UPDATED: Facebook employees and top execs, including chief operating officer Sheryl Sandberg, were fully aware the social giant was overestimating the projected advertising reach it told marketers — and failed to disclose it to clients for years, according to a newly unsealed legal filing.

“Facebook knew for years its Potential Reach [metric] was misleading, and concealed that fact to preserve its own bottom line,” according to the latest filing in the case, released on Feb. 17.

In the fall of 2017, according to the filing, Sandberg “acknowledged in an internal email she had known about problems with Potential Reach for years.” Per the filing, the Facebook product manager for Potential Reach proposed a fix that would have lowered the estimates, but “Facebook’s metrics leadership team rejected his proposal because the ‘revenue impact’ for Facebook would be ‘significant.'”

The filing cites a statement by the Potential Reach product manager that “it’s revenue we should have never made given the fact it’s based on wrong data.” In addition, another employee said that “[t]he status quo in ad Reach estimation and reporting is deeply wrong” and that the only question was, “[h]ow long can we get away with the reach overestimation.”

Reached for comment, a Facebook spokeswoman told Variety, “These allegations are without merit and we will defend ourselves vigorously.” Later, the Facebook rep sent an additional statement: “These documents are being cherry-picked to fit the plaintiff’s narrative. ‘Potential reach’ is a helpful campaign planning tool that advertisers are never billed on. It’s an estimate and we make clear how it’s calculated in our ads interface and Help Center.”

The case, originally filed in August 2018, seeks class-action status. The plaintiffs allege that they bought advertising on Facebook and Instagram based on the false data. The latest filing is available at this link; the case is DZ Reserve et al. vs. Facebook in the U.S. District Court for Northern California in San Francisco.

Facebook has argued that the Potential Reach figure does not affect billing or actual delivery of ads. According to the company, the metric had been based on user-reported information about those who had seen an ad and was never portrayed as census data. In 2019, Facebook said it made changes to the reach estimates, tallying the figure based on people who had actually seen an ad in the last month (from users who were “eligible” to see an ad).

“Over nearly two years of litigation, Facebook has told this Court its Potential Reach metric is ‘simply a free tool’ advertisers can use, whether or not they buy an ad,” the plaintiff’s filing says. However, the plaintiffs continue, “The facts tell a different story. Behind closed doors, Facebook stated: ‘When creating advertising campaigns, advertisers frequently rely on the estimated audience to understand the potential reach of their campaigns and set the bid and budget strategy. Thus, this number is arguably the single most important number in our ads creation interfaces.’”

Facebook knew the problem was largely due to fake and duplicate accounts, but “the company made a ‘deliberate decision’ not to remove duplicate or fake accounts from Potential Reach. And senior executives blocked employees from fixing the problem, because it believed the “revenue impact [would be] significant,” according to the filing.

Separately, Facebook previously admitted it overstating video-viewing metrics over an 18-month period in 2015 and 2016, claiming that it made an honest mistake in calculating the figures. In 2019, the company settled a class-action lawsuit over that issue, agreeing to pay $40 million to advertisers claiming they overpaid for Facebook ads because of the inaccurate data.