Had to read down quite a ways to find the inevitable bullshit in this article.

After applying some complex statistical models to the underlying data, Volckmann finds that, unsurprisingly, relative revenues in the weeks following a crack’s release are lower than the baseline expectation for uncracked games in the same time period. These negative effects of a crack on revenues—which are highly statistically significant (p<0.01)—“impl[ies] that the appearance of a crack reduces revenue relative to the no-crack counterfactual,” Volckmann writes.

I have no doubt of that.

However, the headline clearly implies that that loss is overall or on average 20%. And that’s bullshit.

Just how much money a publisher can expect to lose from a Denuvo crack, though, depends heavily on how quickly the game is cracked, Volckmann finds. A Denuvo-protected game cracked in the first week after release can expect to make about 20 percent less revenue than if the DRM had remained in place, according to the study, while a crack six weeks after a game’s release only costs an estimated 5 percent of theoretical total revenue. After 12 weeks, new sales are so negligible that "developers could eventually remove unpopular DRM schemes with minimal losses

So 20% is actually the maximum, and is only the case if a game is cracked within the first week after release.

Six weeks after release, there’s only approximately 5% difference, and 12 weeks after release, there is no significant difference.

So the headline, unfortunately predictably, is misleading at best.