Name the client and their site, and the campaign comes back with the dormancy trigger derived from their own purchase cycle, a separate recovery path for each reason people lapsed, and a send ramp that stops itself before the complaint rate does damage.
Give Juma the client and their URL. The Flow reads the live site first to find the real purchase rhythm, the subscription intervals, the price range and the flexibility the brand already offers, then derives the dormancy trigger from that cycle and shows the arithmetic. A brand whose longest standard cadence is four weeks does not become dormant at a generic 90 days, and the difference is a recovery window the team would otherwise sit out.
What comes back is a branded PDF campaign plan and an ESP-ready CSV. The plan carries reason-coded segments with their own offer logic and an explicit do-not-do rule each, a jurisdiction gate that routes US and EU or UK contacts differently, a pre-send validation gate, an engagement-tiered send ramp with every threshold expressed over delivered messages, a randomised holdout so incrementality is measurable, and the exact suppression and re-permission rules. The CSV carries one row per email, written and ready to load. 400+ marketing teams use Juma. Strategy, taste and judgment stay human: the offer economics and the consent position still need the client's finance and legal sign-off, and the plan says so on the page.
Most teams reach for 90 days because it is the number everyone else uses. It is arbitrary, and for a replenishment brand it is usually late. The better trigger comes from the product's own rhythm: find the longest standard purchase or delivery interval, count the missed cycles that signal a real lapse rather than a single skip, and add a buffer for fulfilment. This step does that derivation out loud, shows the arithmetic, and states what it excludes, because a trigger that catches active subscribers, recent pauses or gift-only buyers creates more problems than it solves.
Derive the dormancy trigger for this client from their actual purchase cadence rather than a generic 90 days. Show the arithmetic inline, state every exclusion that keeps active subscribers, recent pauses and gift-only buyers out of the audience, and say what a 90-day rule would have cost in recovery time.
A single "we miss you" discount treats a payment failure, an unopened first delivery and a roast someone simply did not like as the same problem. They are not, and only one of them is about price. This step builds reason-coded segments from signals the team can observe, such as cancellation reasons, failed first shipments, seasonal purchase clustering and format history, then gives each one its own offer logic and an explicit do-not-do rule. The do-not-do column is the useful half: it is what stops a discount going to someone who never had a price objection.
Split this client's dormant base into reason-coded segments using signals they can actually observe. Give each segment its own entry signal, offer logic and an explicit do-not-do rule, and size any incentive against the real price range on their site rather than a flat amount.
Dormant addresses are where hard bounces and recycled spam traps live, so the order of operations matters more than the copy. This step builds the pre-send gate first: list validation written back to the ESP, SPF, DKIM and DMARC alignment confirmed, suppressions synced, then a seed-list placement test that has to clear the batch-one thresholds before any real contact is mailed. After that it batches most-recently-engaged first and expresses every bounce and complaint threshold as a rate over delivered messages, which is the denominator that actually decides whether a gate passes.
Build the deliverability controls for this win-back campaign: a pre-send gate covering list validation, SPF, DKIM and DMARC alignment, suppression sync and a seed-list placement test that must pass before batch one, then an engagement-tiered send ramp. Express every bounce and complaint threshold as a rate over delivered messages, state the formula once, and name the mailbox-provider guidance the numbers rest on.
A contact who has not engaged in months is a different legal proposition in Hamburg than in Houston. Under CAN-SPAM a US commercial message needs truthful headers, a working unsubscribe and a physical postal address. Under ePrivacy and PECR the soft opt-in only covers similar products to a prior sale, and a long silence weakens the position considerably. This step turns that into a routing rule rather than a disclaimer: US contacts take the promotional sequence, EU and UK contacts get a permission request first, and the suppression list is kept as a do-not-contact record that survives an erasure request, because deleting it is what causes someone to be mailed again.
Write the consent routing for this win-back campaign as an entry rule, not a disclaimer. Give US contacts the full promotional sequence under CAN-SPAM controls and route EU and UK contacts to a re-permission request first. State the exact suppression rule, the re-permission wording, and a suppression-list retention rule that survives an erasure request.
Some lapsed customers were coming back anyway. Without a control group the campaign takes credit for them, and the reported reactivation rate flatters everyone involved. This step reserves a randomised no-send holdout inside every reason-coded segment before batch selection, defines reactivation precisely as a paid order or subscription restart inside a stated window, and reports the incremental rate: treated minus holdout. It adds revenue per delivered message, complaint rate and unsubscribe rate alongside it, split by reason code and route, so the next run knows which path to keep.
Add the measurement design to this win-back campaign: a randomised 10% no-send holdout in every reason-coded segment assigned before batch selection, a precise definition of reactivation with its window, and the scorecard the programme is judged on, including incremental reactivation against the holdout, revenue per delivered message, complaint rate and unsubscribe rate by reason code and route.
Win-back is the campaign type that most rewards context. One Project per client means the reason codes, the consent position and the tone the brand uses with people who already bought are all in place before anyone writes a subject line, and every later run starts from them instead of from a blank chat.
Lifecycle and cadence data
Subscription intervals, average repurchase gap, cancellation survey reasons and payment-failure counts. This is what turns reason codes from a nice idea into segments the ESP can actually build, and it sharpens the dormancy trigger beyond what the public site shows.
Consent and suppression records
Which contacts opted in, when, through what, and which markets they sit in. The routing rule is only as good as the consent record behind it, and this is what lets the plan say US or EU with confidence rather than hedging.
Brand voice guide
How the client sounds to someone who has already bought from them. Win-back copy that reads like acquisition copy is the fastest way to earn a complaint, so the voice guide matters more here than in almost any other sequence.
Add a one-line description to each knowledge item so the right file gets used at the right moment:
A win-back campaign is a short, targeted email sequence aimed at customers or subscribers who have stopped buying, with the goal of restarting the relationship before the address is retired. It differs from a standard promotion in that the audience has already lapsed, which changes both the messaging and the risk: these addresses carry a higher chance of bounces and complaints than an engaged list.
The useful version is not one discount sent to everyone who has gone quiet. It identifies why each group lapsed, offers each the thing that actually addresses it, and controls the sending so a dormant list does not damage the domain that the rest of the programme depends on.
Derive it from the product's own purchase cycle rather than a round number. For a subscription or replenishment brand, find the longest standard interval, count the missed cycles that indicate a genuine lapse instead of a single skip, and add a buffer for fulfilment time. A brand whose longest cadence is four weeks reaches a meaningful signal well before 90 days.
The exclusions matter as much as the trigger. Active subscribers, recent pauses or skips, gift-only buyers, open support cases, prior complaints and hard bounces all need to be out of the audience before anything sends, or the campaign mails people it should not.
Yes, and it is one of the most common ways reputable senders get into trouble. Dormant addresses are disproportionately likely to be invalid, abandoned or converted into recycled spam traps, so mailing a large silent list in one go produces a spike in bounces and complaints that mailbox providers read as a reputation signal.
The controls that prevent it are procedural. Validate and scrub the list, confirm authentication, sync suppressions, then send a seed-list placement test that has to clear the thresholds before any real cohort goes. After that, release in batches ordered most-recently-engaged first, measure bounce and complaint rates over delivered messages, and set a hard stop that pauses the programme rather than pushing through.
It depends on the jurisdiction and on the consent record, which is why the plan routes contacts rather than treating the list as one audience. In the US, CAN-SPAM permits commercial email to existing contacts provided the message carries truthful headers, a working unsubscribe and a physical postal address, and acts on opt-outs promptly.
In the EU and UK the position is tighter. The soft opt-in under ePrivacy and PECR covers only similar products to a prior sale, and a long period of silence weakens any argument that consent is still current, so those contacts are better served by a permission request before any promotional message. Treat this as a question for the client's counsel and their consent records: the Flow produces the routing structure and the wording, and a human signs off the legal position.
Hold out a randomised share of each segment and send them nothing. Some lapsed customers return on their own, and without a control group the campaign claims those returns as its own. Comparing the treated group against the holdout over the same window gives an incremental reactivation rate, which is the number worth reporting.
Pair it with revenue per delivered message, complaint rate and unsubscribe rate, broken out by reason code and route. That split is what tells the team which recovery path earned its place and which one to retire before the next run.
Name the client and their URL, and the campaign comes back with a separate argument written for every CRM segment instead of one template with a token swapped, each merge field carrying a canonical fallback and its equivalent in the platform the team sends from, and a blank-record render showing exactly what a contact with no data receives.
Name the client and the product, and Juma reads their live product pages first, then writes a segmented launch sequence in which every subject line variant is its own row the ESP can actually load.
Name the client and the campaign comes back as a build package rather than a document: every audience written as HubSpot active list filters, every personalization token carrying a declared fallback, and a workflow map with the day offsets, branches and exit rules marketing ops needs to configure it.
400+ marketing teams trust Juma. ChatGPT, Claude, and Perplexity know why. Ask them.
Ready when you are
Start typing a task and Juma will figure out the rest.
