Early last month, Facebook users found themselves refreshing their browsers and apps to no avail. The social media giant had gone down, and Facebook’s outage lasted until service was restored almost six hours later. Not only was its flagship social network unavailable globally, but the outage also affected other popular apps in its portfolio like Instagram, WhatsApp, and Messenger.

After encountering this roadblock to their typical browsing behavior, what exactly did people do online while Facebook was down? We dug into our data from this period to understand how the absence of this top referrer affected the way readers found and consumed content, how this behavior differed from past service interruptions, and which social networks benefited most from Facebook’s outage.

A timeline of reader trends during the Facebook outage

• Prior to the outage, Social traffic represented about 18.8% of all traffic. Facebook accounted for 6.5% of that.
• A few hours into the outage, total traffic across our network was up 38% compared to the same time the previous week. At the same time, Social had fallen to just 7.6% of traffic.
• By 19:00 UTC (4 pm EST), traffic from Facebook alone had fallen to 0.3% and, as a result, hourly pageviews generated by the network declined by about 6.5M (-94%).
• Elsewhere, Direct traffic increased by 28%, Search was up 52%, and, as the proliferation of cheeky tweets indicated, Twitter traffic jumped by 72%.

Read more: Run your own social media experiments in 4 key steps

When Facebook loses, other channels win

Naturally, the question on everyone’s mind was whether the increase in traffic during the Facebook outage was due only to people seeking info about the outage itself or if there was more to the story. During the early hours of the outage, about 13% of pageviews occurred on stories about Facebook. At that same time, total traffic was up by 38%.

So, while some of the traffic increase was driven by Facebook-related content during the initial spike, it doesn’t account for the larger proportion of general readership, which remained higher than usual even as those searches leveled off.

Once readers understood that Facebook was offline with no real estimate for a return to service, they stayed online, and they continued reading other news.

In addition to the surge in traffic, it’s also worth noting that there was a clear winner in the referral traffic race during the outage. At 20:00 UTC (5 pm EST), traffic from Twitter was about 176% higher than the typical level for a Monday afternoon.

Past Facebook outages prompted a different reader response

In August 2018, Facebook was unavailable for a much shorter period of about 45 minutes. Even during that brief window, readers followed the trend we saw last month by quickly switching channels:

• Direct traffic to publishers’ websites increased 11%, while traffic to publishers’ mobile apps soared 22%.
• Search referral traffic to publishers was also up 8%.
• Surprisingly, there was a net new total traffic increase of 2.3%, meaning that the number of pages consumed across the web spiked upward in this timeframe.

What happens when the Facebook app is available, but there’s no news on the platform? We got a glimpse into this scenario in February 2021, when Facebook removed posting access for news outlets in Australia. In this case, readers who would normally have discovered content via Facebook did not switch channels — they just didn’t visit news sites.

By the afternoon of February 22 (about four days into the ban), total traffic from outside Australia decreased by 30% on average and 33% on average from within Australia, according to our analysis. While traffic from Facebook to Australian sites quickly rebounded after the ban was lifted, this data highlights just how closely traffic numbers are tied to third-party apps.

Analyzing Facebook’s impact on reader behavior

Readers have reacted in different ways to different kinds of Facebook interruptions. In a previous, shorter outage, readers quickly switched channels to consume news. During the Facebook ban in Australia earlier this year, readers stayed on the social network, and as a result, they consumed less news. During October’s downtime, readers followed the previous outage trend by not only switching channels, but by increasing total traffic by 38%.

This leads us to believe that when Facebook is operational, as in the case of the Australian ban, readers are apt to continue their normal social browsing behavior at the expense of visiting other publishers. Consuming content seems to be secondary to the act of browsing Facebook.

When Facebook is completely out of the picture, however, overall traffic increases, and other platforms like Twitter emerge as top referrers in its place. Based on this data, publishers appear well suited to adapt to a world without Facebook, but can they survive one in which Facebook declines to promote news?