed by another campaign until it ends, and every email counts on the bill, so set the group's campaign to a slower rotation than your recent buyers' and let the group definition do the selecting. When someone in it buys, they stop matching and drop out on the next sync; no manual cleanup.

Groups also cap. If a broad group matches more people than you want to spend catalogs on, cap it, and the overflow becomes claimable by the groups below.

For how often to run each group, see Campaign cadence. For what to send each one, see Three campaign strategies beyond "run a sale".