Ecommerce Personalisation: How to Get Started

Start ecommerce personalisation by choosing one customer problem, using data your store already holds and testing one change against the normal experience. A repeat buyer who struggles to reorder needs a different change from a first-time visitor who can’t find delivery information. Buying software comes after you’ve named that problem.

You’ll need access to order records and permission to change the relevant storefront or marketing tool. Assign someone to maintain the rule and someone to review customer-data use. Keep the normal experience available for customers whose details are missing or who decline optional personalisation.

What Personalisation Can Change on a Store

Personalisation changes what customers see or receive according to their circumstances, preferences or shopping history. A store can tailor banners, product order, emails and chat messages. Product recommendations are another common starting point. The guide to product recommendations covers their types and placements.

Segment-based personalisation gives the same change to a group, such as repeat buyers. One-to-one personalisation uses an individual’s details, such as their stated size or last order. Customisation gives the customer a direct choice, such as selecting their preferred department. Keep those choices editable.

Use this sequence for your first rollout.

1. Choose one problem and an observable outcome. For a replenishment store, that might be helping repeat customers complete a reorder.
2. Define who receives the change. Write the rule in plain language, such as customers with a previous completed order in the same category.
3. Choose one surface. Add a reorder shortcut before changing the whole homepage.
4. Confirm the data use is permitted. Record which fields the rule needs and how long you keep them.
5. Preview the change with matching and non-matching customer records. Each must receive the intended experience, with a usable default when data is absent.
6. Compare the outcome with customers who receive the normal experience. Measure completed reorders and margin, alongside complaints or unsubscribes.

Keep the test groups comparable and run the test long enough to cover your store’s usual buying cycle. A higher click rate alone doesn’t show that customers bought more or that the change paid for itself.

Starting With the Data You Already Hold

Location, traffic source, basket contents and past orders can support a first personalisation rule without a new customer questionnaire. Each signal has limits, so choose a change that remains useful when the signal is wrong.

Existing signal	First change to consider	Limit to account for
Delivery region	Show the relevant delivery explanation	A browsing location can differ from the delivery address
Campaign or traffic source	Keep the landing-page message consistent with the advert	The campaign identifies the visit, not the customer’s lasting preferences
Basket contents	Explain delivery or an applicable bundle condition	A basket can contain gifts for someone else
Completed orders	Provide a shortcut to a previously bought item	Exclude cancelled orders and account for returns
Signed-in customer details	Apply the customer’s saved preferences or loyalty tier	Use the authenticated account and honour changed preferences

A customer’s answer to a preference question is more specific than an inference from a single page view. Ask for the detail that changes the experience, explain its use and make answering optional. The zero-party data guide covers preferences that customers deliberately share.

For example, a hypothetical clothing store can ask whether the customer wants petite sizing. It can then filter the relevant range without guessing from photographs. Consent to save that preference must be separate from any unrelated marketing choice.

Customer Segmentation and RFM Analysis

Customer segmentation groups customers with shared characteristics so your store can give each group a relevant experience. Begin with a split you can explain, such as first-time and repeat buyers. A fuller customer segmentation model can use order history and purchasing patterns.

Segmentation type	Store example	Data needed
Demographic	Adults who voluntarily select an age band for age-specific product guidance	An optional, relevant profile answer
Geographic	Customers in a regional delivery zone	A delivery region or postcode
Behavioural	Customers who buy only when a discount applies	Order and discount history
Value-based	Customers whose completed purchases put them in a loyalty tier	Net order value under a written tier rule

Recency, frequency and monetary value (RFM) analysis scores customers by how recently they bought, how often they buy and how much they spend. It uses order data to distinguish a frequent customer who has stopped buying from someone who placed one small order.

A spreadsheet can handle a first RFM analysis. Use a consistent observation period and apply this process.

1. Export completed orders with a stable customer identifier, order date and net order value. Remove duplicates and use one currency.
2. Group the orders by customer. Calculate days since the most recent purchase, the number of completed orders and total net spend for your chosen period.
3. Set score bands for each measure. A simple model uses scores from 1 to 5, with 5 meaning more recent purchases, greater frequency or higher spend.
4. Assign each customer separate R, F and M scores. Keep the underlying values beside the scores so staff can explain the result.
5. Define segments from the score combinations. Decide what action each segment receives and refresh the calculations on a regular schedule.

In a hypothetical store, a customer last bought 150 days ago after making eight orders during the previous year. Their net spend falls in the store’s highest spending band. If those values produce R1, F5 and M5, the customer fits a lapsed high-spender segment. A useful action is a relevant win-back message, subject to their marketing permissions.

The thresholds must reflect your product’s purchase cycle. A furniture buyer who hasn’t returned for several months can be behaving normally. The same gap for a weekly consumable needs a different interpretation. Returns reduce net spend, and a guest checkout must match an existing customer reliably before you combine the records.

You can build these three customer segmentation examples this month.

• Lapsed high spenders have high historical net spend but no order beyond the normal repeat-purchase window. Show a return-to-store message relevant to their previous category.
• Discount-only buyers have completed orders with discounts and no full-price orders in the chosen period. Use that pattern to test an offer’s margin before broadening it.
• First-time buyers from one category have exactly one completed order in that category. Give them category-specific care information or reorder help when appropriate.

A segment can also determine a loyalty tier under your store’s written membership rules. Pass the tier and first name in signed shopper details to an agent that accepts authenticated personalisation. The agent can greet the member by first name and explain the benefits for their tier. The store must supply the tier accurately and keep benefit rules current.

AI customer segmentation finds patterns across browsing and orders that would take staff more rules to describe. It can group customers with similar category interests or estimate who is likely to stop buying. For a small store, the gain is less manual rule maintenance when there’s enough useful history. With sparse orders, simple rules are easier to inspect and correct.

Review the customers in each group before using it for offers. A pattern describes recorded behaviour, so don’t treat it as proof of someone’s identity, finances or needs.

Privacy Rules That Shape Personalisation

For stores covered by the Privacy Act, the Australian Privacy Principles (APPs) govern how personal information is collected, used and shared. Some small businesses are also covered through exceptions to the small-business exemption. Establish your store’s coverage before treating size as an exemption.

The OAIC’s Australian Privacy Principles connect directly to personalisation decisions. APP 1 requires transparent privacy management and a privacy policy. APPs 3 and 5 address necessary collection and collection notices. APP 6 limits use and disclosure beyond the collection purpose unless an exception applies.

For tailored marketing, APP 7 sets conditions on using personal information and requires a simple opt-out. Sensitive information needs consent for direct marketing. Email and SMS marketing also have separate Spam Act requirements, so a purchase record alone isn’t permission to send promotional messages.

APP 8 addresses overseas disclosure, which matters when a vendor receives customer records offshore. APP 11 requires reasonable security and destruction or de-identification when information is no longer needed, subject to retention exceptions. Access and correction rights also affect stored preferences and profiles. Send vendors only the fields the chosen experience needs.

Privacy reform continues. The government passed a first reform package in 2024 and released further exposure draft legislation in 2026. The consultation has closed, and the Attorney-General’s consultation page says the draft remains subject to government consideration. Proposed stronger consent standards and a fair-and-reasonable test are proposals, not a substitute for existing APP duties.

The Australian Privacy Act guide covers the broader obligations. Before adding a new use of customer data, update the collection explanation and permissions that support that use.

Choosing Ecommerce Personalization Software

Choose personalisation software by the experience it must change and the data needed to make that change. A segment builder and a storefront personalisation tool solve different parts of the task.

Platform type	Fits this first job	What to establish before buying
Built-in commerce features	Group customers and apply store-native changes	Whether your edition can display or act on each segment
Email and messaging tools	Send category-specific follow-ups and tailored campaigns	Marketing permissions, active-profile charges and sending limits
Dedicated personalisation platforms	Change content or merchandising across the storefront	Event collection, integration work and module costs
Customer data platforms	Combine customer records from several systems	Identity matching, destinations and ongoing data ownership

Shopify has built-in customer segments using filters and ShopifyQL queries. Customers enter or leave a segment as they meet its conditions. Start there when your first rule uses data already in Shopify. Creating a segment still needs a connected feature or app that delivers the intended change.

Klaviyo builds segments from customer properties and recorded events, making it useful for personalised messaging. Its RFM report needs a Marketing Analytics or Advanced KDP subscription. Klaviyo also specifies minimum order-history and customer-data requirements, so its automated RFM report isn’t a starting point for every new store.

Magento 2 needs an edition distinction. Adobe Commerce has dynamic customer segments based on attributes such as order history and basket contents. This is an Adobe Commerce feature, not the equivalent native feature in Magento Open Source. Budget for the commercial edition or a separately supported extension when that distinction changes your choice.

Use a separate segmentation tool when your rules need data that the existing tool can’t combine or actions it can’t send elsewhere. A guide to customer data platforms becomes relevant when online orders, shop purchases and other systems must describe the same customer. Adding another database won’t repair inconsistent identifiers by itself.

Ask each supplier these questions before comparing proposals.

• Which customer fields and browsing events does the platform need, and which are optional?
• Which storefront changes can staff publish without development work?
• What integration tasks determine setup time, and who owns each task?
• How do segment updates, opt-outs and deletions reach connected tools?
• Is support available during Australian business hours, and does implementation support cost extra?
• What changes the bill, including traffic growth, profile counts, modules or contract renewals?

Shopify publishes subscription plans. Klaviyo’s pricing depends on active profiles and messaging capacity, with separate products charged separately. Adobe’s Commerce pricing page routes package pricing through sales. Compare the full required package, including implementation and ongoing maintenance.

Nosto’s pricing page describes a base platform fee plus a fixed fee based on store volume, including gross merchandise value (GMV) and traffic. Selected modules and support needs also affect the package. Its main pricing page provides a tailored proposal rather than a standard entry amount. Budget from the published model and a written quote for your store, without estimating a rate from another retailer’s bill.

AI Personalisation and Conversational Tools

AI personalisation uses observed behaviour to choose or adapt content, while conversational tools can use a customer’s stated needs to tailor an answer. A returning visitor’s category interest might change the content they see. In a conversation, a customer can directly correct a mistaken preference.

A shopping assistant needs the store’s product data and policy rules as well as the customer details that the store permits it to use. Signed shopper details connect a name or loyalty tier to an authenticated customer. They don’t make the underlying personalisation lawful by themselves.

shoppa is an AI shopping agent that helps customers find products and compare their options using your catalogue and store documents. Customers can also buy in the chat. Its signed personalisation lets it greet customers by name and loyalty tier. You control the agent’s knowledge and settings in your workspace, so the tier explanation can follow your store’s membership rules.

Keep an anonymous path useful too. A customer can tell an assistant which size or delivery region they need without a permanent customer profile. Use the information for that request, and explain any separate choice to save it.

Digital Clienteling Through Conversation

Digital clienteling is the online version of a store associate who remembers a regular customer and helps them choose. Chat, messaging or a shopping assistant carries that relationship across visits. A useful conversation starts with relevant context and lets the customer change it.

For a clothing store, that context can include past orders, a size the customer stated and their saved preferences. Pass only the fields needed for the interaction, under a clear collection notice and permitted purpose. Obtain consent when required for the sharing or use, and make optional saved preferences easy to remove.

A hypothetical customer asks for another work shirt in the size they saved. The assistant can use that stated size and a previous product reference to narrow the catalogue. It must still show the current item’s size information, because a previous purchase doesn’t prove that every new style fits.

Keep internal segment labels out of customer-facing greetings. A customer can receive relevant help without being told they are a lapsed high spender. Give staff a way to correct the context when an order or preference is wrong.

Personalisation for B2B Stores

Business-to-business (B2B) personalisation gives trade buyers the catalogue, prices and reorder options allowed for their company account. Start with account permissions before tailoring messages. A buyer’s company entitlement can determine the transaction itself.

Customer-specific catalogues limit the products a trade account can buy. Price lists apply the company’s agreed prices, while reorder shortcuts restore frequently purchased items and quantities. If an enterprise resource planning (ERP) system owns those terms, sync its approved records and decide how stale prices are handled.

For a hypothetical trade supplier, an authenticated buyer sees their company’s approved range and can reorder a previous supply list. An employee from another company must receive that company’s own range and prices. Personalisation must never let a message or customer-supplied tier override account permissions.

Before releasing your first change, test a matching account, a non-matching account and an anonymous visitor. The matching account must get the intended experience, and the others must retain a usable default without private details. If the wrong customer receives an offer, inspect identity matching and the segment rule before changing the copy. Expand the rollout only after the rule behaves correctly and the measured outcome supports it.