Email goes from 9% of revenue to a third, adding $420K a month
One weekly broadcast to everybody, two abandoned automations, and 38% of the list dormant for a year. Rebuilding it models email-attributed revenue from $90K to $483K a month.
| Measure | Before | After |
|---|---|---|
| Email-attributed revenue / month | $90K | $483K |
| Added cash flow per month | — | +$420K |
| Email share of total revenue | 9% | 34% |
| Revenue per recipient, per send | $0.11 | $0.68 |
| Subscription churn, monthly | 7.4% | 4.1% |
The problem
Email was contributing 9% of revenue and treated as an announcement channel: one send a week to the whole list, regardless of whether someone had subscribed that morning or churned eighteen months ago. Deliverability was degrading under the weight of unengaged contacts, and the two flows that did exist — a welcome email and a cart reminder — had not been touched in two years. Subscription churn was running at 7.4% a month.
What the model found
Order-interval data per product is modelled to estimate how fast each customer actually consumes it, rather than using the label's suggested duration. Reminder windows are set per customer from that estimate. This one change accounts for most of the churn improvement.
The plays
Model consumption per customer
Order intervals per product feed a per-customer consumption estimate, which sets each replenishment reminder window individually. Getting this wrong in either direction is the biggest source of avoidable churn.
Sunset the dead weight
A four-email reactivation attempt, then suppression of the 28,000 contacts who do not respond. Sending to fewer people raises inbox placement, and revenue per send rises immediately.
Build the flows that carry the revenue
Nine automations: welcome with a first-purchase path, browse and cart abandonment, post-purchase education, replenishment, win-back, and a pre-churn save for subscribers.
Segment on behaviour, not demographics
Sends keyed to purchase recency, category affinity and subscription status. The weekly broadcast becomes four variants of a weekly broadcast.
Test the whole flow, not subject lines
Offer structure, sequence length and timing — the things with real effect sizes — rather than another month of subject-line tests.
The lever
Replenishment timing. Nine flows ship, and the one that maps reminders to how fast people actually use the product out-earns the other eight put together.
How the model was calibrated
Sized against Litmus’ benchmark of 45:1 email ROI in retail, e-commerce and consumer goods — the highest-return sector in that study, ahead of agencies at 42:1 and software at 36:1. That is the reason this work is sequenced last in a program and valued first in the model: it is the cheapest place to put attention that has already been earned somewhere else.
Sources
- Litmus — State of Email / ROI of Email Marketing — Return per $1 spent by sector: retail, e-commerce and consumer goods 45:1; agencies 42:1; software and technology 36:1.
- Semrush — AI Search & SEO Traffic Study — ~90% of ChatGPT search citations point to pages ranking 21+; AI search visitors convert at 4.4× ordinary organic; 50% of ChatGPT links go to business and service sites.
Want the same read on your category?
Send a brief and we will run twenty prompts before we talk.