Credit George Etheredge for The New York Times
Facebook is moving to restore confidence among advertisers and publishers that it is accurately measuring how often and how long users see videos, news articles, ads and corporate messages on its social network.
The steps, announced on Wednesday, followed the company’s apology in September for an error that resulted in considerably overstating how long, on average, users spent watching videos on Facebook.
Facebook said that it would increase its use of outside measurement services for display and video ads. The company said it is expanding its use of metrics services like comScore, Moat, Nielsen and Integral Ad Science, and looking to add others.
Facebook said it would form a “measurement council” of ad agency executives and marketers to develop metrics suited to the needs of advertisers.
To communicate more openly, Facebook established a blog, Metrics FYI, to post regular updates on the metrics it uses and what its analysts have found.
“This is an important evolution for Facebook,” said Randall Rothenberg, chief executive of the Interactive Advertising Bureau, a trade group of media and technology companies, including Facebook. “This is Facebook saying, Yes, we know that we need to be and are getting more involved in advertising transparency.”
In the first blog post, Facebook, after a lengthy internal audit, said it had uncovered a few more “bugs.” One such miscalculation involved double counting of repeat visitors to view what are known as marketers’ organic messages — regular Facebook posts by companies rather than paid ads, the post said. Correcting that problem, it said, would have reduced the measurement of the number of unique users seeing the marketers’ messages by 33 percent on average over a seven-day period, and by 55 percent over the 28-day span.
The Facebook data analysts also found that the time spent on what are known as instant articles — fast-loading news items tailored for easier reading on smartphones — was overstated on average by 7 to 8 percent, the blog said.
But another bug resulted in undercounting a video metric. The Facebook research found that the number of people viewing video ads from beginning to end was 35 percent higher than it had previously calculated.
Facebook emphasized that that newly uncovered flaws involve only a few of the more than 220 metrics Facebook deploys.
The impact of the disclosure of further calculation miscues, after the September apology, is uncertain. Advertisers and publishers have little choice but to embrace Facebook given its huge audience and its dominance, with Google, of online advertising. Early this month, Facebook reported that its third-quarter profits surged 166 percent from a year ago to nearly $ 2.4 billion.
But analysts said concerns about the accuracy of Facebook’s metrics could prompt some to question whether they should invest so much time, money and resources with the company.
“It’s not like you can not use Facebook as an advertiser or as a publisher,” said Brian Wieser, a media industry analyst at Pivotal Research. “But it is a question of who do you prioritize and how many resources do you send that way?”
The moves Facebook has taken are intended to quell such doubts and restore its credibility among advertisers and publishers.
“Our goal going forward,” Facebook said in its web post, “is to communicate more regularly about our metrics, so that our partners can focus on doing what they do best — servicing their customers — with the best insights possible.”
Yet a more skeptical view of digital advertising is emerging, not only for reasons of measurement but some major advertisers wonder if it is as effective as they had assumed.
“The ascendant view had been ‘if it’s digital, it’s better,’ ” Mr. Wieser said. “And I think with every passing incident, those who have not been appropriately skeptical or appropriately critical have come down to earth a bit.”