Retention is more than sending campaigns
An ecommerce retention marketing agency should help customers find a reason to return—not simply increase the volume of messages they receive. Its work connects customer data, merchandising, creative and channel operations to improve the experience after someone subscribes or buys.
The distinction matters when comparing proposals. A campaign-production service can be useful, but it is not the same as lifecycle strategy. Equally, an ambitious strategy is of limited value if nobody owns implementation, quality assurance and ongoing maintenance.
Start with the commercial problem: customers never make another purchase, replenishment reminders arrive too late, discounts dominate sales, or subscription cancellations are rising. The agency should explain which problems it can influence and which depend on product, fulfilment or customer service.
What the work should include
Customer and data diagnosis
Before proposing more activity, the agency should review how customer information moves between your store, messaging platforms and other systems. Expect an assessment of:
- Purchase history, customer identities and duplicate records.
- Marketing permissions, suppression lists and unsubscribe handling.
- Product, inventory and order events used to trigger messages.
- Existing segments, automated journeys and campaign performance.
- Returns, cancellations and discounts that affect revenue reporting.
The output should be a prioritised action plan, with dependencies and named owners. Ask whether fixing integrations is included or whether your team must supply development support.
Lifecycle strategy and automation
Lifecycle work maps messages to customer needs and behaviour. Depending on the business, this can include welcome journeys, browse or cart abandonment, post-purchase education, replenishment, cross-selling, win-back and subscription support.
Not every journey belongs in every store. A durable-goods retailer should not inherit a replenishment programme designed for consumables. Someone waiting for a delayed order should not receive an inappropriate sales push.
For each proposed journey, expect the agency to define its audience, trigger, exclusions, content, exit conditions and success measure. Confirm who updates it when products, policies or integrations change.
Campaigns, creative and channel operations
Campaign delivery should cover more than a calendar. Specify who handles merchandising briefs, copy, design, production, approvals, link checks and testing across devices.
Channel operations should include deliverability monitoring, list hygiene, frequency controls and coordination between automated and scheduled messages. SMS may be relevant, but it should not appear in a proposal merely because the agency sells it.
Consent and compliance responsibilities need explicit ownership. The FTC’s CAN-SPAM compliance guide explains US commercial-email obligations and warns that businesses cannot contract away their legal responsibility by hiring someone else to handle email marketing. Other markets and channels require their own assessment.
Use the email and retention agency directory to identify relevant providers, then examine whether their actual scope matches these responsibilities.
How retention work should be measured
Start with customer outcomes
A useful scorecard separates customer behaviour, financial outcomes and channel health. Ask the agency to define each metric before agreeing to targets.
- **Repeat purchase rate:** Which customers are eligible, and over what observation period?
- **Time to next purchase:** Does the analysis reflect the product’s normal buying cycle?
- **Cohort revenue:** How do customers acquired in comparable periods behave as they mature?
- **Contribution after marketing costs:** How are discounts, returns, product costs, platform charges and agency fees treated?
- **Subscription retention:** Are cancellations, pauses and failed payments separated?
- **Channel health:** Are complaints, unsubscribes and delivery problems visible alongside sales?
Customer lifetime value needs particular care. Ask whether the figure is historical or predicted, whether it measures revenue or contribution, and how incomplete customer histories are handled.
Do not confuse revenue with retention
A historical public-company example illustrates the distinction. Amazon’s annual filing for 2024 reports subscription-services net sales of:
- **2022:** $35,218 million.
- **2023:** $40,209 million.
- **2024:** $44,374 million.
The filing describes this category as including Prime membership fees and other subscription services. These are revenue figures, not renewal rates or evidence of an agency’s effectiveness. They cannot tell a buyer whether growth came from retaining customers, acquiring subscribers, pricing or a changing service mix.
Apply the same discipline to proposals: rising channel revenue alone does not establish improved retention. These historical figures illustrate a measurement distinction, not a benchmark for your store.
Separate attribution from incrementality
Messaging platforms credit purchases according to their attribution rules. That is useful operational information, but it does not establish that the message caused the purchase. Customers may already have intended to buy, and other channels may claim the same order.
Ask for documented attribution windows, an explanation of overlapping claims and reconciliation with store-level sales. Where feasible, use randomised holdout groups to compare customers eligible for a programme with comparable customers who do not receive it.
For smaller audiences, a test may not produce a decisive result quickly. A credible agency should explain that limitation rather than present ordinary fluctuations as proof. Opens and clicks can help diagnose messaging, but neither should substitute for business outcomes.
How to compare proposals fairly
Put every proposal against the same brief
Give shortlisted agencies the same business context: product category, buying cycle, customer-data setup, current problems, internal capacity and commercial constraints. Our guide to choosing an ecommerce marketing agency can help structure the broader selection process.
Request an itemised scope covering:
- Discovery, data checks and initial recommendations.
- Journeys being created, repaired or maintained.
- Campaign production and revision allowances.
- Segmentation, testing and reporting responsibilities.
- Integration work, migrations and technical support.
- Client approvals, account access and escalation procedures.
Distinguish deliverables from ambitions. “Improve loyalty” is an objective; a documented post-purchase programme with clear ownership is a deliverable.
Understand the commercial model
Compare fees against work and accountability, not just headline price. Separate setup costs, ongoing management, platform subscriptions, messaging charges and out-of-scope development.
For performance-linked fees, inspect the revenue definition. Ask how the agreement handles refunds, discounts, existing automated sales, seasonality and attribution changes. A fee tied to platform-attributed revenue can reward activity without demonstrating additional profit.
If the proposal crosses into merchandising, acquisition or onsite conversion, clarify whether you need a broader ecommerce marketing agency rather than a narrowly scoped retention partner.
Evidence, ownership and warning signs
Request case studies that explain the starting problem, scope, measurement method and limitations. A revenue screenshot without dates, definitions or context is weak evidence. References are most useful when the client has a similar buying cycle and operating setup.
Before signing, establish that your business controls its accounts, customer data and reusable assets. The agreement should explain documentation, intellectual-property rights and handover arrangements if the relationship ends.
Watch for guaranteed lifts without diagnosis, unexplained attribution settings, constant discounting, missing suppression rules and reluctance to share how results are calculated. Consider conversion optimisation support when checkout friction is the underlying obstacle; messaging cannot fix every store problem.
The strongest proposal makes the work inspectable: what changes, who delivers it, how success is assessed and what happens when the evidence does not support the original plan.