E-commerce personalisation

E-commerce personalisation is the practice of serving different content, products, or offers to different customers based on what you know about them. Done well, it makes a customer feel understood. Done poorly, or done without the right data, it makes a customer feel surveilled without the benefit of relevance.

The difference between the two is not the technology. It is the data underneath it, and the judgment about when and how to use it.

What good e-commerce personalisation looks like

Good personalisation is built from a customer's actual history with your brand, not from inferences drawn from browsing session alone.

A returning customer who has bought your skincare range three times should see product recommendations in that category when they return, not a homepage designed for a first-time visitor. A customer who regularly buys in a specific price band should not receive promotions exclusively featuring your premium range. A customer who bought a gift last December should receive a timely prompt the following October, not a generic campaign.

These are simple examples. The underlying principle is that personalisation should reflect the relationship: what the customer has bought, what they browse, when they typically purchase, what they have responded to before. First-name personalisation in an email subject line is not personalisation in any meaningful sense. It is a mail merge.

Examples of e-commerce personalisation that works:

• Abandoned cart emails that reference the specific item left behind, sent within two hours, with social proof or urgency relevant to that product category
• Post-purchase flows that recommend complementary products based on what was just bought, not generic bestsellers
• Replenishment reminders timed to the customer's typical repurchase interval, based on their purchase history
• Homepage personalisation that surfaces the product category a returning customer buys from most
• Segment-specific promotional emails that offer relevant products rather than the full range at the same discount

The data you need to personalise properly

Personalisation in e-commerce requires two types of data: transactional data and behavioural data.

Transactional data is what your customers have bought: product, category, price point, date, frequency. This is stored in your order management system and is typically available in Shopify, WooCommerce, or whatever platform you trade on. It is the most reliable signal for personalisation because it reflects actual purchase decisions, not intent signals that may not convert.

Behavioural data is what your customers do between purchases: which pages they visit, which products they view, which emails they open and click. Klaviyo, in a Shopify context, captures this through its tracking pixel and stores it against customer profiles. It is useful for identifying interest signals in real time (an abandoned cart, a repeated product view, a category browse) but should be layered on top of transactional history rather than used in isolation.

The third layer that most businesses do not use: demographic and household-level data that enriches your customer profiles beyond what they have done on your site. This is where address-level data and consumer classification data, such as Experian Mosaic, allows you to personalise by household context as well as purchase history.

Personalisation in Klaviyo and Shopify

Klaviyo personalisation. Klaviyo's strength is lifecycle email flows built from customer behaviour and purchase history. The most impactful flows for e-commerce personalisation are:

• Welcome series that reference the customer's first purchase category
• Post-purchase sequences that vary by product type or purchase value
• Win-back flows that address lapsed customers differently based on how long they have been inactive and what they typically buy
• Browse abandonment and cart abandonment flows triggered by specific product or category engagement

Klaviyo's segmentation allows you to split sends by RFMRFMRecency, frequency, monetary value: a scoring approach that ranks customers by how recently they bought, how often, and how much they spend. A practical bridge from simple rule-based segments to scored segmentation.View in glossary segment, purchase category, predicted LTVCustomer lifetime value (LTV)The total gross profit you expect from a customer across the whole relationship, not a single order. It tells you how much you can afford to spend to acquire and keep them.View in glossary, or any custom property you have attached to customer profiles. This is where the quality of your segmentation data determines the quality of the personalisation.

Shopify personalisation. Shopify's native personalisation capabilities are more limited than Klaviyo's but have improved. The most practical applications:

• Personalised product recommendations using Shopify's recommendation API, trained on purchase and view history
• Collection-level personalisation that surfaces relevant products to returning customers based on their category history
• Dynamic discount codes or offers delivered through Klaviyo flows and redeemed on Shopify, personalised by segment

Measuring whether personalisation is working

Personalisation adds complexity to measurement. If you show different customers different content, you need a framework for comparing outcomes across those groups.

The clearest measure is incremental revenue: the difference in purchase rate, average order valueAverage order valueThe average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".View in glossary, or repeat purchase rate between customers who received personalised communications and those who received generic ones. This requires either holdout testing (a control group that receives the non-personalised version) or a period of A/B testing before full rollout.

A simpler starting point: track click-through rate and conversion rateConversion rateThe share of people who take a defined next step, measured between two funnel stages (e.g. visit → signup). Always read it alongside the volume it is calculated from.View in glossary on personalised email flows versus broadcast campaigns to the same segment. If a personalised post-purchase flow drives a higher repeat purchase rate than a generic newsletter, the personalisation is working. If it does not, the personalisation logic needs review, which usually means the underlying segmentation is not precise enough.