A Misapplied 'Esports' Label: Reading Gacha Monetisation Architecture With a Spreadsheet

At ten in the evening I opened a file labelled “Esports” in our internal archive. My spreadsheet was already open with its four familiar columns: version, roster, schedule, performance data. I started filling it in, as I have done several hundred times before. Three minutes later all four columns were still empty.

There were no matches inside. No teams, no players, no standings, no fixtures. Only a banner list for Genshin Impact: phase two of version 7.0, then phase one and phase two of version 7.1. Names such as Vesna, Vodyanitsa, Flins, Ineffa, Skirk, Escoffier. A pity mechanic, a 50/50 rule, and a note admitting the exact banner schedule was still to be confirmed.

I closed the laptop for a moment. When home advantage stops being home advantage, I am forced to rewrite every assumption. This time the stadium had not disappeared. My analytical frame was the thing that did not fit.

The document was a stage-two deep analysis of an article about the Genshin Impact banner schedule. It contained 28 information points. Twenty carried no source. One cited an official announcement from the publisher. Three were the author’s opinion. The rest was scheduling, mechanics and speculation about the future.

Every dataset is a scripture, and I am a slow reader. I read this one out of that habit, even though the subject sits outside the pitch. Genshin Impact is published by HoYoverse, an open-world role-playing game monetised through gacha: players spend premium currency for a randomised chance at characters or weapons. The title has no professional tournament circuit, no clubs, no transfer mechanism, no competitive balance patches. What it has is a revenue engine running on a version cycle.

And yet the document carried the label “Esports”. Feed that label straight into an esports model and a chain of false conclusions generates itself: characters read as players, banner phases read as tournaments, character power read as competitive form. I refused to do that, because a model is only worth something when its assumptions of similarity are tested first. But once the wrong label is peeled off, what remains is still worth reading: a monetisation architecture designed with real care.

The operating rhythm splits each version into two phases of roughly 21 days, each with its own banner. Seen through a data lens, this is a time-boxed purchase window. Players do not spend in a vacuum; they spend inside a framed period with an opening and a closing date. In football, the closest mechanism is the transfer window. A player is still there in March, but pressure only appears when the countdown clock runs. Compressing time is a way of manufacturing decisions.

The second layer is pity: a five-star character is guaranteed within 90 pulls. Many read this as generosity. I read it differently. Pity is a soft price floor. It turns a random outlay into an estimable one, and when players can estimate, they plan; when they plan, they spend more than when they are blind. A release clause in football works on the same logic: a concrete figure does not make a deal easier, it makes the deal easier to calculate, and whatever is easy to calculate gets done more often.

The third layer is the 50/50 structure. The first five-star on an event banner has a 50% chance of being the limited character and a 50% chance of being a standard-pool character; hit the standard pool and the next five-star is guaranteed to be the limited one. Look at the mean and the system is cheaper than it appears. Look at the variance and it is far more expensive. Every industry lives on variance, not on averages. A team with a 60% win probability still loses 40% of the time, and that 40% funds the entire betting market. The 50/50 structure does the same thing with a coin flip.

The fourth layer is shared pity across banners of the same type. When two same-type banners share accumulated pull counts, players lose no progress when they switch direction. The marginal cost of switching falls, and when marginal cost falls, spending frequency rises. This is a revenue-smoothing mechanism between banner types, not a concession.

The fifth layer is the absence of a fixed rerun schedule. Some characters are absent for more than a year; others return after only a few versions. If this were a tournament, people would call it an arbitrary fixture list and demand the organiser explain itself. But this is a market, and undated scarcity is the product. A star returning from injury only sells out the stadium when the return date is not known too far in advance. Know the month for certain and nobody queues early.

The sixth layer is the Chronicled Wish, a separate banner type for older characters with its own rules. It can be read as a secondary lane that re-monetises legacy characters without disturbing primary banner cadence. In sport, this is the memory-driven business: testimonial matches, anniversary shirts, products that do not need a strong current squad to sell.

The interesting pressure point sits elsewhere: phase two of version 7.0 was a rerun, while phase one of version 7.1 introduced two new characters at once. The peak currency-allocation strain falls on phase one, and the group that loses out is the one that already spent on the earlier rerun. No power data accompanies the ordering to justify it. That is an observation about cash-flow architecture, not about character strength.

The biggest gap sits somewhere else again. The document describes timing, not power. There is no kit analysis, no stat comparison, no basis for the advice to save resources for version 7.1. That advice is a calendar statement presented as a value judgement. A player’s value is only a number — until you find the error in how it was calculated. Here there is not even a number to read.

When 20 of 28 information points carry no source, and several names and version numbers cannot be cross-checked against known game state, the problem stops being whether the prediction is right or wrong. The problem is that there is nothing to verify. In my work, a model that is 80% right and delivered on time is still more useful than a perfect model delivered after the match. Here, not even the 80% model exists.

Based on my experience watching matches, I always separate observation from inference before writing a single line. In this document, observation accounts for about a third: dates, rules and milestones. The remaining two thirds is inference presented as fact.

Scarcity and schedule ambiguity are usually treated as two separate flaws. They are one design. An unfixed schedule creates scarcity; scarcity creates fear of missing out; fear of missing out creates fast decisions. Such a system simultaneously retains the planners and harvests the non-planners. That is why I do not file uncertainty under operational risk. I file it under features.

The correlation between banner order and character strength is zero. Coming first does not mean weaker; coming later does not mean stronger. Yet readers still translate “coming soon” into “worth saving for”. It is a familiar interpretive error, and it shows up identically in football: transfer data models inflate the potential of young players because their metrics are measurable, while what stays unmeasured is dressing-room chemistry. A handsome highlight reel says nothing about whether that player will run at minute 80 when his team is a goal down. In this document there is no highlight reel at all, only dates.

The part worth keeping is the power structure: the publisher writes the rules, publishes the rules and collects the money. No independent third party confirms that real rates match disclosed rates. In football we have VAR, and I still say it plainly: VAR does not reduce controversy; it moves controversy from the pitch into the review room and into the grey zones of the law. But at least there is a review room. Here there is no equivalent.

Three signals will decide whether this document retains any reference value. The official announcement from HoYoverse naming the characters for each phase of version 7.1, the only point in the document with an official source. The appearance of the names Odette, Flins, Ineffa, Vesna and Vodyanitsa in official materials; if they do not appear, the entire schedule section must be struck from the reference set. And regulations on probability transparency in games with randomised rewards, because that is the only variable capable of changing this architecture from the outside.

I do not predict the future with intuition; I only read the traces the data leaves behind. The problem with this document is that no trace yet exists to read. For anyone patient enough to wait a season to prove a number — this time, the season begins by verifying a name.