What happens, week by week
Every engagement follows the same spine: measure first, suppress second, build third, report forever. The sequence exists because each stage de-risks the next; flows built on an unaudited list inherit its complaint problems.
Weeks 1–2: audit sprint
The send-vs-engagement map, a GlockApps seed test across Gmail, Outlook, and Yahoo, complaint rates read against Google's 0.1% guidance, SPF/DKIM/DMARC review, and a CASL consent register. Output: a ranked fix list, priced.
Week 3: suppression pass
High-churn-risk and long-inactive segments come out of routine sends. This is the fastest reputation lever we have: reversible in one click and visible in complaint data within two weeks.
Weeks 3–4: predictive segments
CLV tiers, churn-risk gates, and expected-date-of-next-order cohorts, built inside your ESP. Accounts below Klaviyo's predictive thresholds (500+ customers, 180 days of history) get RFM proxy segments instead, swapped for models later.
Weeks 4–10: flow builds with parallel runs
The five core flows, built and validated one at a time. Each new flow runs in parallel against a held-back cohort (the pet-supplies replenishment flow ran 12 days) and cuts over only after beating baseline on placed-order rate.
Monthly: the working session
The AI sweep runs across every campaign and flow; an analyst verifies findings; you get one page of numbers, one page of decisions, and 45 minutes of argument about them. Reporting is in placed-order rate and revenue per recipient, never MPP-inflated opens.
Why this order and not flows first?
Because sequencing is a compounding bet. Suppression raises inbox placement, which raises every flow's effective reach, which shortens the parallel runs, which gets the full build live sooner. Run the same work in reverse order and the early flows fight a reputation headwind the whole way. Thecoffee-subscription case studyshows the pattern, including the awkward fortnight where sends dropped before conversion caught up. Details of each stage live on theservices pages.
How long before the first change ships?
Suppression changes ship in week three, immediately after the audit reads out. They are low-risk, reversible in one click, and touch every send, which is why they go first in the sequence. The coffee-subscription engagement cut 22% of send volume in its first suppression pass, and complaints fell from 0.19% to 0.06% within eight weeks.
Who from our team needs to be involved?
One decision-maker for a weekly 30-minute check-in and whoever approves copy. That is it. We build inside your ESP with your logins, so there is no integration project and no engineering dependency. Copy approval speed is the main variable your team controls: three-day turnarounds keep the 8-week flow timeline intact.
What happens if the audit finds deliverability damage?
Repair jumps the queue, because building flows on a damaged domain wastes them. Expect suppression of complaint-generating segments, authentication fixes, and a monitored volume ramp: in one referenced repair, complaints fell from 0.42% to 0.08% over six weeks while send volume dropped 38% during the suppression phase. Growth work resumes after reputation stabilizes.
When does an engagement end?
The build program has a defined finish line: the five core flows clear their parallel-run baselines and cut over, typically at the 8-to-12-week mark. After that you either run the program yourself with the runbooks we leave, or keep the monthly intelligence retainer, which is month to month with 30 days notice.
Start at week zero
The working session is the audit sprint's first hour, free: your data, our questions, a written scope at the end.
Book a working session