Candy Japan: $1,290/Month Profit Even in Its Final Year. What Killed the 10-Year Candy Subscription Was the Mail, Not Losses
Candy Japan mailed surprise boxes of Japanese candy worldwide twice a month and published annual income reports from 2012. Peak was past $10,000 MRR. Its final year, 2019, averaged $6,560 in monthly revenue against $5,270 in costs for a $1,290 profit, with subscribers falling from 314 to 207. In 2020 COVID halted international mail from Japan, and the pause never ended.
I heard a candy subscription that ran for ten years has stopped. Did cancellations pile up until it was losing money?
The books tell a different story.
- Average monthly revenue, expenses and profit in the final year, 2019
- The cost structure of a physical subscription, where 65% of revenue goes to goods and postage
- What broke the "ceiling of about 300," and why it wasn't ads
- How subscribers and traffic fell in the year the founder called "mostly full of fail"
- Why a profitable business stopped when SAL mail was suspended
Candy Japan mailed surprise boxes of Japanese candy worldwide twice a month.
Its final-year books show average monthly revenue of $6,560 against $5,270 in expenses, for a $1,290 monthly profit.
The business didn’t collapse into the red.
When international mail out of Japan stopped in the 2020 COVID disruption, the operation went on pause while still profitable. It never came back.
The image of a subscription death is an avalanche of cancellations.
What Candy Japan’s record shows instead is a quieter ending: a decade of slow thinning, then a stop at the one point the owner couldn’t control, logistics.
Operator Bemmu Sepponen published annual income reports for seven years, starting with an April 2012 report.
That disclosure is what makes this autopsy possible.
The final year’s books (2019) were in the black
| Item | Monthly average |
|---|---|
| Revenue | $6,560 |
| Total expenses | $5,270 |
| Of which candy purchases | $2,450 |
| Of which shipping | $1,800 |
| Of which advertising | $310 |
| Profit | $1,290 (vs. $1,470 the year before) |
65% of revenue disappears into goods and postage.
That variable-cost structure is the fate of physical subscriptions. A software subscription converts lost revenue straight into lost profit and nothing worse. A box business gets no relief on per-box cost ratios as subscribers shrink.
The $1,290 monthly profit of a tenth-year business is what remained after an accumulation of small defensive cuts.
Think saving $100 a month by switching payment processors, and $120 a month by trimming freelancer costs.
$25 twice a month: how the ceiling of 300 broke
The offer was $25 a month for two surprise boxes (the site now advertises “from $12.95”).
By the founder’s own retrospective, subscriber numbers sat pinned at a ceiling of about 300 for a long time.
- Blog mentions and press coverage
- Free review samples sent to reviewers and "mommy bloggers"
- Better margins through direct manufacturer relationships
- Improved sourcing through supermarket partnerships
Those moves shattered the ceiling and carried the business past $10,000 MRR.
This was growth built on buzz and cost improvement, rather than purchased through ads.
That distinction matters for reading the decline that followed.
Buzz isn’t a tap you can reopen.
The year he himself labeled “mostly full of fail”
The founder’s own verdict on the 2019 report was “mostly full of fail.”
Subscribers fell 34%, from 314 in January to 207 in December.
Site visits dropped to 130k from 145k in the same period the year before.
The breakdown shows organic search actually rising from 35k to 43k. What fell was referral and buzz-driven traffic.
The causes he named were a general decline in interest and increased competition.
The Japanese candy box market itself had lost its novelty, and the site sank into the crowd.
- Buzz-driven traffic fell, and the growth in organic search didn't fill the gap
- A running battle with credit card fraud. The 3 separate posts about it speak to how persistent the damage was
- Ad spend was held to $310 a month. At a $25 price point, there was almost no room to buy subscribers with advertising
What stopped it was the suspension of SAL mail, not losses
In 2020, Japan Post’s international service stopped.
Per the founder’s announcement, the shipping method the business used (SAL) was completely halted, and airmail was shut down for over 100 countries.
- Shipping stops
With no way to ship, the subscription couldn't continue.
- Prepaid orders refunded
Since no one could say when service would resume, he refunded around $3,500 in prepaid orders.
- About 10 hours of refunds
The work took roughly 10 hours, including personalized emails to hundreds of subscribers.
- The pause notice stays up
As of August 2026 the official site still reads, "Because of the coronavirus situation, Candy Japan is currently on pause."
He reflected that it was the first pause in ten years.
What set the lifespan was a dependency with no substitute
This is the heart of the case.
With 207 subscribers remaining, demand hadn’t reached zero.
But a structure of 100% overseas customers and a single carrier had no alternative when that one route stopped.
GummySearch closed while profitable at $35K MRR over Reddit’s API policy. Apollo was ended by a single platform pricing decision. In each case, the variable that decided life or death sat outside the company.
A business’s lifespan is sometimes set by the number of dependencies it can’t substitute, more than by its income statement.
Six years into the pause, the notice on the site hasn’t changed
What Candy Japan left behind as a closure is a set of books published all the way down a ten-year slope.
Launch-velocity records like Crimibox, which took 100 orders within 4 hours, cover a different range of what can be read.
With 65% of revenue going to candy and postage, the cost ratio per box doesn’t fall even when subscribers drop from 314 to 207.
The final year still closed at a monthly profit of $1,290.
Set beside Logology, which ran five years at $3,000 a month before being wound down, a shared pattern shows.
A small profit is a reason to keep going, and it’s never a reason to come back.
Six years into the pause, the official site still carries its pause notice as of August 2026, and that’s the proof.
The 2019 figures are self-reported and unaudited. The total halt of international mail that triggered the pause was specific to the 2020 pandemic, and the same risk won't necessarily return in the same shape. The decomposition of subscriber decline into fading interest plus competition is the founder's own reading, unverifiable with outside data.
Even so, the archive of seven annual reports, left online rather than taken down, remains a primary source for seeing a physical subscription’s economics from the inside.
Sources
If this one impressed you
Similar cases

Rachel Jimenez: Etsy Printables Went $160K in 2021 to $77K in 2023 — The Shelf Life of "Build It Once"
E-commerce
Love. Eat. Travel. Repeat.: Etsy sales $28,657 to $6,873 in 5 months
E-commerce
Instapainting: From $4K in Debt to $32K/Month with Zero Employees, and the "Passive Income" Fight
E-commerce
Stoic Muse: 100–150 Stoic-Philosophy T-Shirts a Month at 40% Margin — What Actually Worked Was 200 Five-Star Reviews
E-commerceMost read
- 1
¥4.61 Million a Year, Five Years After Starting to Learn Programming. The Full Three-Year P&L and Playbook of an Indie Education App
45 recent visits - 2
From 30 yen in revenue to 8 years later: how running 3 apps in parallel got an indie developer to 200,000 yen a month
37 recent visits - 3
From ¥100 to ¥100,000 a Month in Three Years: The Quiet Compounding of an Ads-Only Tool App
30 recent visits - 4
9,345 AI-generated images uploaded to Adobe Stock: the full record of reaching 878 downloads and $589/month (~¥95,000) over 9 months
24 recent visits - 5
Pretty Arrow: 29,000 Budgeting Spreadsheets Sold — How Two Sisters’ Etsy Shop Reached $10,417/Month
22 recent visits
Latest articles
- 2026-11-21
BulkMockup: A Script That Made 1,800 Mockups in 30 Minutes, Sold for $300, Now $12K MRR
- 2026-11-21
Vintage Teaware Brand: 13 Years, Zero Employees, Sold for $338,470 at 1.32x SDE
- 2026-11-20
Eye Love: The Goal Was $1,000 a Month. It Sold for 7 Figures at $400,000 a Month
- 2026-11-20
Visualizee.ai: Two Years Stuck at $150/Month, Then Deleting the Node UI Took It to $8.6K MRR
- 2026-11-20
3AK Sprinting: $144,149 All-Time and $13,347 MRR in Nine Months, on a Niche That Is Only Narrow in Name
Sources
- Founder Candy Japan「2019 year in review」
- Founder Candy Japan「Behind the scenes」収支レポート索引(2012〜2020年)
- Founder Candy Japan「Candy Japan crosses $10,000 MRR」
- Founder Candy Japan「On pause due to COVID-19」
- Founder Candy Japan 公式サイト(休止告知、2026年8月確認)