Case study · Food & Beverage
Subscribers who pause instead of cancelling
A coffee roaster was losing subscribers who only wanted to skip a month. We rebuilt subscriptions on Shopify so customers control their own schedule — and churn fell 38%.
Engagement · Build & rollout
Subscription commerce where retention is the entire business model.
- Industry
- Food & beverage
- Solution
- Subscription commerce
- Platform
- Shopify
- Engagement model
- Dedicated product team
- Scope
- Subscriptions, portal, dunning, forecasting
- Model
- Recurring, customer-scheduled
- Constraint
- Roast and dispatch planning
- Target
- Reduce avoidable churn
Outcomes
Where subscription revenue actually leaks
Most churn in a subscription business is not a decision to leave. It is friction, or a card that expired.
-
−38%
Subscriber churn
Customers can pause, skip, change frequency and swap the coffee themselves. The cancellations that were really "I have too much coffee this month" stopped being cancellations.
Measured across monthly churn rate for 12 months after launch
-
82%
Of failed payments recovered
Card failures are retried on a schedule tuned to how bank declines actually behave, with the customer prompted to update details before the subscription lapses rather than after.
Measured across failed renewal attempts after launch
-
+2.4 mo
Average subscriber lifetime
Lower churn and better payment recovery compound: the average subscriber now stays nearly two and a half months longer, which is the whole economics of the business.
Measured across cohort lifetime after 12 months
Context
A subscription business losing subscribers to admin
The coffee was not the problem. The only self-service action was "cancel".
The business
A speciality coffee roaster selling freshly roasted subscriptions direct to consumers, roasting to order on a weekly cycle.
The starting point
Subscriptions ran on a bolt-on app with almost no customer self-service. Changes went through support by email.
The trigger
Support was handling hundreds of "can you skip this month" emails, and a meaningful share of cancellations were customers who had given up asking.
What they wanted
Full customer self-service over the subscription, proper payment recovery, and a roasting forecast they could plan against.
Constraints
Coffee is roasted to order, so schedule changes affect production · freshness means dispatch timing matters · card failures are a large share of lost revenue · customers want to change their coffee, not just their date.
System
What it runs at today
The subscription programme as it runs today.
-
−38%
Churn
Against the pre-launch rate
-
82%
Payment recovery
On failed renewals
-
Self-serve
Schedule control
Pause, skip, swap, reschedule
-
Weekly
Roast forecast
From live subscription data
The engineering problem
Four ways subscription revenue disappears
Only one of them is a customer deciding they no longer want the product.
-
Cancel is the only self-service option
When pausing requires an email and cancelling takes one click, the interface makes the decision for the customer.
What we did
A subscriber portal where pause, skip, reschedule, change frequency and swap product are all one action, and cancel is not the easiest one on the page.
-
Cards expire and nobody chases properly
A failed renewal that is retried once and then abandoned is a lost subscriber who never decided to leave.
What we did
Dunning tuned to real decline behaviour: retries spaced to match how issuers behave, with the customer prompted before the subscription lapses.
-
Customers get bored of the same coffee
A single-product subscription is fragile. The customer who wants something different cancels rather than asks.
What we did
Product swapping inside the subscription, with rotating and surprise options, so variety is a change rather than a restart.
-
Production planning by guesswork
A roaster planning against last month's volume either wastes coffee or misses dispatches, and both cost.
What we did
Forward roast and dispatch forecasting from live subscription data, including known pauses and scheduled changes.
Architecture
How it fits together
Simplified — the shape of the system rather than every service in it.
-
Subscription model
- Plans
- Schedules
- Products
The subscription as a schedule a customer owns rather than a contract they endure.
-
Subscriber portal
- Pause & skip
- Reschedule
- Swap
- Payment details
Every routine change self-service, without contacting anybody.
-
Billing & dunning
- Renewals
- Retry schedule
- Recovery prompts
Payment failure treated as a recoverable event, not a cancellation.
-
Forecasting
- Roast volume
- Dispatch planning
- Known changes
Production planned against what is actually scheduled.
Pausing is the most important feature in the build. Every subscription business that makes pausing hard is converting temporary friction into permanent churn.
Solutions
What we built
A subscription the customer controls.
-
Subscription engine
Plans, schedules and products as a model the customer owns.
-
Subscriber portal
Pause, skip, reschedule and swap, all self-service.
-
Dunning
Retry scheduling matched to decline behaviour.
-
Renewal notices
Advance notice with a change link.
-
Product swapping
Rotating and surprise selections inside the subscription.
-
Forecasting
Roast and dispatch volume from live schedules.
Key capabilities
What it does day to day
Six capabilities across the subscription lifecycle.
| Capability | Runs | Refresh | What it does |
|---|---|---|---|
| Plan management | Customer | Anytime | Frequency, quantity and product, changed self-service |
| Pause & skip | Customer | Anytime | Skip one delivery or pause for a period, without cancelling |
| Product swapping | Customer | Per delivery | Change coffee or take a rotating selection |
| Dunning | Automatic | On failure | Retry schedule matched to decline behaviour, with prompts |
| Renewal notices | Automatic | Before charge | Advance notice with a change link, not just a receipt |
| Forecasting | Automatic | Weekly | Roast and dispatch volume from live schedules |
Integrations
How the moving parts plug in
A subscription the customer steers.
Before renewal
- Advance noticeWith a change link
- Skip or swapOne action
- PauseInstead of cancel
Renewal
- Payment taken
- On failureDunning begins
- Retry scheduleMatched to declines
Production
- Roast forecastFrom live schedules
- Known pausesAlready accounted for
- Dispatch planFor freshness
Sending an advance notice with a change link, rather than a receipt after the charge, moved a large share of would-be cancellations into skips.
Security & data
What protects subscribers
Recurring billing has a duty of clarity as well as security.
-
No stored card data
Payment details stay with Shopify and the payment provider — we never handle them.
-
Advance notice
Customers are told before they are charged, with a one-click route to change it.
-
Easy exit
Cancellation is genuinely available, because a subscription business built on trapped customers does not last.
-
Change history
Every schedule and plan change recorded, so a billing question can be answered.
The brief
Make pausing easier than cancelling
The roaster assumed its churn was a product problem and was considering changing the coffee. The data said otherwise: cancellations clustered around holidays and around the point where a customer had too much stock at home.
Those are pause events dressed as cancellations, and the old system had no pause button.
- Full self-service over schedule and product
- Dunning tuned to real decline behaviour
- Advance renewal notice with a change link
- Roast forecasting from live subscription data
What the build had to respect
- 01Coffee roasted to order on a weekly cycle
- 02Freshness, so dispatch timing matters
- 03Card failures as a major revenue leak
- 04Customers who want variety, not just a different date
Process
We read the cancellation reasons first
The roaster had a year of cancellation emails nobody had ever aggregated.
-
Churn analysis
A year of cancellation emails and support threads coded by reason, which is where the pause hypothesis came from.
-
Portal first
Self-service scheduling built and launched before anything else, because it addressed the largest churn category.
-
Dunning tuning
Retry schedules tested against historical decline data rather than set to a default.
-
Product swapping
Added once scheduling was stable, to address the variety category of churn.
-
Forecasting
Built last, once there was enough clean schedule data to forecast against.
Technology
Shopify with a subscription model that fits the business
Commerce and payments on Shopify, subscription logic built for a roastery.
Commerce
- Shopify
- Subscriptions API
- Storefront API
Portal
- Customer portal
- Schedule management
- Product swapping
Billing
- Recurring billing
- Dunning
- Recovery prompts
Planning
- Roast forecasting
- Dispatch planning
- Cohort reporting
Business impact
What changed for the roaster
Three outcomes over the first year.
-
Churn down 38%
Mostly by converting cancellations into pauses and skips.
-
82% of failed payments recovered
Revenue that was previously just gone.
-
Production planned properly
Roast volume forecast from real schedules rather than last month.
The result
Subscribers stay about two and a half months longer
Churn is down 38%, 82% of failed payments are recovered, and average subscriber lifetime is up by 2.4 months.
Support went from hundreds of scheduling emails a month to almost none, because customers now do it themselves in under a minute.
- Churn down 38%
- 82% of failed payments recovered
- Average lifetime up 2.4 months
- Roast volume forecast weekly
What we hold to in subscription commerce
- 01Make pausing easier than cancelling
- 02Treat a failed card as recoverable, not as churn
- 03Notice before the charge, not after
- 04Read the cancellation reasons before changing the product
Free discovery call
Have an idea? Let's turn it into AI-powered software.
Book a free discovery call with our experts. Share your idea and we will help you shape the scope, timeline and budget, under NDA.
- Free consultation
- NDA before we talk
- Transparent estimate