Work Ferngrove Coffee
Ferngrove Coffee: find out which emails keep subscribers.
A small roaster sends four emails a month to its subscribers. Open rates look fine, but nobody knows whether any of them stop people cancelling.
01The brief
Client
Ferngrove Coffee: a small coffee roaster selling a monthly bean subscription online.
Goal
Keep more subscribers past their third month.
What was happening
Ferngrove sends four emails a month. Opens look healthy, but nobody can say whether any email stops people cancelling, or prompts them to.
Constraint
A small team. The report has to take five minutes to read.
02The creative
Strategy, copy and design by Forward.
FFerngrove Coffeeto you
Your next bag ships Thursday. Keep it or swap?
Arriving Thursday
Ethiopia Guji, washed
Bright and tea-like, with peach and a clean finish. Roasted on Monday, so it will taste its best from day 7 to day 10.
Keep this bagSwap the roast
Going away? Pause next month in one click.
- The subject says what and when. Subscribers know what’s coming and when it ships before they open it.
- One decision per email. Keep or swap. Each button is tagged, so the report can see what people chose.
- Pause sits next to the choice. Someone going away pauses for a month instead of cancelling for good.
Monthly report · sample
The discount reminder may be prompting cancellations.
Example figuresCancellations per 100 subscribers in the 7 days after each send
| Got it | Holdout | Difference | |
|---|---|---|---|
| “Next bag” email | 0.9 | 1.8 | −0.9 |
| Roast news | 1.2 | 1.6 | −0.4 |
| Brew guide | 1.7 | 1.7 | 0.0 |
| Discount reminder | 2.6 | 1.8 | +0.8 |
Next change: Pause the discount reminder for a month and see whether cancellations fall.
Photography for the subscription emails.
03The decision
Judge each email by what subscribers do in the seven days after it, not by opens.
An open tells you an email was seen. It doesn’t tell you whether it kept anyone.
So every email gets tagged, and we count skips, swaps, pauses and cancellations in the week after each send, against a small holdout group that doesn’t get that email. The “next bag” email also gets a one-click pause, so a subscriber going away pauses instead of cancelling.
04The outcome
What we test
A 10% holdout group for each of the four emails, for two months.
How we measure it
Cancellations per 100 subscribers in the seven days after each send, compared with the holdout group.
Baseline
Three months of cancellation data from before tracking started.
What counts as a win
At least one email that measurably lowers cancellations, and a clear call on any email that raises them.
What happens next
Emails that lower cancellations stay and get tested further. Any that raise them get rewritten or dropped.
What would you like your marketing to do better?
Tell us in a few sentences. James, our strategy lead, replies by email within two business days.