Call of Duty: Modern Warfare 4 and the 93% Steam Rating: The Line Between Buyer Sentiment and Esports Health

For three consecutive nights after the campaign mode of Call of Duty: Modern Warfare 4 opened early to pre-order players, I sat in front of two monitors in a rented flat in Chaoyang District, Beijing, logging the number of new Steam reviews every 60 minutes. Colleagues in the analytics department call this the hobby of a man with too much time. I call it the only way to know whether a metric is rising because the product is good, or because of something that isn't in the product at all.

By the third night, the positive review rate settled at 93%. International media called it the franchise's best showing in 16 years, benchmarked against 2026. I read the headline, reopened my tracking sheet, and found three things that did not line up.

First: 93% and the Overwhelmingly Positive label do not sit in the same band. Valve's classification system sets Overwhelmingly Positive at 95% and above; the 80–94% band is Very Positive. A report citing both figures at once has either pulled them from two different moments or misread the label. For an analyst, that is a signal to verify the source before quoting it, not a signal to celebrate.

Second: the campaign opened early. Anyone writing a review inside that window had already paid in advance, had waited months, and was playing the exact portion of content this franchise has historically been strongest at. The sample is skewed toward the favourable.

Third: the 2026 benchmark. Call of Duty was absent from Steam between 2026 and 2026. When a data series breaks for four years, every "best since" claim becomes a comparison over a far narrower sample than the headline implies.

None of that makes 93% meaningless. It makes it narrower — and narrow is an important adjective in my line of work.

When the stands are empty, I can hear every single unit of budget. When a game launches, I can hear every single review too — but only when I know who wrote it, when they wrote it, and which part of the product they were playing.

What this metric measures, and whose interests it serves

Steam does not publish the weighting of its review system, but the structure is transparent. Each account that owns the game gets one vote. Valve aggregates those into a positive percentage and assigns a label against fixed thresholds. There is no weighting by hours played, no weighting by time of posting, and no filter by game mode.

Which means: a player with 12 hours of campaign and a player with 300 hours of multiplayer carry identical weight. For a title with multiple modes, that is a structural blind spot. For Call of Duty, the blind spot is amplified, because campaign and multiplayer serve different audiences, evaluate different things, and have entirely different lifecycles.

Based on my experience following CDL matches and the seasons I have spent handling club budgets, I always separate three layers of data when assessing a title.

Layer one is the product metric — review score, refund rate, average playtime. It measures buyer satisfaction in the first 30 days.

Layer two is the operational metric — concurrent players, 30-day and 90-day retention, the conversion rate from campaign to multiplayer. It measures whether the product survives its first quarter.

Layer three is the competitive metric — the weapon and map pool adopted into the competitive ruleset, meta stability, the number of teams entering qualifiers. It measures the health of the esports ecosystem.

These three layers do not substitute for one another. They are not tightly correlated. And in most launch coverage, all three get collapsed into one.

The 93% figure sits entirely in layer one. It says nothing about layer two, and even less about layer three. An ordinary reader is entitled not to distinguish between them. An investor, a sponsor, or a club executive is not — because they will use that number to decide where money goes.

There is an example I still use when training new analysts. In 2026, during the European Championship, I wrote a quick note on Leonardo Spinazzola built on 10 successful crosses into the box across his first four matches, against an average of 5 for wingers of comparable level. The note was shared more than 2,000 times. But if I had published the number 10 without stating the sample size of four matches, without stating opponent conditions, that number would have become a false assertion. Spinazzola does not take free kicks; he leaves the imprint of a new valuation rule — and a rule only holds value when it travels with its conditions of application. A Steam review works the same way: it does not score a goal, it only leaves the imprint of an emotional state at a single point in time.

The transmission line from a percentage to a league

The popular hypothesis in this industry runs: good game, more players, higher viewership, more sponsorship, healthier league. The chain sounds reasonable and almost always fails at the second link.

Call of Duty esports actually transmits through four tiers, each with a different decision-maker.

Tier one is the publisher. A well-received title strengthens their negotiating position with retailers, with distribution platforms, and with brand-level sponsors. This is the only tier where 93% has a direct effect, and the effect is short-term.

Tier two is the league system. Whether a new title becomes the official competitive title of the Call of Duty League is an operational decision, not an automatic consequence of unit sales. It depends on the map pool, the game modes, the movement mechanics, and above all on whether the publisher wants to bet on a fresh meta cycle.

Tier three is teams and players. This is where the risk concentrates. A new meta cycle means the accumulated skill of a professional can lose value inside a single season. Players specialising in an older movement mechanic suddenly become less competitive — not because they got worse, but because the yardstick changed.

Tier four is viewers and sponsors. This tier reacts slowest, typically one to two quarters behind tier two.

A Steam rating only touches tier one. It does not automatically touch tier two. And the gap between tier one and tier two is precisely where bad investment decisions are manufactured.

One technical detail rarely mentioned: Call of Duty esports runs predominantly on console and controller, while Steam is a PC platform. These are two different communities, two different habits, two different content ecosystems. A PC player reviewing a campaign on Steam tells you nothing about a controller professional preparing for the next season. This is a channel mismatch, not a quality mismatch — and channel mismatch is the hardest error to detect, because it produces no contradiction inside the data.

The valuation lesson I paid for in real money

In 2026, at 25, I started doing financial analysis for Beijing Guoan. During the summer window I proposed spending 12 million euros on midfielder Jonathan Viera, based on key pass and expected assist figures from La Liga. I built a beautiful model. That model omitted a single variable: adaptability to the Chinese football environment. Six months later the club sold him for 8 million euros. A 4 million loss. In a closed-door meeting the head coach said it to my face: numbers cannot replace direct observation.

The market does not forgive, it only records — and I paid for that with the 2026-18 season.

That mistake taught me a principle I now apply to every metric, including game review scores: before using a number, establish who generated it, under what conditions, and how it loses validity when those conditions change.

Applied to 93%: the number was generated by early buyers, mostly playing campaign. The condition was the launch window, before full multiplayer and before monetization was exposed. The number loses validity the moment those two conditions shift — and the original author conceded exactly that, noting the score may change once the full version launches and player counts rise.

That is an honest concession, and it sits in the body text, not in the headline.

There was another occasion where I was wrong in the opposite direction. In January 2026, an acquaintance inside the City Football Group system asked me whether I could believe a 21 million euro price for Julian Alvarez. I reviewed six months of statistics: 14 goals, 6 assists in Argentina, a low true tackle figure. I concluded the risk was too high. In 2026-23, Alvarez scored 17 Premier League goals. I was wrong. I learned valuation from one mistake, and never needed a second lesson — but I did need a third, and that lesson was this: data can mislead in both directions, the direction of overvaluation and the direction of undervaluation.

With Modern Warfare 4, the risk sits in the first direction. A high metric inside the launch window creates pressure to conclude early. And concluding early in this industry is always expensive.

The arithmetic nobody runs

Here is a simple calculation anyone managing a budget should run before quoting 93%.

Steam keeps 30% of revenue at the standard tier, dropping to 25% once a title passes 10 million USD in revenue, and 20% above 50 million USD. For a AAA title from a major franchise, it is almost certainly at the 20% tier within weeks.

In Vietnam, a standard-edition Call of Duty typically prices between 1.3 and 1.7 million dong under regional pricing. After the platform's share and applicable taxes, the publisher's net lands somewhere around 60–65% of the listed price — under 1.1 million dong per unit.

Set against that the operating cost of an esports season: venue hire, broadcast production, player travel, operational staff salaries. A regional-scale event carries an operating budget equivalent to tens of thousands of game units sold.

My point is not to compare two figures in different units. My point is this: game sales revenue and esports ecosystem revenue are two different cash flows, flowing into two different pockets, decided by two different departments. A title selling 20 million copies does not automatically produce a profitable league. It produces a large enough pool of interest for someone to attempt building one.

The distance between "large enough to attempt" and "large enough to profit" is the entire job of people like me.

Where Vietnam sits in this picture

I have followed the Vietnamese market long enough to know one thing: the picture here is drawn by different titles. Arena of Valor, Free Fire, PUBG Mobile, CS2, Dota 2, Valorant, League of Legends. Call of Duty has never been a main current in Vietnam's competitive ecosystem — no large-scale domestic circuit, no widely sold CDL broadcast rights, and Southeast Asian servers still hosted abroad, pushing latency above what a professional will accept.

Which means for a Vietnamese operator, 93% on Steam carries close to zero informational value for an investment decision. It is a datapoint about the global PC market, not about the Vietnamese esports market.

But it carries a different value, and that value is real: it is an exercise in reading a metric. Vietnamese esports is at a stage where a great many decisions are made on a feeling about buzz rather than on the structure of cash flow. A small lesson from a misread percentage can save a club several hundred million dong of sponsorship directed at the wrong place.

A tight budget does not produce poverty; it produces sharpness. The Vietnamese clubs that survive difficult seasons are the ones that learned to read metrics before signing contracts, not after those contracts collapsed.

One cultural detail is worth noting. Most PC players in Vietnam access games through internet cafés, where a personal Steam account is not the primary tool. A player logging 200 hours in a café may never have left a single review. Which means the Steam review sample is already skewed toward players with personal accounts, international payment cards, and a habit of writing reviews. That is a much narrower group than the actual player base — and in Vietnam, the gap between those two groups is larger than in most other markets.

The contrarian angle: a high score is a weak signal

This is the part I consider most important, and also the part most likely to irritate.

In most cases, a high review rate inside the launch window is a weaker signal than an average one. The reasoning is specific.

The launch window has three structural features: early buyers are a self-selected group who already believed in the brand enough to pay in advance; the content being reviewed is the most tightly controlled content in the product's entire lifecycle; and the problems around balance, monetization and servers have not yet been exposed. Together, those three features manufacture an artificially high floor.

An average review rate inside the launch window, by contrast, is a strong signal, because it shows the product hit trouble even when every condition was favourable.

In the other direction, I equally distrust the argument that "a low score means esports will die." There are seasons where a title heavily criticised by the multiplayer community produced the most interesting competitive meta, because the publisher was forced to intervene, and intervention patches tend to generate volatility — and volatility is the raw material of sport.

What I actually track is not the score. It is three other things.

One: the 90-day retention curve. If concurrent players hold above 30% of peak after three months, the product has a floor. If it falls below 15%, every launch metric was noise.

Two: the campaign-to-multiplayer conversion rate. This is the least published and most valuable metric of all. A title with a strong campaign and a dead multiplayer produces a large buyer pool and a small competitive community — the worst possible configuration for esports.

Three: the official competitive ruleset announcement. A title becoming the Call of Duty League's competitive title is a decision with an owner and a date. No speculation required. Just wait for the document.

And here is the point I want to press: the biggest risk in this story is not that Modern Warfare 4 is bad. The risk is that it is good just enough to trigger an investment wave built on an unrelated metric. As the "best in 16 years" headline spreads, it will be read by people who do not distinguish campaign from multiplayer, PC from console, game sales revenue from ecosystem revenue. And some of them will sign sponsorship contracts.

There is a precedent I keep returning to. In March 2026, when every competition in China was suspended, I was working at Shanghai SIPG in a mid-level role. I proposed cutting 35% of non-essential operating costs — cancelling the private bus lease, renegotiating the data analytics contract with the provider. The plan saved 2.3 million RMB in the second quarter, enough to retain two Brazilian assistant coaches who had initially been on the exit list. For two weeks I worked 18-hour days, building a contingency plan detailed down to individual line items.

The lesson was not "crisis creates opportunity." The lesson was: when everything is euphoric, nobody builds a spreadsheet. When everything collapses, everybody builds one — and by then it is too late to have options.

What to watch in the next 60 days

I am not forecasting whether Modern Warfare 4 succeeds or fails. I am offering four checkpoints, because checkpoints are worth more than forecasts.

First, track the review rate once the full build opens. If the rate slides below 85% in the first two weeks of the multiplayer phase, the "return to form" story has passed its expiry date. If it holds above 88%, the product has a real floor.

Second, read the negative reviews, not the positive ones. Positive reviews inside the launch window carry almost no information, because expectation dominates them. Negative reviews are always specific: a movement mechanic, a weapon, a map, a price.

Third, verify the title name and the timeline. The Modern Warfare line already has 2026, 2026 and 2026 entries with non-linear naming. Any comparison against "2026" needs independent cross-checking against the Steam release history before being quoted. And remember that between 2026 and 2026 this franchise was not on Steam at all, so the historical series is broken.

Fourth, wait for the official competitive ruleset document. Any meta analysis before that point is a probabilistic guess.

There is one small detail I kept from that third night. When the positive rate crossed 90%, I opened the comment section of the single remaining negative review and read all of it. The author complained about a technical detail in the campaign. Beneath it, seventeen people replied, mostly in objection. None of them argued about the detail. They argued about whether a negative review should have been written at all.

A community in that state is not yet a competitive community. It is a community defending something it just bought. Those two states can convert into one another, but it takes time and it takes a multiplayer product that holds up.

Which is why I keep my conclusion unchanged: 93% is a real datapoint, narrow, and with a short shelf life. Its value lies in forcing us to state clearly what we are measuring. In an industry where most decisions are made on a feeling about buzz, stating clearly what you are measuring is already half of the competitive advantage.

The other half is knowing when to stay quiet and wait for the second datapoint. With Modern Warfare 4, that second datapoint has not arrived. It will arrive when multiplayer opens fully, when monetization is exposed, and when the first professional player says publicly that he cannot play this title at a competitive level.

That is the day worth writing about.