GTA6’s 25m forecast needs a 22x ramp

On 13 October Sensor Tower’s pre-order tracker put Grand Theft Auto VI at 6.08 million copies and $594 million, split 4.51 million and $441 million on PlayStation against 1.57 million and $153 million on Xbox. Those are estimates, the tracker covers the two console storefronts only and is net of returns, and Take-Two has published nothing. The figure that has travelled with them since is a forecast: Karl Kontus of Sensor Tower told GameSpot that pre-orders could land “in the area of 25 million” by launch, and that at 25 million the revenue would work out to “about $2 billion”, enough to cover the development budget and most of the marketing. That pair of numbers is now being repeated in English and in Chinese coverage, and it is the part of the story nobody has divided. Take-Two’s own position has not changed: Strauss Zelnick has said raw pre-order data “could potentially be misleading”, and that the company has not sold a single unit because a pre-order can be cancelled.

The first division is the book’s own unit price. $594 million across 6.08 million copies is $97.70; PlayStation works out at $97.78 and Xbox at $97.45, a spread of 33 cents. This is arithmetic, not a disclosure, and it assumes the tracker’s revenue is storefront gross at list prices, which is the reading the totals support. Two things follow. The first is that the average has not moved at all: on 8 September the same tracker showed 5.28 million units and $516 million, or $97.73. Five weeks and 800,000 copies later the unit revenue differs by three cents. The second is what that level implies — an arsenal weighted almost entirely to the expensive edition. Nine in ten buyers at $99.99 and one in ten at $79.99 produces $97.79, which is where the book sits. The mix reported in August, when Sensor Tower put the Ultimate share near 89%, has therefore not shifted in any way the revenue can detect.

The forecast breaks against that. The same interview has Sensor Tower expecting the edition split to even out to roughly 50/50 between the $79.99 and $99.99 versions by launch, on the theory that early buyers are enthusiasts and the late wave is ordinary. At that mix the average falls to $89.99, and 25 million copies produce $2.25 billion. At the book’s current $97.70 they produce $2.44 billion. The $2 billion being quoted is 25 million multiplied by exactly $80.00 — a price that requires the Ultimate edition to sell essentially no copies at all. The two figures cannot both hold: either the unit forecast is nearer 20.5 million at the price the book is actually clearing, or the revenue is $2.25 billion to $2.44 billion. The most likely explanation is not an error of multiplication but a collision of subjects, because $2 billion is also the development-and-marketing figure that has been in circulation all year and that Take-Two has declined to confirm. That reading is inference, and it is offered as such.

The second division is the pace. From 6.08 million on 13 October to 25 million on 19 November is 18.92 million copies in 37 days, or 511,351 a day. The book’s own slope is 800,000 copies over the 35 days from 8 September to 13 October, which is 22,857 a day; the forecast therefore requires 22.4 times the rate the tracker has been recording. The campaign has produced exactly one documented step change: on 27 August, the day of An Extended Look, daily pre-orders ran at more than five times the previous day’s and the total jumped to about 4.86 million. Hold five times the current rate for all 37 remaining days and the book lands near 10.3 million, which is 41% of the forecast. Kontus’s accompanying expectation — that the majority of pre-orders arrive in the last week or so — is harder still: 13 million copies in seven days is 1.86 million a day, 81 times the current pace. One caveat runs against the forecast rather than for it. Sensor Tower changed its model on 5 October to reflect the split by country and says the change moved units, revenue and the console mix, which means an unknown part of those 800,000 copies is a re-based ruler rather than new buyers, and the ramp required is steeper than the raw arithmetic shows.

The comparison the coverage reaches for has the same problem in miniature. Sensor Tower puts Grand Theft Auto V at about 7 million pre-orders in 2013, and the line that has circulated is that VI could pass it before launch. It needs 920,000 more copies in 37 days, or 24,865 a day — 9% above the observed slope, which is close enough to be plausible and far enough to be worth stating as a requirement rather than a likelihood. At the flat 22,857 a day the seventh million arrives 40 days after 13 October, which is 22 November: three days after the game is on sale, and therefore a record set by launch sales rather than by pre-orders.

What survives all of this is smaller than the headlines and firmer. The book is $594 million from 6.08 million digital pre-orders, net of returns, excluding retail codes and every market the tracker does not model. The forecast is a vendor’s model, and as published it does not divide by its own unit price. Nothing in either number is a company statement, and nothing in either changes what a buyer does: the two editions remain $79.99 and $99.99, digital pre-load opens on 12 November, the unlock is per store region on 19 November, and the retail box holds a code rather than a disc. The next opportunity for a real number is 9 November, when Take-Two reports its second-quarter results — the last scheduled disclosure before launch, and the first at which the adjectives could be replaced by a figure.