Cohort Segments: What the Month a Buyer Arrived Predicts

Two buyers who look identical today can behave completely differently, and the difference is often when they arrived. Cohorts are the segment nobody builds because they look boring.

A cohort segment groups buyers by when they first arrived, usually the month, and tracks how each group behaves over time. Cohorts reveal whether a change in the business helped or hurt, because they separate the buyers who came in under the old offer from the ones who came in under the new one.

Why arrival month matters more than age

Two buyers who look identical today can behave completely differently, and the difference is often when they arrived. A buyer who arrived during a big sale was trained to expect discounts from the first purchase. A buyer who arrived at full price was trained to pay full price. Both are buyers, both bought the same product, and both look the same in a demographic segment. The cohort is the segment that separates them, and the separation matters because the two groups will behave differently for the rest of their relationship with the store.

The conditions at arrival shape behavior in ways that last. A buyer who arrived through a referral behaves differently from a buyer who arrived through a paid ad, because the referral buyer came with trust already built in. A buyer who arrived during a product launch behaves differently from a buyer who arrived during a clearance sale. The cohort captures the conditions at arrival, and the conditions at arrival predict the future better than age, gender, or location ever will.

Related reading on this site: SegmentMuse and Segment Overlap: When One Buyer Sits in Three Segments cover the neighbouring problems in more depth.

What cohorts reveal that RFM hides

RFM groups buyers by what they did: how recently, how often, how much. It is a good segment, and it tells you who is valuable now. It does not tell you whether your recent buyers are getting better or worse, because RFM mixes buyers from every arrival month into the same buckets. A store whose RFM looks stable can be hiding a decline: the old buyers are still valuable, the new buyers are worse, and the average has not moved yet because the old buyers are still carrying it.

Cohorts show the trend. A cohort of buyers from January, tracked over six months, shows how January buyers behave. A cohort from June shows how June buyers behave. If the June cohort is worse than the January cohort at the same age, the business got worse, and the RFM average will not show it for months. The cohort is the early warning system, and it is the only segment that shows it.

Building your first cohort segment

Start with monthly cohorts for the last twelve months. That is enough to see trends without drowning in groups. Each cohort is the buyers who first purchased in that month, tracked forward: how many bought again, how much they spent, and how long they stayed. The table is simple, and the story it tells is clear. A cohort that flattens after the first purchase is a cohort that did not stick. A cohort that keeps buying is a cohort that did.

Expand to weekly only if a specific launch or campaign needs its own cohort. A Black Friday cohort, a product launch cohort, a referral campaign cohort: each one is a group that arrived under specific conditions, and each one deserves its own line in the table. The rest of the time, monthly is enough. The point is not to build a perfect cohort system. The point is to stop averaging buyers from different eras together and calling the average a segment.

Frequently asked questions

What is a cohort segment?

A group of buyers defined by when they first arrived, usually the month of their first purchase. Each cohort is tracked over time, so you can see how buyers from January behave six months later versus buyers from June.

Why segment by arrival month?

Because the conditions at arrival shape behavior. Buyers who arrived during a big sale behave differently from buyers who arrived at full price. Without cohorts, those two groups average together and hide the difference.

How are cohorts different from RFM segments?

RFM groups buyers by what they did: how recently, how often, how much. Cohorts group buyers by when they arrived. RFM tells you who is valuable now. Cohorts tell you whether your recent buyers are getting better or worse.

How many cohorts do I need?

Start with monthly cohorts for the last twelve months. That is enough to see trends without drowning in groups. Expand to weekly only if a specific launch or campaign needs its own cohort.