Segmentation
Segmentation splits store visitors into cohorts you can analyse and act on — by source, device, geography, or behaviour. It's the prerequisite to any working personalization program.
Segmentation
Dividing store visitors into meaningful cohorts — by source, device, geography, behaviour, or purchase history — so you can analyse and act on them separately.
Segmentation is the practice of grouping visitors and customers into cohorts that behave similarly enough to be worth treating as one unit. In an online store that usually means slicing traffic by acquisition source, device, country, new-vs-returning status, on-site behaviour, or order history.
Good segmentation is the foundation that personalization sits on: until you can reliably tell a paid-social first-time mobile visitor apart from a returning desktop customer with three prior orders, any attempt to show them different content is guesswork. Segmentation is the analysis layer; personalization is what you do with it.
Most stores already segment without naming it — looking at mobile-only conversion rate, or comparing Meta versus Google traffic. Formal segmentation just makes that habit consistent: the same cohort definitions, applied across reports, tests, and email flows.
The useful segments are the ones that move differently. If two cohorts convert at 2.1% and 2.0%, splitting them buys you nothing. If one converts at 1.2% and the other at 4.8%, you've found a lever — and probably a hypothesis for your next A/B test.
Segment Value = Segment Size × Segment Conversion Rate × Segment AOV
Segment Size
Visitors in segment
Monthly sessions or users matching the segment definition.
Segment Conversion Rate
Conversion rate
Orders divided by sessions for this segment only.
Segment AOV
Average order value
Mean order value for this segment's purchases.
A Shopify apparel store evaluates its 'returning mobile visitors from email' segment to decide whether it's worth a dedicated landing experience.
Segment Size (monthly sessions): 18,000
Segment Conversion Rate: 5.4%
Segment AOV: €72
→ €69,984 / month
That segment alone drives roughly €70k/month. A 10% lift on it is worth ~€7k/month — easily enough to justify a dedicated mobile landing page and a tailored discount ladder.
Use segment value to prioritise. Small segments with huge conversion rates are interesting but rarely worth a custom experience; large segments with mediocre rates are usually where the money is hiding.
Typical Shopify-store conversion rates by segment (apparel & beauty, AOV €40–€90)
| Segment | Sessions share | Conversion rate | AOV |
|---|---|---|---|
| New visitor, paid social, mobile | 32% | 1.1% | €48 |
| New visitor, organic search, desktop | 14% | 2.6% | €64 |
| Returning visitor, direct, mobile | 11% | 4.2% | €58 |
| Returning visitor, email, mobile | 9% | 5.4% | €72 |
| Returning customer (2+ orders), any source | 7% | 7.8% | €86 |
| Paid search, branded keyword | 6% | 6.1% | €61 |
| Referral / affiliate | 5% | 2.9% | €55 |
| All traffic (blended) | 100% | 2.4% | €58 |
Notice the spread: branded paid search and returning-customer cohorts convert 5–7× better than cold paid-social mobile. A single blended conversion rate hides all of that — and hides every decision worth making.
Frequently asked questions about segmentation
Segmentation is how you group visitors; personalization is what you serve each group. Segmentation lives in the analytics layer and is fundamentally about understanding behaviour. Personalization is the action layer that sits on top — different copy, products, or offers for each segment.
For most stores under €15M revenue, five to eight active segments is the sweet spot. Fewer and you miss real behavioural differences; more and you can't staff the experiences. Start with traffic source × device × new-vs-returning and expand from there.
For analytics, roughly 1,000 sessions/month is enough to spot meaningful patterns. For A/B testing a segment, you typically need enough traffic to reach statistical significance within four weeks — usually 5,000+ sessions and 100+ conversions in the segment per variant.
RFM is excellent for email and retention strategy — it sorts existing customers by how valuable they are right now. It's less useful for on-site segmentation because most sessions are anonymous or first-time visitors who have no R, F, or M yet.
A cohort is a segment defined by a shared start point in time — for example, 'customers who first ordered in March'. All cohorts are segments, but not all segments are cohorts. Cohorts are the right tool when you care about behaviour over time (retention, repeat-purchase curves).
Yes — by device, geography, traffic source, landing page, on-site behaviour, and any first-party signals you collect in-session. You can't segment them by purchase history until they identify themselves at checkout or via email capture, which is why progressive identification matters.
GA4 handles descriptive segmentation reasonably well — comparisons, audiences, exploration reports. Where it falls short is segment-level experimentation, fast iteration on segment definitions, and triggering personalized experiences on-site. Most stores layer a dedicated tool on top for those use cases.
In practice: new vs returning, mobile vs desktop, paid vs organic source, high-intent landing pages (PDPs, category pages) vs low-intent (blog, homepage), and customers by order count. These five splits surface most of the actionable variance on a typical Shopify store.
Quarterly is a reasonable cadence. Definitions drift as traffic mix shifts — a 'paid social mobile' segment in 2022 behaves very differently in 2025 after iOS attribution changes. Review when a major channel or campaign type changes meaningfully.
Yes. Even with no personalization, segmentation sharpens every decision: which channels to invest in, which landing pages to rebuild, which audiences to test offers against, and where to focus CRO experiments. Personalization is the eventual payoff, not the prerequisite.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.