How To Build A Win-Back Automation Without Harming Deliverability

A win-back automation is a controlled attempt to reconnect with subscribers or customers whose relationship with your brand has gone quiet. It is not a licence to send a last-minute discount to every old address in the database.

Done well, it has two equally valuable outcomes: some people return and re-engage; others make it clear, through non-response, an unsubscribe or a complaint, that they should no longer receive routine marketing. Both outcomes can improve the quality of the active audience.

That matters because mailbox providers assess more than technical setup. They see how recipients react to mail. Google advises senders to confirm that people still want messages periodically and to consider unsubscribing recipients who do not open or read them. Its current sender guidelines also require bulk senders to keep reported spam rates below 0.3% and make marketing mail easy to leave. Google’s email sender guidelines are therefore useful operational reading, not just a technical compliance document. (Google’s sender guidance)

The practical goal is simple: give a previously permissioned person a small number of relevant, clearly branded opportunities to return, while protecting people who have opted out, bounced, complained or simply stopped showing signs of interest.

Start by defining what “inactive” means for your business

There is no universal number of days after which a subscriber becomes inactive. A weekly content publisher, an annual insurance provider and a B2B software company each have very different buying and usage cycles. A 90-day gap may be serious for one business and completely normal for another.

Build the definition around the events that genuinely indicate an ongoing relationship. Depending on your model, those may include:

• a purchase, renewal, booking or donation;
• a product login, meaningful feature use or account activity;
• a form submission, support interaction or preference update;
• a click from a marketing email; and
• email opens, used cautiously and alongside stronger signals.

Opens are a weak sole trigger. Image blocking, privacy features, corporate security systems and plain-text reading can all mean that a person reads an email without recording an open. Conversely, an open does not necessarily show real attention. Treat clicks, purchases and product activity as stronger evidence where they are available.

A retailer might define an inactive customer as someone who has not purchased, clicked or visited while signed in for 120 days, despite being eligible for regular campaigns. A subscription app may instead use product inactivity, such as no login for 45 days, because that is more closely connected to retention risk. The point is to choose a definition that reflects a meaningful break in the customer relationship, rather than copying a generic threshold.

Make eligibility stricter than the trigger

The automation trigger identifies people who may need a win-back. Eligibility decides who is safe and appropriate to contact. Keep these separate.

Before the first message can send, exclude recipients who are:

• unsubscribed from the relevant marketing category, or globally suppressed;
• known complainers, hard bounces or addresses with persistent delivery failures;
• currently in a customer-service, dispute, bereavement or fraud workflow where promotional messaging would be inappropriate;
• already receiving another high-priority lifecycle sequence, such as an abandoned basket, onboarding or renewal journey;
• recently sent a campaign, where the win-back would breach the brand’s contact-pressure rules; or
• not covered by a valid permission or other applicable basis for the marketing in their jurisdiction.

For UK marketing to individuals, the rules under PECR normally require consent for email marketing, with a limited existing-customer exception often called the soft opt-in. In either case, the sender must identify itself and provide a valid means to opt out. The ICO also says organisations should not require someone to create or log into an account just to unsubscribe. The ICO’s electronic mail marketing guidance explains the position in more detail. (ICO guidance)

If you market in the United States, the Federal Trade Commission says commercial email must include a working opt-out mechanism and opt-out requests must be honoured within 10 business days. Operationally, it is better to process these immediately or near-immediately: delay creates unnecessary complaints and can also conflict with mailbox-provider expectations. The FTC’s CAN-SPAM compliance guide sets out the US requirements. (Ftc source)

Use a short sequence with clear exit conditions

A win-back journey should usually be finite. Repeatedly pursuing people who have not responded turns a recovery attempt into a source of unwanted mail.

For many programmes, two or three messages over a few weeks is enough. The exact spacing should reflect the normal send frequency and purchase cycle. The important part is the decision logic between sends: each message should be conditional on no meaningful re-engagement and continued eligibility.

These timings are examples, not a prescribed cadence. A brand that normally emails twice a month may want wider gaps; a service with a rapid repeat-purchase cycle may need a shorter window. Avoid adding “just one more” message unless there is a specific, defensible reason to do so.

Define a meaningful exit event

Do not use “opened an email” as the only signal that someone has been won back. An open may be useful as a supporting indicator, but it should not automatically return a recipient to the full marketing cadence.

A stronger approach is to tier the outcomes:

• High-intent return: purchase, renewal, booking, product login or a conversion event. Exit the journey and re-enter the appropriate customer or engagement lifecycle.
• Potential interest: click to useful content, browse activity or a preference update. Exit the win-back, then move the recipient to a reduced-frequency nurture segment until further activity appears.
• Explicit decline: unsubscribe, complaint or category opt-out. Suppress promptly and ensure the instruction reaches every relevant sending system.
• No response: do not quietly put the address back into the normal broadcast audience. Apply a sunset rule or a much more cautious, separately governed re-permission approach where lawful and appropriate.

Give each message a distinct job

The most common weakness in win-back programmes is a sequence of near-identical promotions. If the first offer did not matter, a slightly more urgent version is unlikely to rebuild relevance.

Start with why the recipient originally signed up or became a customer. That might be useful content, replenishment, new product capability, a saved benefit, a change that addresses a previous friction point, or a personalised reminder based on a legitimate and understandable customer relationship.

The first email can be simple: acknowledge the gap, remind the recipient what they can expect and provide one primary action. For example, an education platform could point to newly added courses in the topics the learner chose at registration. A retailer might use recent category interest only where that personalisation is accurate and not unsettlingly specific.

The second email should test a different value proposition. If the first highlighted new products, the next might offer help choosing, useful editorial content or an account benefit. A discount can be appropriate, but it should not be the default answer. Constantly escalating incentives can train customers to wait for a win-back offer and tells you little about whether they value the underlying relationship.

The final email is often best used as a preference decision rather than another sales push. Give people meaningful choices: fewer messages, selected categories, a pause, or complete unsubscribe. A preference centre is useful only if the choices alter future sending in practice.

Build the automation around contact pressure, not in isolation

Every automated programme competes with campaigns and other journeys for the same inbox. Without central contact-pressure controls, a subscriber can receive a win-back message on the same day as an abandoned-basket reminder, a sale announcement and a product update. The individual workflow may be logical; the combined experience is not.

Set an overall marketing frequency cap, then decide which messages can override it. A password reset can override it. A routine win-back email generally should not. If a person is in a sales campaign, you may suppress the win-back temporarily. If they are in a highly relevant, time-bound lifecycle sequence, decide which programme takes precedence and document the rule.

Also apply a recency guard. A purchase, login or meaningful site event occurring minutes before a scheduled win-back send should stop it. This sounds obvious, but it requires the automation platform, customer data source and send system to share events quickly enough for the decision to be reliable.

Protect sending reputation before you reactivate a large dormant segment

Dormant audiences are intrinsically higher-risk than regularly engaged audiences. People may have forgotten the brand, changed jobs, abandoned an account or no longer recognise the subscription. Sending to the entire segment at once can create an avoidable concentration of bounces, unsubscribes and spam complaints.

Check the basics before activating the journey: SPF, DKIM and DMARC alignment; a stable and recognisable From identity; valid reverse DNS and TLS where you operate the sending infrastructure; and working unsubscribe processing. Google requires SPF and DKIM for bulk senders, DMARC alignment for direct bulk mail, and one-click unsubscribe for marketing and subscribed messages. It also warns that non-compliant traffic may face temporary or permanent rejections as enforcement ramps up. Google’s sender-guidelines FAQ provides the current detail. (Google’s sender guidance)

For one-click unsubscribe, use the standard intended for it: RFC 8058 requires an HTTPS URI in the List-Unsubscribe header and a List-Unsubscribe-Post: List-Unsubscribe=One-Click header, covered by a valid DKIM signature. The endpoint must action the request without asking the recipient to log in or complete another confirmation. RFC 8058 explains the implementation and security considerations, including the use of opaque, hard-to-forge identifiers. (RFC specification)

For a large segment, send in measured cohorts rather than switching on the whole automation at once. Start with the most recently active portion, monitor delivery results by mailbox provider and cohort, and only expand when recipient feedback is acceptable. Google recommends increasing volume gradually and monitoring feedback; its Postmaster Tools data can help identify Gmail-specific spam-rate and compliance issues. (Google’s sender guidance)

Watch more than conversion rate. Review hard and soft bounces, deferrals, spam complaints, unsubscribe rate, inbox placement where available, and performance split by provider. A cohort that produces weak revenue but high complaints is not a marginal success: it is a reason to narrow, slow or stop the programme.

Measure whether the automation improves the audience

Judge a win-back programme over a defined observation window, not by a single email’s click rate. Useful measures include:

• the proportion of entrants who complete a meaningful return event;
• time from entry to purchase, login or other desired action;
• incremental revenue or retention, using a sensible control group where volume permits;
• unsubscribe and complaint rates by message, segment, provider and source of consent;
• bounce, deferral and spam-folder trends; and
• the share of non-responders removed from routine mail under the sunset rule.

A small holdout group can be particularly useful. If a randomly selected, eligible portion receives no win-back messages, compare later conversion and retention against the contacted group. This helps distinguish genuine incremental recovery from people who would have returned anyway. Be careful to exclude people receiving other overlapping lifecycle campaigns, or the comparison will be muddied.

Review results by acquisition source as well. A consented newsletter subscriber acquired through a content download may behave very differently from a customer who opted in at checkout. If one source repeatedly produces complaints or non-response, the lesson may concern the original sign-up promise rather than the win-back email itself.

When a managed deliverability review is useful

If the dormant audience is large, complaints are rising, or provider-level delivery is already unstable, solve those problems before treating win-back as a revenue lever. This is where a deliverability review can be useful: authenticate and inspect the infrastructure, validate suppression handling, assess the audience, segment by likely human engagement, and introduce provider-aware pacing where the evidence supports it.

Email Foundry’s managed recovery work is relevant in that narrower situation, particularly when reactivating an audience could worsen an existing reputation problem. The sensible objective is not to force old contacts back into the inbox. It is to protect the engaged audience while establishing which inactive recipients can still be contacted responsibly and which should be retired.

A win-back automation earns its place when it gives people a credible reason to return and makes leaving easy. Build it as a bounded decision system, with clean exits, reliable suppression and a willingness to stop mailing the people who remain silent. That is better for customers, better for future campaign performance and better for sender reputation.