--- title: "Segmentation and personalisation | Optimize All Academy" description: "Relevance drives results Sending the same email to everyone is simple but blunt. Subscribers who receive irrelevant emails ignore them, unsubscribe or…" url: https://optimizeall.com/learn/email-marketing-and-automation/segmentation-and-personalisation updated: 2026-10-05 --- Email Marketing & Automation · Consent-based list building and segmentation · lesson 2 of 16 · 14 min # Segmentation and personalisation ## Relevance drives results Sending the same email to everyone is simple but blunt. Subscribers who receive irrelevant emails ignore them, unsubscribe or mark them as spam, which hurts deliverability for everyone on your list. **Segmentation** means dividing your list into groups that receive different messages; **personalisation** means adapting content to the individual. ## Data you can segment on | Data type | Examples | How you get it | |---|---|---| | **Declared (zero-party)** | Interests, preferences, skin type, budget, language | Sign-up forms, quizzes, preference centres | | **Behavioural** | Emails opened or clicked, pages viewed, products browsed | Email platform and website tracking (with consent where required) | | **Transactional** | Purchases, order value, frequency, last purchase date | E-commerce or CRM integration | | **Lifecycle stage** | New subscriber, first-time buyer, repeat customer, lapsed | Combination of the above | | **Location and language** | Country, city, preferred language | Forms, store data | Only collect data you will use and that your privacy notice covers. ## High-impact starter segments 1. **New subscribers** (joined in the last 30 days) – still forming an opinion; send onboarding content. 2. **Engaged subscribers** (clicked in the last 60–90 days) – your most responsive audience. 3. **Unengaged subscribers** (no clicks in many months) – candidates for re-engagement or suppression. 4. **Customers vs non-customers** – different messages and offers. 5. **Repeat or high-value customers** – loyalty perks, early access. 6. **Interest segments** – for example "menswear" vs "womenswear", or "beginner" vs "advanced". 7. **Language or region** – Arabic and English versions for Gulf audiences; UK vs US spelling, currency and holidays. ## The RFM model for e-commerce **RFM** scores customers on: - **Recency:** how recently they bought. - **Frequency:** how often they buy. - **Monetary value:** how much they spend. Customers who bought recently, often and with high spend are your champions; those who used to buy often but have not recently are at risk. Many email platforms calculate RFM-style segments automatically. ## A note on open-based segmentation Since Apple's Mail Privacy Protection (introduced in 2021), emails opened in Apple Mail can be pre-loaded, which records an "open" even if the person never read it. This inflates open rates and makes "opened in the last 30 days" segments unreliable. Use **clicks, site visits and purchases** as stronger engagement signals. ## Personalisation that helps, not creeps Useful personalisation: - First name in the greeting (with a sensible fallback like "Hi there" when the name is missing). - Product recommendations based on browsing or purchase history. - Content blocks that change by interest or location (dynamic content). - Send-time optimisation where the platform supports it. - Language and currency matching. Avoid personalisation that feels invasive ("We noticed you looked at this item 7 times at 2am") or that reveals sensitive information. Never infer or reference sensitive characteristics such as health or religion without explicit, appropriate permission. ## Preference centres A **preference centre** lets subscribers choose topics and frequency instead of unsubscribing completely. Offer options such as "weekly digest only", "offers only" or "pause for a month" (useful around exams, Ramadan or busy seasons). This keeps more subscribers and improves relevance. ## Worked example A UK and UAE online bookstore segments by: - **Language:** English or Arabic newsletters. - **Genre interests:** from a sign-up quiz (fiction, business, children's). - **Lifecycle:** new subscribers, first-time buyers, repeat buyers, lapsed buyers. The weekly newsletter uses dynamic blocks to show each reader's preferred genres, and repeat buyers get early access to signed editions. Unengaged subscribers move to a monthly digest before any suppression. ## Common mistakes - Blasting every campaign to the whole list. - Relying on opens as the main engagement signal. - Collecting data you never use. - Personalisation errors like "Hi {FIRST_NAME}" (test merge tags and fallbacks). ## Hands-on: segment definitions you can build in any ESP Most email platforms (Klaviyo, Mailchimp, Brevo, HubSpot and others) let you build segments from conditions. Write them in plain language first, then build: ```text ENGAGED 90D Clicked any email in last 90 days OR placed order in last 90 days OR (signed up in last 30 days) UNENGAGED 180D Received >= 10 emails AND no clicks in 180 days AND no orders in 365 days NEW CUSTOMERS Placed exactly 1 order, first order in last 60 days REPEAT / VIP Placed >= 3 orders OR total spend in top 10% of customers AT RISK Placed >= 2 orders AND last order > 1.5 x their usual reorder gap BROWSED NOT BOUGHT Viewed product in category X in last 14 days AND no order in 14 days LANGUAGE = AR Preferred language (from form or site) = Arabic ``` Use **clicks and orders**, not opens, as engagement signals (the next module explains why opens are unreliable). ## Hands-on: RFM scores in Google Sheets Export customers with columns: customer_id (A), days since last order (B), number of orders (C), total spend (D). Score each 1–5 by quintile: ```text R score E2: =6-MATCH(PERCENTRANK.INC($B$2:$B$1000,B2),{0,0.2,0.4,0.6,0.8}) F score F2: =MATCH(PERCENTRANK.INC($C$2:$C$1000,C2),{0,0.2,0.4,0.6,0.8}) M score G2: =MATCH(PERCENTRANK.INC($D$2:$D$1000,D2),{0,0.2,0.4,0.6,0.8}) Segment H2: =IF(AND(E2>=4,F2>=4),"Champion",IF(AND(E2<=2,F2>=4),"At risk", IF(E2>=4,"Recent",IF(E2<=2,"Lapsing","Steady")))) ``` (Recency is inverted: fewer days since the last order scores higher.) Many platforms calculate similar segments automatically; the sheet helps you understand and check them. Use customer IDs, not names or emails, in any file you share. ## Worked example 2: a Karachi electronics retailer The retailer used to send every promotion to its whole list of 60,000. It creates four segments: engaged buyers (weekly offers), engaged non-buyers (buying guides and reviews), at-risk customers (a service check-in and accessory offer) and unengaged subscribers (monthly digest, then re-engagement). Sends to the unengaged group fall sharply, complaint rates drop and revenue per recipient rises on every campaign – with fewer emails sent. ## How to measure success - **Revenue or clicks per recipient by segment**, compared with the old whole-list sends. - **Share of the list that is engaged** (clicked or bought in 90 days) trending up. - **Unsubscribe and complaint rates** falling on promotional sends. ## Video lecture: Segmentation and personalisation: relevance without the creepiness Lecture coming soon · 12 chapters · about 8 minutes. Read the full transcript below. 1. Segmentation and personalisation 2. Five data types 3. Starter segments 4. Opens are unreliable 5. RFM scoring 6. Example 1: UK/UAE bookstore 7. Example 2: Karachi electronics (illustrative) 8. Helpful, not creepy 9. Zero-party data 10. Preference centre + mistakes 11. Measure success 12. Recap and try this now ## Lecture transcript ### Segmentation and personalisation Imagine a shop assistant who greets every customer with the same line, whether they're a first-time visitor, a loyal regular or someone returning a faulty product. Awkward, right? That's what sending the same email to your whole list feels like. Irrelevant emails get ignored, unsubscribed from or marked as spam, and that hurts deliverability for everyone on your list. In this lecture you'll learn what data you can segment on, the high-impact starter segments, how RFM scoring works with a real spreadsheet, and how to personalise in ways that help rather than creep people out. ### Five data types Let's define terms. Segmentation means dividing your list into groups that receive different messages. Personalisation means adapting content to the individual. And you can segment on five kinds of data. Declared data, also called zero-party, which people tell you directly, like interests, skin type or language, from forms, quizzes and preference centres. Behavioural data: emails clicked, pages viewed, products browsed. Transactional data: purchases, order value, frequency and last purchase date. Lifecycle stage: new subscriber, first-time buyer, repeat customer, lapsed. And location and language. One rule: only collect what you'll use and what your privacy notice covers. ### Starter segments Now the starter segments that give the biggest gains. New subscribers from the last thirty days, still forming an opinion. Engaged subscribers who clicked or bought in the last ninety days, your most responsive group. Unengaged subscribers with no clicks for many months, candidates for re-engagement or suppression. Customers versus non-customers, who need different messages and offers. Repeat or VIP customers, for loyalty perks and early access. Interest segments, like menswear versus womenswear. And language or region, like Arabic and English versions for the Gulf. The plain-language definitions are in the lesson text, ready to build in any email platform. ### Opens are unreliable One important warning about engagement. Since Apple's Mail Privacy Protection arrived in twenty twenty-one, emails opened in Apple Mail can be pre-loaded automatically, which records an open even if the person never read it. That inflates open rates and makes opened in the last thirty days segments unreliable. So build engagement segments on clicks, site visits and purchases instead. We'll cover this properly in the deliverability module. For now, remember: an open is a hint, not a fact. A click or an order is a fact. ### RFM scoring Now RFM, a classic model for e-commerce. It scores customers on Recency, how recently they bought; Frequency, how often they buy; and Monetary value, how much they spend. Customers who bought recently, often and with high spend are your champions. Those who used to buy often but haven't recently are at risk. Many email platforms calculate RFM-style segments automatically, but building it once in a spreadsheet helps you understand it. Export customer ID, days since last order, number of orders and total spend. Score each from one to five by quintile, and label segments like champion, at risk, recent and lapsing. The formulas are in the lesson text. ### Example 1: UK/UAE bookstore Let's do a simple worked example. A bookstore selling in the UK and UAE segments by language, English or Arabic newsletters; by genre interest from a sign-up quiz, fiction, business or children's; and by lifecycle, new subscribers, first-time buyers, repeat buyers and lapsed buyers. The weekly newsletter uses dynamic blocks to show each reader their preferred genres. Repeat buyers get early access to signed editions. And unengaged subscribers move to a monthly digest before any suppression. One newsletter template, many relevant versions. That's the power of combining segmentation with dynamic content. ### Example 2: Karachi electronics (illustrative) Now a realistic business scenario, with illustrative outcomes. A Karachi electronics retailer used to send every promotion to all sixty thousand subscribers. It creates four segments. Engaged buyers get weekly offers. Engaged non-buyers get buying guides and reviews. At-risk customers get a service check-in and an accessory offer. And unengaged subscribers get a monthly digest, then a re-engagement sequence. The result: far fewer emails go to the unengaged group, complaint rates drop, and revenue per recipient rises on every campaign, with fewer emails sent in total. Relevance isn't just nicer. It's more profitable. ### Helpful, not creepy Personalisation should help, not creep. Helpful: first name in the greeting, with a sensible fallback like hi there when it's missing. Product recommendations based on browsing or purchases. Content blocks that change by interest or location. Send-time optimisation. Matching language and currency. Creepy: we noticed you looked at this item seven times at two a.m. And never infer or reference sensitive characteristics, like health, pregnancy or religion, without explicit, appropriate permission. A good test: would you be comfortable if the customer saw exactly why they received this email? If not, don't send it. ### Zero-party data Where does the best segmentation data come from? Often, from simply asking. Zero-party data is information people give you on purpose: a skin-type quiz, a what are you shopping for question at sign-up, a preferred language, a budget range. It's accurate, it's consented, and it makes your first emails relevant from day one, before you have any purchase history. Keep it short, one to three questions, and show people the benefit: tell us your skin type and we'll send routines that suit you. Then actually use the answers, because asking and ignoring is worse than not asking at all. Many email platforms also offer predictions, like expected next order date. Use them, but sense-check them against real behaviour. ### Preference centre + mistakes Offer a preference centre too. It lets subscribers choose topics and frequency instead of unsubscribing completely: weekly digest only, offers only, or pause for a month, which is useful around exams, Ramadan or busy seasons. That keeps more subscribers and improves relevance. Common mistakes: blasting every campaign to the whole list, relying on opens as your main engagement signal, collecting data you never use, and broken personalisation like hi first name in capital letters. Always test merge tags and fallbacks before sending. One broken merge tag in a greeting can undo months of trust. ### Measure success How do you measure success? Compare revenue or clicks per recipient by segment with your old whole-list sends. Track the share of your list that's engaged, meaning clicked or bought in the last ninety days, and watch it trend up. And watch unsubscribe and complaint rates on promotional sends fall as relevance improves. If a segment's performance is no better than the whole list, it's either too broad or the message isn't really different. A segment is only useful if it changes what you send. ### Recap and try this now Let's recap. Segmentation divides the list; personalisation adapts content to individuals, and both improve relevance and deliverability. Segment on declared, behavioural, transactional, lifecycle and location data that your privacy notice covers. Use clicks and purchases, not opens. Score customers with RFM, personalise helpfully, never infer sensitive traits, and offer a preference centre. Here's your try this now. Write five segment definitions for a business you know using the plain-language templates in the lesson text, and for each one, one email idea you'd send only to that group. ## Key takeaways - Segmentation divides the list; personalisation adapts content to individuals; both improve relevance and deliverability. - Segment using declared, behavioural, transactional, lifecycle and location data your privacy notice covers. - Use clicks, visits and purchases over opens because Apple Mail Privacy Protection inflates opens. - Offer a preference centre and keep personalisation helpful, not invasive. ## Try it Define five segments for a business's list, including the data source and one tailored email idea for each. - [Previous: Building a list with consent](https://optimizeall.com/learn/email-marketing-and-automation/consent-based-list-building) - [Next: Authentication: SPF, DKIM and DMARC](https://optimizeall.com/learn/email-marketing-and-automation/authentication-spf-dkim-dmarc) - [All lessons of Email Marketing & Automation](https://optimizeall.com/learn/email-marketing-and-automation)