Adobe Commerce’s customer segments are the bridge between “we have customer data” and “the store behaves differently for different customers”. Segments update dynamically as customers match or stop matching rules - and they drive banner content, cart price rules and related-product blocks. Used well they replace a surprising amount of email-only personalisation with on-site relevance.
What a Segment Is
A segment is a saved set of conditions evaluated against customers continuously: address, order history, cart contents, customer attributes. “VIP” might be: lifetime orders over £2,000, or 3+ orders in 90 days. “Lapsing” might be: no order in 180 days and lifetime orders above zero.
Customers flow in and out automatically. That dynamism is the point - a VIP who stops ordering leaves the VIP segment and stops seeing VIP treatment without anyone maintaining a list.
Where Segments Drive Behaviour
- Cart price rules: restrict promotions to segments (“VIP early access pricing”) without coupon codes
- Dynamic blocks / banners: show different CMS content per segment - hero for new visitors, loyalty message for repeat buyers
- Related products and upsell rules: segment-scoped product suggestions
Note the platform boundary: segments are an Adobe Commerce (paid) feature. Open Source achieves similar outcomes with customer groups plus cart rule conditions, third-party personalisation platforms, or custom modules - the segment UI itself is Commerce-only.
Practical Patterns That Work
- New vs returning: different homepage hero. Simple, visible, low risk.
- Category affinity: segment by past category purchases; show relevant cross-sells. High effort, high reward.
- B2B vs B2C on shared storefronts: segment by company association; hide consumer promotions from trade accounts (and vice versa - trade customers seeing retail discounts causes awkward phone calls)
- Geographic: region-specific shipping messaging before checkout, which measurably reduces cart abandonment surprises
Performance Reality
Segment matching happens per session against the customer’s data, and segment-dependent content busts full page cache for personalised blocks. Magento handles this via customer-data sections (hole-punching), but every personalised block is cache-exempt work. Two or three personalised areas per page is fine; twenty is a TTFB problem. Personalise the high-value real estate - hero, cart message, one recommendation rail - not everything.
Measuring Honestly
Personalisation programmes die on unmeasured claims. Before/after conversion rate by segment, A/B-tested where the tooling allows, is the only honest accounting. The patterns above earn their keep in most retail contexts; the fancier the personalisation tree, the more sceptical you should be.
Start with two segments and one personalised block. Prove the uplift, then expand. That discipline, not the tooling, is what separates personalisation that pays from personalisation that just adds cache misses.