Background
Caffeine & Legends is a coffee/tea/cocoa subscription box with a fandom angle — including the quarterly Romantasy box themed around romance-fantasy book releases. Patty runs the brand with a 32K-subscriber list and steady MRR. The business has two interlocking growth challenges: catching first-time buyers and converting them into the 3+ order retention cohort that actually drives subscription economics.
The problem
CPA had spiked to several times the target range. The team didn't know why — the homepage was performing well, ad spend was steady, and creatives weren't obviously fatigued. Under the hood, the Facebook conversion pixel had stopped firing during a homepage A/B test running through Shopify. Without conversion signals, Facebook started experimenting with audiences and the Romantasy box and retargeting campaigns were all running well above the target CPA range.
Retention was a separate problem: only 57% of first-time buyers were reaching 3+ orders. The target was 70%. And email was on a monthly tool with a bounce rate that was eating deliverability.
What we did
Pixel diagnosis, in one call. Caught the missing pixel using MetaPixel Helper. While on the call: shut down the worst-performing creative. Disabled the elevated-CPA retargeting campaigns. Reinstalled the pixel on the new homepage.
Restructured the FB account. Two campaigns were targeting the same audience with different budgets. Advantage Plus wasn't properly configured. Split tests were set up against the wrong variables (LAL vs Fantasy interest). Cleaned up all three.
Habit-building email sequences. Built sequences keyed to buyer cohort:
- First-time buyers: day 3 educational (brewing guides), day 18 subscription pitch
- Two-time buyers: day 12, personalized by drink type (tea/coffee/cocoa)
- 3+ order cohort: special messaging, early preview access to new boxes, exclusive subscriber notifications
Klaviyo migration. Better segmentation, direct sync to Facebook/Google ads, centralized audience management — and dropped the prior tool's monthly cost. Used a one-time list-cleanup service for the 32K list — better than the built-in cleaning.
Demographic enrichment. Layered in a third-party demographic enrichment service to Klaviyo so audiences could be segmented more precisely.
Scaled Google Ads. The PMark sales campaign was converting roughly an order of magnitude cheaper than Facebook. Recommended and shipped a meaningful budget increase. Killed branded search (minimal value).
Results
| Metric | Before | After |
|---|---|---|
| Monthly recurring revenue | Baseline | ~2× growth |
| Cost per acquisition | ~5× over target | Back in target range |
| Retention to 3+ orders | 57% | Targeting 70% |
The MRR doubling didn't come from one fix. It came from catching the tracking gap before the BOGO campaign torched inventory, then compounding habit-building retention emails through cohort after cohort.
In their words
"When you do an A/B test on your homepage and lose the pixel, you don't always know. Andrei and the team found it in a week and saved the BOGO campaign." — Patty, founder, Caffeine & Legends