369 Cups, 15 Kilos and One Viral Video: How NPB's Beer Model Got Trapped by Its Own Success

The stairs at Belluna Dome are not a smooth rake. Every two or three rows there is a riser, and on that riser you balance with fifteen kilos strapped to your back. Mana Kato, 21, did exactly this for three years — seat to seat, cup to cup. Her first game produced 100 cups of draft beer. By her third year her personal record was 369. Roughly 3.7 times.

In early October 2026 a video came out of a stand in Yokohama. A spectator ordered so many cups that they had to be poured into a container — the container normally used for discarded cups. Japanese social media found a word for it: "tarugai", buying the whole keg. It circulated first as a joke, then as an accusation.

A transparency note is required here. The material behind this report is interviews published by The Japan Times — with Kato and with Yuichi Ito, a 56-year-old regular spectator. That material contains no football: no formation, no set-piece design, no coach, no xG. So no tactical claim is made below, and where the source has no number I have labelled the arithmetic as an estimate rather than dressing it as data. What we have is a stadium-operations and labour-economics case. I will audit it the way I audit dead-ball routines — I started this one with a stair count, not a theory.

NPB's in-seat draft service is not a legacy quirk; it is a deliberate product differentiator. MLB delivers beer down the aisle — cold can or bottle, transactional, you queue for it. Japan delivers it to the seat: the vendor carries a draft system on his or her back to your row, pours on the spot, and next time knows your name. Kato sold Kirin draft beer. Freshness, immediacy and relationship are three parts of one design. The design is not cheap. The cost is paid in labour, and specifically in the knees of a 21-year-old student.

Kato applied in 2026 after watching TikTok videos, and by her own account knew almost nothing about baseball. She left the job in October 2026 to prepare for graduation. She recruited her younger sister into the same work. Ito sits at the other end of the same distribution: a 56-year-old regular, seven or eight beers a game, and hard self-policing — no covert photographs, no following a vendor out into the corridor, no crossing the line. Both testimonies are tails of one distribution. Neither Kato's experience represents the whole workforce, nor does one viral clip represent the whole crowd.

Now the measurement. The first sum nobody runs: the reload cycle. If a laden pack of roughly fifteen kilos carries on the order of ten to fifteen litres of net beer, and a pour is 350–400 ml, then one load yields twenty-five to thirty-five cups. 369 cups implies ten to fifteen trips back to refill per game — a logistics chain that includes leaving the bowl, re-entering through the crowd, and rebuilding the same seat sequence. The source describes neither keg-swap mechanics nor reload timing. The calculation is an estimate and I will call it one. But if it holds, the real product of "beer at your seat" is not beer; it is a stair-logistics system whose only engine is a human body.

The second sum is the one that tells you most: 100 to 369. A low-skill commodity role does not produce that spread over three years. The learning curve says stadium vending is a high-ceiling job, not a low-skill one — it demands physical endurance, social memory and route planning simultaneously. In a piece-rate system, dispersion of that size usually produces two outcomes: novices exit inside a season, and a small, high-earning veteran core remains. Kato was that core. The model marketed as "student labour" is in fact a screening machine, sorting on body and conversation.

The third: name-recall capital, which sits with the vendor, not the club. Kato's sales engine was her regulars. She introduced herself, remembered names, and went straight back to the same seat next time. This is familiar ground. When I built reception maps, the question was never where the ball went but where it was welcomed. Kato's seat map looks less like a sales route and more like a nervous system — a few nodes she returns to constantly, and whole sections in between where nobody stops.

Here is the structural fault. Graduation ends the job by design, not by resignation. When Kato leaves, the regular-customer map leaves with her. The club had a revenue-generating relationship and nobody owned it — not the club, not the operator, not the vendor under any contract. Every season a new cohort arrives and the map is rebuilt from zero. It is the lesson Bengaluru FC's performance unit taught me in 2026: the most valuable file after a match is the clip pack the club keeps, not the player. Here that file walks out of the building every third year.

The fourth point is the uncomfortable one. The piece-rate structure and the behaviour now being condemned are two halves of the same design. Income partly depends on cups sold. The selling method is relationship. A customer who orders ten cups and drinks none is therefore not a bad customer — he is the best customer in the section. In the Yokohama video the beer was poured into a discard container, which is to say it was destroyed. The real product of that transaction was not beer; it was attention, and attention converts an employee into a spectacle she cannot step out of. The purchase is legitimate, which is precisely why this sits in a regulatory grey zone and precisely why no clean response exists yet.

The fifth: the document that is not there. Nothing in the available material describes a supervisor, a floor manager, a security escort, a complaints channel or an incident form. The two things Ito disciplines himself against — covert photography, following a vendor into the corridor — tell you those behaviours exist, or there would be nothing to restrain. Yet no rule text is quoted, no operator statement appears, no sanction is cited. Caution is warranted and I will date-stamp it: the absence of an operator voice in a republished feature does not prove the absence of a policy. It proves nobody was quoted. "No comment" and "no story" are different things, and this may be the first.

The sixth: recruitment now runs through lifestyle content, not baseball fandom. A 21-year-old applying because of TikTok is a cheap pipeline and a fragile one. The same imagery that fills it manufactures the disillusionment — the video sells atmosphere, the job delivers fifteen kilos and humidity. Where hiring doesn't happen through content, it happens through referral: Kato brought her sister. That is an informal network, not HR, and an informal network can dry up for a season after one viral negative clip.

The seventh: the shape of the clock. Student labour means the workforce age curve is driven by the university calendar, not the baseball season. Experience drains every three or four years. That is a permanent operating cost no balance sheet shows. I learned the same thing inside the 2026 empty-stadium ISL bio-bubble: an empty stadium is not silence, it is a laboratory with better acoustics. The same applies here. When the noise of 40,000 people makes a vendor inaudible, safety will be solved by process, not by volume.

Now the contrarian read. The video social media has cast as this story's crisis is not the story. The story is that the model pays for the exact behaviour it now condemns. If the operator introduces a strict fan-conduct rule next month, the first casualty will be the relationship-based selling that produced the 369-cup record. Writing a rule is easy; holding revenue under it is not. The likeliest outcome is that one individual is sanctioned — a stadium ban for the fan in the clip — and the model survives intact. The best outcome is that the incident pulls a protective framework into place: staff-only interaction zones, a photography ban, a reporting line.

The second contrarian point: Kato's remark that she was lucky never to have encountered a frightening or unpleasant situation is not reassurance, it is a warning. It attributes safety to chance rather than to protection. And there is a selection error to name: she is a top-1% performer whose experience happened to be clean, so generalising her account to the workforce would be a sampling mistake. There is no base rate for harassment here — one video, one vendor, one spectator. n equals one. Social-media heat is already far ahead of the evidentiary base, which is the classic overshoot condition.

The third, and nobody will write it: the dates do not reconcile. Applying in 2026, a three-year tenure, and a resignation in October 2026 makes four years. It may be a typo, a translation artefact, or a rounded tenure figure. Until the publication date and the tenure arithmetic are verified, the flag belongs on any external citation. This is not a marginal note: if the tenure count shifts, the entire assessment of experience accumulation and of the screening model shifts with it.

What was not measured needs saying out loud. The effect of fatigue on late-game service quality. Alcohol-service oversight — no cut-off rule is mentioned against a seven-or-eight-beer-per-head baseline. The employment classification of vendors, employee or contractor. And the pay rate itself, for which the source gives no figure. Without those, no labour-market conclusion can be drawn responsibly. That is the emptiest box in this audit.

What to watch next. A dated operator statement — and if it comes, whether it contains written words on photography and corridors or only an appeal to good manners. Second, whether the piece rate acquires a floor, because the economy of "tarugai" survives precisely in the delay before it does. Third, the next time such a clip appears, count the days before the operator responds; the lag is itself a number. I followed the small variables until the stadium showed me its instructions. The stadium has. The open question is whether anyone is reading them.