How To Segment By Revenue And Purchase History

Revenue and purchase history are among the most useful signals available to a permission-based email programme. They show what someone has bought, how often they buy, how much they have spent and, crucially, whether their relationship with the brand is active or fading.

Used well, these signals help replace broad promotional sends with messages that make sense for the recipient: a replenishment reminder for a consumable product, a complementary-item recommendation after a first order, early access for a long-standing high-value customer, or a carefully limited win-back message for someone whose normal buying cycle has passed.

The important qualification is that purchase data does not explain everything. A customer with a high lifetime spend may have returned their last order, stopped engaging with marketing, or bought a durable product they will not need again for years. Revenue segmentation works best when it is combined with consent status, product context, timing and sensible contact-pressure controls.

Start by deciding what “revenue” means

Before building segments, define the number that will drive them. “Customer revenue” is often treated as a simple field, yet the calculation can vary substantially between systems.

For most retention and lifecycle email decisions, net merchandise revenue is the most useful starting point: paid product revenue after cancellations, returns and refunds. Whether shipping, tax, gift cards, store credit, discounts and subscription renewals are included should be an explicit business rule, not an accidental consequence of an integration.

For example, a fashion retailer that ranks customers by gross checkout value may label a frequent returner as a VIP. A business that sells annual subscriptions may need to distinguish initial acquisition revenue from successful renewals. A marketplace may need to use commission revenue rather than the full basket amount. None of these choices is universally correct; they should reflect the commercial decision the segment is intended to support.

Document the definition alongside the segment. A useful record might say: “Net paid product revenue, excluding VAT, shipping, cancelled orders and refunded items; refreshed daily; attributed to the customer profile used at checkout.” This makes the segment interpretable when someone later asks why a particular contact was included.

Check the data before trusting the segment

A revenue segment is only as reliable as the events behind it. Reconcile a small sample of customer profiles against the ecommerce platform or CRM. Check that an order is not recorded twice through separate integrations, that refunds reduce the appropriate total, that test orders are excluded and that guest checkouts can be connected to an existing subscriber where the customer has identified themselves.

Also establish the event timing. Some platforms create an order at checkout; others distinguish order placed, paid, fulfilled and refunded. A post-purchase cross-sell should generally rely on a completed, paid order, while a delivery-related message may depend on fulfilment status. Do not use a segment based on an event whose meaning the team has not agreed.

If sales occur in multiple currencies, convert values consistently for reporting and thresholds, while retaining the original order currency for offers and content. It is confusing to place someone into a “spent over £100” segment on the basis of a foreign-currency conversion and then show them a poorly localised offer.

Use purchase history as a pattern, rather than a single score

Total spend matters, but it is incomplete on its own. A customer who spent £500 yesterday is in a very different position from one who spent £500 across several orders four years ago. The practical answer is to consider three dimensions together:

• Recency: how long it has been since the most recent qualifying purchase.
• Frequency: how many qualifying purchases the customer has made in a defined period.
• Monetary value: the amount they have spent, usually within that same period or over their lifetime.

This is commonly called RFM analysis. It is a useful framework because it separates recent customers, loyal repeat buyers and historically valuable but inactive customers. RFM tools typically derive groups from recency, frequency and monetary value, but the labels and thresholds are not universal. They should fit the brand’s actual purchase cadence and distribution of customer value. Klaviyo’s explanation of RFM analysis is a helpful illustration of the inputs and the need to adjust scoring thresholds to the data.

For a low-cost consumable product, a customer who has not bought for 60 days might be overdue. For furniture, 60 days may simply mean the customer has finished furnishing a room. Start with product economics and observed reorder behaviour, rather than importing arbitrary 30-, 60- or 90-day windows.

A practical segment structure

A small number of clearly different segments is normally more useful than a complicated matrix that nobody can explain or maintain. The following structure gives many retailers and subscription businesses a workable foundation.

First-time customers

Definition: exactly one qualifying purchase, made recently.

Purpose: turn a first purchase into a second one. The content should help the customer get value from what they bought, set expectations for delivery or usage, request a review at an appropriate point, and introduce genuinely relevant complementary products.

Important exclusion: do not send a generic “welcome” promotion that conflicts with the post-purchase flow. The customer has already converted; the next message should acknowledge that fact.

Recent repeat customers

Definition: two or more qualifying purchases within a period that reflects normal buying behaviour, with a recent last order.

Purpose: nurture habit and increase relevance. This group may be suitable for replenishment prompts, product education, loyalty benefits, category-specific updates or early access. A repeat buyer is often more valuable than a high-spending one-off purchaser, particularly where margins and future demand matter.

Use product history to prevent clumsy recommendations. Someone who has just purchased a three-month supply should not immediately receive a weekly reminder to buy the same product again.

High-value active customers

Definition: customers above a chosen revenue threshold or percentile, whose last purchase is still within their expected active window.

Purpose: recognise the relationship without assuming that every high spender wants more messages. Benefits can include relevant previews, access to specialist support, carefully chosen premium products or an invitation to manage preferences.

“VIP” should describe an actual experience, not merely a hidden segment name. If the business cannot offer a meaningful benefit, it may be better to use the group for relevance and send prioritisation than to announce a tier it cannot sustain.

High order value, low frequency customers

Definition: above-average spend per order, but only one purchase or relatively few purchases.

Purpose: understand whether the customer is a one-off occasion buyer, a durable-goods buyer, a trade customer, or someone who needs a different next step. This group often responds better to service, compatibility guidance and considered recommendations than to repeated discounting.

It is also a useful diagnostic segment. If these customers never return, investigate delivery experience, product fit, pricing or the absence of a credible second purchase, rather than simply increasing promotional frequency.

At-risk repeat customers

Definition: previous repeat buyers whose time since last purchase is longer than their own or the cohort’s expected reorder cycle.

Purpose: offer a reason to return. The message might be a reminder based on a consumable product, a new range in a previously purchased category, or a straightforward request to update preferences.

There is no need to treat every lapse as a discount opportunity. A discount can train customers to wait, reduce margin and mask a product or service problem. Test non-price messages against offers and measure incremental orders, not just clicks.

Lapsed customers

Definition: no qualifying purchase for a materially longer period than the normal cycle.

Purpose: run a limited reactivation effort, then reduce or stop promotional mail if there is no meaningful response. It is usually better to send one or two well-considered messages than to retain a permanently unresponsive group in every campaign audience.

Mailbox providers assess whether recipients want the mail they receive. Google advises senders to send only to people who want messages, monitor spam reports and consider removing recipients who do not open or read mail; its bulk-sender requirements also include accessible unsubscribe mechanisms for promotional messages. See Google’s sender guidelines and its sender-guidelines FAQ. Purchase history can identify former customers, but it does not override poor engagement or an unsubscribe request.

Add product and timing context

Revenue-based segments become much stronger when they answer a specific merchandising question. Rather than emailing everyone who has spent more than £200, identify the business reason for reaching them.

• A skincare brand might target customers who bought a 30-day product 25 to 35 days ago, have not repurchased it and have not made a recent complaint or return.
• A homeware retailer might target customers who bought a bed frame but not compatible bedside tables, excluding anyone whose frame order is still awaiting delivery.
• A subscription business might distinguish customers approaching renewal from those who have already cancelled, failed payment or requested no marketing communications.
• A retailer with seasonal ranges might use the previous year’s category purchase, but only where the item is genuinely seasonal and the recipient remains engaged.

Many email platforms support conditions for total order count, average order value, total spend, product, category and purchase date. For example, Mailchimp’s purchase-activity segmentation documentation describes these conditions and the ability to combine them with engagement criteria. The feature names will differ by platform, but the underlying logic is portable.

Be careful with “has bought product X” as a permanent rule. The relevant period may be a week, a year or never, depending on the item. A person who bought baby clothes several years ago may not welcome an assumption that the same purchase intent remains current.

Combine value with permission and engagement

Every revenue or purchase-history segment should be intersected with operational eligibility rules. At a minimum, exclude contacts who are unsubscribed, suppressed, hard bounced, marked as spam, in a complaint state, or otherwise ineligible under the organisation’s consent and preference rules.

Then consider recent meaningful engagement. Opens can be incomplete and affected by privacy features, so they should not be treated as proof that a person read an email. Clicks, site activity, app activity, purchases, preference updates and replies may provide stronger signals where they are available and appropriate to use. The aim is not to create an inflexible “opened or nothing” rule; it is to avoid repeatedly mailing people who show no evidence of wanting promotional email.

High-value customers should not be exempt from this discipline. A large historical spender who has ignored marketing for a long time may be better approached through a restrained re-permission or preference message than through every sale announcement. This protects customer trust and avoids concentrating unwanted mail on a commercially important group.

Set thresholds from your own distribution

Absolute thresholds such as “customers who spent more than £500” are easy to understand, but they may be too broad for one business and too narrow for another. Review the distribution of net revenue, order count, average order value and days since purchase. Percentile-based thresholds can be useful when value is highly skewed: for example, the top 10% of active customers by 12-month net revenue.

Use a minimum audience size where needed for privacy, meaningful reporting and reliable testing. Very small segments may be commercially important, but they can create awkward personalisation, reveal too much about what the business infers, or produce results that are too volatile to interpret.

Revisit thresholds as the catalogue, price points and customer mix change. A threshold set before a price increase or an international expansion can quietly become misleading.

Measure whether the segment improves decisions

Do not judge segmentation only by campaign revenue. High-value segments will naturally generate more attributed revenue than low-value groups, even if the email had little effect. Compare similar eligible customers where possible: a holdout group, a different creative treatment, or a change in timing. Look at repeat purchase rate, net revenue after refunds, margin where available, unsubscribe rate, complaints and longer-term customer behaviour.

Measure the impact on the whole programme too. A targeted campaign may have a good conversion rate yet still be harmful if it adds frequency to recipients who are already receiving several messages from overlapping automations. Establish precedence rules: for instance, a replenishment flow pauses when a customer enters a high-priority service flow, and promotional campaigns exclude people who received a campaign in the last few days.

If a programme has developed broad inactivity, rising complaints or inconsistent infrastructure, better segments alone may not resolve delivery problems. An independent deliverability review can help establish whether authentication, audience quality, suppression, contact pressure and mailbox-provider signals are supporting the intended strategy. That is where Email Foundry’s audience verification, contact-pressure controls and managed recovery work can be relevant: they address the sending conditions around segmentation, rather than treating segmentation as a shortcut around recipient choice.

Privacy and transparency in the UK

Purchase history linked to an identifiable customer is personal data. If it is used to decide which direct marketing a person receives, the organisation is profiling for direct marketing. The ICO’s direct marketing guidance says organisations should make such profiling fair and transparent, have an appropriate lawful basis, keep the data accurate and not excessive, and respect objections to direct marketing and related profiling.

Explain the practice in privacy information in plain language. A statement such as “we use your purchase history to send offers and product information that may be relevant to you” is more useful than hiding the activity behind a vague reference to personalisation. Take particular care with any information that could reveal or infer special category data, and seek specialist privacy advice where the use case is sensitive or uncertain.

Segmentation should make email more proportionate. It should never be used to ignore opt-outs, conceal profiling, extend retention indefinitely or pressure people based on sensitive circumstances.

Build slowly, then automate what has proved useful

Begin with a small set of saved, dynamically updating segments and clear rules for inclusion and exclusion. Validate the customer counts and a sample of profiles. Send a restrained test, review commercial and deliverability outcomes, then refine the definition.

The strongest revenue segments are not the most intricate. They are the ones where the team can answer four questions without hesitation: who is included, why they are receiving this message, what data supports that decision and when they should stop receiving it.

Frequently asked questions

Should revenue segments use lifetime spend or recent spend?

Use the measure that fits the decision. Lifetime spend is useful for recognising long-term value, while 6- or 12-month net revenue is usually more relevant for current promotional eligibility. Many programmes use both: recent spend to identify active value and lifetime spend as supporting context.

How often should revenue and purchase-history segments update?

For campaign audiences, daily updates are often sufficient. Time-sensitive post-purchase, replenishment or suppression decisions may need event-based or more frequent updates. Confirm how your platform handles late refunds, order edits and duplicate events before relying on real-time logic.

Is RFM better than simple purchase-history segments?

RFM is useful when the customer base has enough order history to distinguish recency, frequency and spend. Simple segments can be clearer and more effective for a focused use case, such as first-time buyers, customers due to replenish a consumable, or customers who have not bought in an expected time window.

Should high-value customers receive more email?

Usually no. They may merit more relevant treatment or priority access, but higher value does not equal unlimited attention. Apply the same consent, engagement, preference and contact-pressure rules that protect the rest of the audience.

Sources and further reading

• Google’s email sender guidelines — Google
• Google’s email sender guidelines FAQ — Google
• ICO direct marketing guidance — Information Commissioner’s Office
• Segment an audience by purchase activity — Mailchimp
• Getting started with recency, frequency and monetary analysis — Klaviyo