DTC technology × growth systems — built to compound.
DTCUP / FIELD NOTES / OPERATE

Give Customers a Good Reason to Return Before Building a Retention Program

Choose useful retention benefits, subscriptions or repeat-purchase reminders from customer needs, permission, reliable controls and complete relationship economics.

Published May 17, 2026Updated Sep 20, 202613 min read
Jack Hu
Jack Hu

7 years of experience building and operating brand websites, specializing in website development, paid advertising and SEO. Built a brand website from launch to US$10 million in annual sales.

The Short Answer

Identify the natural reason for another purchase and where the first experience fails. Fix product guidance, fulfillment or service problems before adding messages. Organize communication around the customer's current situation and choose benefits that deliver understandable value at a supportable cost.

Define entry, stopping and overlap rules so automation responds to purchases, support issues and changed preferences. Test the actual customer experience. Compare customer groups after similar elapsed periods, examine contribution and service outcomes, and distinguish attributed orders from additional effect. Give one person responsibility for the connections across teams and use the program's learning to improve the product and operating experience. Compare recurring subscriptions with reminders that require a fresh purchase, using actual consumption, control, complete customer paths and comparable economic outcomes.

Key takeaways

  • Different products create different natural reasons and intervals for another purchase.
  • Choose subscription or reminders from consumption and control needs, not a desire to label every customer recurring.
  • Recheck customer context and communication choices before automated actions.
  • Judge repeat behavior alongside contribution, customer outcomes and uncertainty.

A retention program is supposed to improve the relationship after the first purchase. Yet many programs begin with a different question: how many automated messages can we send, and which discount will make someone buy again?

That approach can produce orders while leaving the real problem untouched. The product may be difficult to use. Customers may not understand which refill fits. Delivery may be unreliable. A points balance cannot make those experiences satisfactory.

I would start by asking what a successful ongoing relationship looks like for this particular product. Then build the service, communication and commercial offer around it. Email, loyalty points and subscriptions are possible methods. None is the definition of retention.

Understand the natural reason for another purchase

Different products create different relationships. A regularly used consumable may need replenishment. A durable product may need maintenance, a compatible accessory or an occasional replacement. A gift purchase may lead to another gift months later, rather than personal use next week.

Do not judge all of those customers by the same short repurchase window. Someone who does not buy another sofa within ninety days is not necessarily disappointed. Someone who stops buying a frequently used refill may deserve a different investigation.

Write down the customer's likely next need and the evidence behind it. Use actual purchase intervals, product usage questions, support contacts and conversations. Treat the interval as a distribution, not a timer that applies to everyone.

For an illustrative household-cleaning brand, a starter kit and refill have different roles. The starter kit must work well enough to establish a routine. The refill must be easy to identify and order when it is needed. Sending a refill offer cannot compensate for an unclear dilution instruction in the original box.

The customer-choice guide is useful here: a good second purchase still depends on understanding fit, compatibility and value.

Find where the first experience fails

Map the period between placing the first order and reaching the first useful outcome. Did the package arrive as promised? Were all components present? Could the customer set up or use the product? Did help resolve a problem without repeated explanations?

Read those experiences before choosing a campaign. A cluster of missing-part complaints calls for a fulfillment or packaging correction. A confusing instruction needs clearer guidance. A reminder sent while a replacement is unresolved is likely to feel careless.

Separate the underlying problem from its communication. Better order updates can reduce uncertainty, but they do not make an unreliable delivery process reliable. Both may need work, with different owners and different evidence of improvement.

Choose the most consequential failure that the team can correct. Then verify whether customers reach the intended outcome more easily. The post-purchase experience guide connects these improvements to the wider service journey.

Organize communication around the customer's current situation

A calendar says when the brand wants to speak. A useful program also considers what has happened to the customer. New, delivered, using, ready to replenish, waiting for support and no longer interested are meaningfully different situations.

Use reliable signals to distinguish them. Do not invent a detailed personal profile when the data only says that an order was placed. If delivery confirmation is unavailable, do not write as though the product definitely arrived yesterday.

For the cleaning example, a simple program might work like this:

Customer situation Useful response Reason to stop or change it
Waiting for the starter kit Accurate order and delivery information The order is cancelled or the delivery promise changes
Beginning to use the product Clear setup and usage guidance The person reports a problem requiring support
A plausible refill need approaches Help identifying the compatible refill A refill has already been ordered or communication is declined
A service issue remains open Follow through on the resolution The issue is actually resolved, with the next step confirmed

This is a planning example, not a ready-made sending schedule. Select timing and channels from the product, permission and available evidence. Keep essential order information distinct from optional marketing in the approved setup.

Choose email, SMS or neither for the specific moment

Choose a communication channel after defining the customer's task, the relevant permission and the value of the interruption. A comparison of average channel revenue does not answer whether this customer needs this message now.

Email can carry a longer explanation, illustrations and information someone may want to revisit. A short text message can suit a concise, relevant prompt where the customer has agreed to that use. Neither format guarantees attention, delivery, understanding or an additional purchase.

Customer task A channel choice to evaluate What could make it unsuitable
Understand setup or maintenance An accessible email with a clear guide, where permitted Too much information, inaccessible content or an unresolved product problem
Receive a concise requested reminder SMS where the purpose and channel are permitted and useful Inaccurate timing, unclear sender or unnecessary interruption
Compare products before a repeat purchase A useful page supported by an appropriate email or requested message Squeezing essential limitations into an unclear short message
Resolve a current service issue The suitable support route for that issue Continuing promotional messages while the underlying problem is ignored

These are decision prompts, not an instruction to send every customer both channels. Having an email address and phone number does not by itself establish permission for marketing through either. Verify the applicable market, provider and message-purpose requirements with the responsible team.

For example, Twilio's messaging policy specifies consent, sender identification and opt-out requirements for its services. That is a provider policy, not a complete global legal checklist or a substitute for checking the actual sending arrangement. Keep permission records and ensure a changed preference affects future actions.

Compare costs on the actual use case. Include the available audience, message charges, relevant production and support work, discounts and the contribution from additional orders. Account for the provider's real billing units and destination pricing rather than assuming every message has one universal cost.

If you want to learn whether SMS adds value to an email program, define the eligible group and compare the existing program with the proposed addition using a suitable experiment where feasible. Counting every SMS-attributed order as additional value overlooks customers who would have bought after email or without either message.

Set priorities across the channels before launch. A recent purchase, open complaint or preference change may make both planned messages irrelevant. Test suppression and delayed-data cases, then review complaints, opt-outs, support effort and contribution together. “Send nothing now” should remain a valid decision.

Choose benefits that support the relationship

Before adding points, calculate what the benefit does for the customer and what it costs the business. A useful benefit might reduce effort, improve access to relevant support or make a regular purchase more convenient. A discount is one option, not the only source of value.

Make benefits understandable before asking someone to join. Explain eligibility, how the benefit is earned and used, and any actual limits. Do not design a program whose apparent value depends on customers misunderstanding its rules.

Compare a subscription with a replenishment reminder before treating recurring billing as the next growth milestone. In this comparison, a subscription creates an agreed recurring order or service arrangement. A reminder invites the customer to decide whether to make another purchase; receiving it should not itself create a paid order. The difference is the commitment being made, not simply the message format.

I would begin with consumption and control. Ask whether the customer can predict the next need, whether the quantity stays reasonably stable and whether automatic replenishment removes useful effort. Check those questions with actual customers and order histories rather than assuming that every consumable should renew monthly.

Customer situation Arrangement worth considering What to establish before choosing
Regular use with a reasonably predictable quantity A clearly explained subscription The schedule fits and the customer can manage changes reliably
Irregular use or long gaps between occasions A requested reminder or easy repeat order The reminder helps without assuming the product has run out
Product choice changes between purchases A reminder with useful selection guidance The next choice remains compatible and easy to understand
Customer values continuity but needs flexibility A subscription with suitable supported controls Skips, quantity changes and other promised actions actually work
No current repeat need No replenishment prompt The program can leave the customer alone rather than force a sale

For the fictional cleaning brand, two households can consume the same refill at different rates. One knows its regular requirement and dislikes remembering to order. The other cleans an occasionally used property and has unpredictable gaps. The first might value an adjustable recurring arrangement; the second might prefer a reminder it can act on or dismiss. Those are hypotheses to investigate, not evidence that either option suits all households.

Make the first choice clear on the product page. Explain what the customer is agreeing to, the price basis, timing and available controls using the actual supported arrangement. A one-time purchase, optional reminder and recurring purchase should not look interchangeable at the moment of confirmation. Confirm applicable requirements with the responsible team before release.

Test each path through its consequence. A reminder should lead to the appropriate product and a fresh purchase decision, using current availability and price. If a customer has already reordered, changes communication preferences or still has an unresolved delivery issue, the sending logic should consider that new information. Test the final action rather than only the date when the reminder was scheduled.

For subscriptions, follow the scheduled renewal into payment, fulfillment and the next account state. Customers need to understand and use the promised skip, change, pause or cancellation options. Check actions close to renewal and distinguish changing a future subscription from changing an order already processed. Making the relationship difficult to leave is not evidence of loyalty.

Compare the economics on the same observation window. Include actual collected revenue, discounts, relevant fulfillment and service costs, refunds and the customer's current state. A reminder click is not a repeat purchase; subscription enrollment is not proof of several successful renewals. Keep scheduled future revenue separate from money actually collected.

Customers who voluntarily choose a subscription may already be more frequent users than customers who prefer reminders. A higher repeat rate for that group does not, by itself, prove that the subscription created the difference. Use a suitable controlled comparison where feasible, and report the limits where selection or small numbers prevent a firm conclusion.

Start with a manageable customer group and a complete experience for the chosen arrangement. Read the reasons people skip, decline or leave alongside the numbers. If customers accumulate unused product, review cadence and quantity before increasing discounts. If reminders arrive too late or too often, repair the timing and controls. Keep both options when they serve genuinely different needs and the team can operate them clearly; do not add both merely to fill a feature checklist.

Start with a benefit the team can fulfill reliably. A complicated tier structure creates ongoing obligations even when the software makes it look easy to launch.

Put an economic boundary around the program

Measure the cost of the whole benefit, not just the software subscription. Discounts, reward redemption, shipping support, service time and additional returns can change the result. Include costs once, using definitions that finance and operations understand.

Consider a hypothetical refill with $16 of contribution before a promotional discount and the program's additional service costs. A $5 discount leaves $11 before those other costs. That arithmetic alone does not tell you whether the discount is worthwhile: the customer might have ordered anyway, ordered earlier, or bought a larger useful quantity.

The comparison should ask whether the program creates enough additional value over an appropriate period to cover its costs. Avoid counting all revenue associated with a message as revenue caused by that message.

Also examine who receives the benefit. Rewarding frequent purchasers may be commercially sensible, but a program that subsidizes purchases already likely to happen can look strong in attributed revenue while adding little contribution.

Use the business economics guide to connect the order calculation to operating costs and cash. A retention forecast should use observed behavior where possible and label uncertain future purchases clearly.

Prevent automation from ignoring new information

A person can qualify for a reminder on Monday and place an order before Thursday's message. The program needs to consider the changed situation at the relevant action, not only when the person first entered the sequence.

For example, Klaviyo distinguishes trigger and profile filters; its flow-filter documentation explains that profile conditions are checked on entry and before actions. This is an implementation example, not a reason to choose a particular vendor. Verify the equivalent behavior in the system you use.

Document entry, stopping conditions, repeat-entry behavior and priority when programs overlap. A welcome sequence, replenishment reminder and promotion can all reach the same person unless someone owns the combined experience.

Test ordinary cases and uncomfortable ones: a recent purchase, a refund, an unresolved complaint, a changed preference, duplicate records and delayed order data. Check the final message and destination, not just whether the workflow appears connected in the editor.

Respect the person's current communication choice. Use only the approved data and channels, keep preference changes effective, and ensure support can identify why an inappropriate message was sent. More automation should reduce avoidable work for customers as well as the team.

Measure relationships at comparable ages

Group customers by a meaningful starting point and compare them after similar amounts of time. This is often called cohort analysis. A group acquired last week has had less opportunity to return than a group acquired six months ago.

Shopify's customer-report documentation describes a first-order cohort view. The same analytical idea can be applied in other systems: define the group, the elapsed period and the outcome before drawing a conclusion.

For a practical review, examine repeat-purchase counts, contribution over the observed period, product returns and relevant service issues. Show group sizes. Separate first-product or acquisition differences when they could explain the result, but avoid splitting a small dataset into meaningless fragments.

Use a suitable comparison when evaluating a change. A properly designed holdout can provide stronger evidence than attributing every subsequent order to the program. When volumes or the situation do not support a convincing test, combine directional numbers with customer and operating evidence and state the uncertainty.

Look for benefits and damage together. More orders accompanied by more complaints, unnecessary shipments or expensive discounts may require revision. A quieter program that helps customers use the product successfully may create a healthier relationship even before a later purchase is visible.

Give one person ownership of the complete experience

Retention spans product, fulfillment, support, marketing and reporting. Give someone responsibility for the connections, even if different people perform the work. Otherwise each team can optimize its own measure while the customer receives a disjointed experience.

Begin with one important lifecycle moment and a limited, well-defined change. Record the problem, expected customer benefit, commercial boundary, owner and evidence required for a decision. Confirm that operations can honor the promise before expanding the audience.

Review the actual experience after launch. Read replies and complaints, inspect unexpected exclusions, check whether people receive conflicting messages and assess outcomes after a relevant period. End the review with a decision to keep, change, stop or investigate further.

A useful retention program becomes a way to learn what customers need next. Feed that learning into product guidance, packaging, inventory and service. The most valuable improvement may be a better first experience rather than another message asking for the next order.

Questions about retention programs

Do we need a points program to improve retention?

No. Start with the reason customers should return and the problems that prevent a good experience. Clear product guidance, dependable service and convenient repeat ordering may address the need more directly. Add points only when their value and ongoing cost are understandable.

How soon should we send a replenishment reminder?

Use the product's actual usage pattern and observed purchase intervals, then test a sensible approach. Customers use products at different rates. Stop or change the reminder when a new order or preference makes it irrelevant, rather than applying one fixed interval to everyone.

Is revenue attributed to retention messages proof that the program worked?

No. Attribution assigns credit under the reporting rules, and some customers may have purchased without the message. Compare customer outcomes and contribution over a relevant period, use credible experimental evidence where possible, and report uncertainty when the additional effect is unclear.

Explore the customer experience and retention guides to connect communication with service, product value and repeat behavior.

Sources

  1. help.klaviyo.com — 115002779051
  2. help.shopify.com — customers-reports
  3. www.twilio.com — messaging-policy

Still have a question?Ask the DTCUP Community ↗

FAQ

Questions related to this insight.

Clear scope, practical constraints and direct answers for the decision ahead.

Do we need a points program to improve retention?

No. Start with the reason customers should return and the problems that prevent a good experience. Clear product guidance, dependable service and convenient repeat ordering may address the need more directly. Add points only when their value and ongoing cost are understandable.

How soon should we send a replenishment reminder?

Use the product's actual usage pattern and observed purchase intervals, then test a sensible approach. Customers use products at different rates. Stop or change the reminder when a new order or preference makes it irrelevant, rather than applying one fixed interval to everyone.

Is revenue attributed to retention messages proof that the program worked?

No. Attribution assigns credit under the reporting rules, and some customers may have purchased without the message. Compare customer outcomes and contribution over a relevant period, use credible experimental evidence where possible, and report uncertainty when the additional effect is unclear.

Your workspace

Selected services

DTCUP / YOUR NEXT PERSPECTIVE

Stay curious.
Stay connected.

Ideas for the work behind growth—websites, brands and the apps that connect them.