How many of your subscribers get an offer that doesn’t apply to them? On a list of 20,000 contacts, a good chunk never opens, another clicks on everything, and a third barely exists anymore. Segmenting means no longer sending the same message to these three audiences. This guide breaks down the criteria that work and the RFM method. One angle is missing from most articles on the topic though: the quality of the data itself.
Segmenting by demographic and geographic data
It’s the simplest filter to set up. Age and language form the basis, location fine-tunes the sending window, and industry adds useful context in B2B. These fields already exist in most CRMs and sync directly to the ESP via API or webhook. An email in English sent to a French-speaking contact loses its open before it’s even read.
Two limits to keep in mind. Too many fields at signup scares off the prospect before the first send even happens. And a demographic criterion alone says nothing about real engagement: a 35-year-old contact in Paris might just as easily ignore every campaign.
Segmenting by behavior: opens, clicks, browsing
Behavior reveals what demographics can’t show. A contact who clicks on three consecutive campaigns deserves different treatment than one who hasn’t opened an email in six months. Most ESPs, Brevo included, let you build this type of segment without any development work:
- Define the observation window: 30, 60, or 90 days depending on send frequency.
- Create an “engaged” segment (at least one open or click within the window).
- Create an “inactive” segment (zero interaction over the same period).
- Isolate clicks on specific links to trigger automated scenarios (cart recovery, content tied to a detected interest).
Responsiveness to previous emails carries more weight than any demographic field when it comes to predicting the next open.
Combining recency, frequency, and spend with the RFM method
The RFM method ranks each contact according to three measures. Recency indicates the date of the last purchase or interaction. Frequency counts the number of purchases over a given period, and a third indicator, total amount spent, completes the score.
Each axis is scored from 1 to 5, producing a three-digit score. A 555 contact is a champion: recent purchase, frequent, high cart value. A 111 contact has almost no measurable value. In between, intermediate segments point to distinct scenarios: reactivation for at-risk contacts, premium offers for dormant big spenders.
This is the basis for calculating customer lifetime value (LTV). A B2B email campaign that treats a qualified lead like a standard newsletter subscriber wastes its sending budget on the wrong segment.
Segmenting by data quality and validity
Demographic, behavioral, and RFM criteria all share one weakness: they assume the email address behind each contact still works. That’s not always true. More than 20% of an email contact database becomes obsolete every year, between job changes and closed mailboxes. Typos made at signup add to that total, according to data published by Mailgun1.
A segment built on dead addresses skews everything else. The open rate calculated on that segment looks catastrophic. The real cause: mailboxes that no longer exist. Every send to a vanished address generates an SMTP rejection (hard bounce). Mailbox providers use this to judge sender reputation.
Data quality is a segmentation criterion in its own right, not an afterthought. In practice: isolate a “to verify” segment. It groups addresses never opened since being added, plus recent signups with no first open. Domains with questionable syntax round out the filter. Have this segment verified before the next send, not after a reputation alert on your sending domain. That’s the difference between a scheduled check and a scramble after a deliverability drop.
What segmentation actually changes
The most cited study on the topic remains Mailchimp’s, conducted on roughly 2,000 accounts that sent nearly 11,000 segmented campaigns to 9 million recipients2. Compared to non-segmented campaigns, segmented campaigns show:
- +14.31% open rate
- +100.95% click rate
- -4.65% bounces
- -3.90% spam complaints
- -9.37% unsubscribes
The breakdown by segmentation type confirms the previous point: activity-based segmentation lowers bounces by 9.23% on top of improving clicks. Open rate remains the most closely watched metric. The drop in bounces, however, weighs just as heavily on sender reputation over the medium term.
GDPR: what consent changes in segmentation
Every piece of data used for segmentation must correspond to explicit consent (opt-in), given at the exact moment it was collected. An “industry” field filled in for a newsletter can’t be used to build undisclosed commercial targeting.
Document what’s processed, for what use, and for how long. That also protects the relationship with the subscriber. A contact who understands why they’re receiving a given email rarely flags the campaign as spam.
Metrics to track after implementation
Four numbers are enough to manage a segmentation strategy over time:
- Open rate by segment: compares subject line appeal across groups.
- Click rate by segment: measures real interest in the content sent.
- Conversion rate: concrete actions (purchase, signup, reply) after the email.
- Unsubscribe rate: signals fatigue or a poorly matched segment.
Engagement rate combines several of these signals into a single measure. It’s useful for comparing segments of different sizes without opening four separate dashboards.
A well-segmented list is still a living list: it needs to be verified as much as it needs to be sliced up.
