Step 1 (days 1–2): baseline, attribution window, cap and suppression list in Setup
The first step with a Retention FTE is to record the numbers the month will be judged against and to fix the rules before any flow changes. The baseline is repeat-purchase rate at the chosen horizon, revenue per recipient for each live flow, list size and recent unsubscribe rate, all read from the sending platform and the store as they stand. The attribution window is chosen now and written into Setup, so no later report can pick a window that flatters.
The frequency cap and the suppression list are also set now. The Retention FTE never asks for what onboarding already captured, such as the store platform or the brand name, so Setup is short: voice, discount rules per segment, the cap, the suppressed customers, and who approves.
Step 2 (days 3–7): review the flow audit in the Flows tab
The second step is the flow audit, which the Retention FTE runs on connection and reports in the Flows tab. Every live flow is mapped: its trigger, its messages and their spacing, its segment and size, and its recent revenue per recipient. Beside each timing sits what the order data says: the median reorder gap for replenishment SKUs, the typical return gap for win-back, the delivery window for post-purchase. Where the flow and the data disagree, the audit says by how much.
The owner's job is to read the audit and set the order of work. The Retention FTE proposes starting with the flows where the gap between current timing and the data is largest, because those are the changes with the clearest reasoning.
Step 3 (days 8–21): approve flow changes in Approvals
The third step is the first proposals in the Approvals tab. Each proposal is one flow change: a retimed replenishment reminder, a rebuilt post-purchase sequence, a new lapse threshold for win-back. It shows every message as the customer will see it on the device it targets, the segment definition and size, the reasoning from the audit, and the frequency cap already applied. Copy is in the store's voice from the remit and makes no product claim the catalog does not support.
The owner approves, edits or sends back with a note. Send-backs become remit rules: a phrase to avoid, a segment that must never see a discount. Reviewing a proposal takes minutes because the preview and the reasoning are on one screen.
Step 4 (days 8–30, daily): watch cap and suppression enforcement in the brief
The fourth step runs daily from the second week: the morning brief in the Today tab reports sends against the frequency cap, any sends held back because they would have breached it, and confirmation that the suppression list was applied. A campaign the owner scheduled that collides with a flow for part of a segment appears here as a held-back count with the reason.
The brief also lists signals the Retention FTE noticed: an unsubscribe spike after a change, a deliverability warning from the platform, a segment that grew unexpectedly. Each signal carries what the FTE can see and a proposed action if one is warranted.
Step 5 (day 30): the review in Performance and the autonomy decision
The fifth step is the monthly review in the Performance tab. It compares repeat-purchase rate and revenue per recipient per flow against the baseline, against the attribution window printed on the report, and names the flows that underperformed with a hypothesis and a proposed change. It proposes holdouts on the flows where an incremental claim would matter, so next month's review can say more than this one.
The autonomy decision follows the approval record. A flow that ran for weeks with approvals and no send-backs can be promoted to run its scheduled sends without a gate; campaigns and new flows usually stay gated. Any promotion can be reversed from Setup.
What does done look like after 30 days?
Done after thirty days with a Retention FTE looks like flows whose timing comes from the store's own order data, a suppression list and cap enforced on every send, a report that prints its attribution window, and at least one holdout running so the next review can speak to incrementality. It also looks like a remit that has grown by the rules the send-backs taught it.
Done does not look like a revenue claim without a holdout. Lifecycle numbers are easy to inflate with a generous window, and the first month's honest result is a baseline the second month can beat with evidence.