2012 was a year of significant changes to Facebook’s sharing policy. Initiatives such as Open Graph Publishing and Seamless Sharing raised concerns from the public and the media that Facebook users would resist the increase in automated sharing on their behalf, particularly around authorizing Facebook Connect with third party applications.
At Sociable Labs, we last studied Facebook Permissions authorizations in December 2011, before Facebook’s most recent raft of changes, and we thought it would be insightful to revisit the question a little over a year later. The question we examined is “What percentage of users will accept a Facebook Permissions authorization prompt and how has that figure changed in the last year?” Our research indicates that there has been no meaningful change in authorization rates in the last year, with an average rate across our customers of just under 60%.
Data for the study was collected with Facebook Insights using all Sociable Labs customers that had deployments of Sociable applications on their websites in both 2011 and 2012. We compared accept rates during the one-month period from 19 November to 18 December of both 2011 and 2012 to determine if there had been a meaningful change in the rate at which users authorize Facebook Permissions prompts. Averages were calculated as the ratio of total accepts to total impressions, summed across all customers.
Additionally, we calculated acceptance rates for all Sociable Labs customers for whom data was available during the last 12 months, again using data collected with Facebook Insights. For this same period, we also calculated the acceptance rate when the authorization was being used for registration only.
During a one month period spanning November and December, there was no significant change in Facebook Permissions acceptance rate between 2011 and 2012. The 2011 accept rate was 55.6% and the 2012 rate was 56.6%. Total impressions during these periods was just under 400,000.
Overall acceptance rate for the entire 12-month period from December 2011 to December 2012 for all Sociable Labs customers for whom data was available is marginally higher at 59.4% of nearly 3.8 million permission request impressions.
Permissions for Registration Only
When the authorization was requested for site registration only, as employed for “Login with Facebook”, the accept rate rose to 68.8%. The actual accept rates are even higher when discounting one outlying customer that had an abnormally low acceptance rate but accounted for a significant portion of the data (30.8% of all impressions for the period). Therefore, the nonweighted average of all customer acceptance rates, 75.7%, is a more realistic picture of the acceptance rate when permissions are requested solely for the purpose of using Facebook for site registration.
The results show that despite growing concerns about privacy on Facebook, particularly as sharing becomes more automated, users are continuing to authorize Facebook Permissions at high rates. Compared to one year ago, the current rates are about a percent higher during a one month period, and over the last year in its entirety, authorization rates are nearly 60%, and more like 75% if the permissions are requested only those required for the purpose of “Login with Facebook” in lieu of creating a site password.
This finding is promising because it indicates that the fundamental premise of social commerce, that users are willing to share personal social data with online retailers, is a) sound and b) relatively inelastic. That is, even in the face of controversy surrounding Facebook’s evolution of how social information is shared, users seem unfazed and are still willing to authorize permissions. It is worth noting that, as always, good user flows, application design, and a clear benefit to the user will incentivize them to authorize permissions.
Today we released a research study that we believe will be very helpful for any online retailer who is considering implementing Log In with Facebook. In particular, we reviewed how the top 500 online retailers integrated Log in with Facebook as an alternative option for their account creation, site registration, and site login processes. Sociable Labs’ study of Log in with Facebook usage by the IR 500 was restricted to this use case only — simply whether Log in with Facebook is integrated with the core e-commerce site for user account creation. Sociable Labs did not study other use cases for Facebook Platform, such as Social Plugins (i.e. Like buttons) or other onsite social applications that utilize Log in with Facebook.
Sociable Labs found that adoption of Log in with Facebook by the IR 500 is still in its beginning stages with 30 of the top 500 online retailers having implemented Log in with Facebook into their account creation process today. This remains an untapped opportunity area for most of the IR 500, who have yet to take advantage of the many benefits that Log in with Facebook affords: simplified account creation, elimination of required site-specific passwords, increased onsite personalization, richer CRM data, and increased referral traffic from Facebook. Sociable Labs believes that the major factors hindering adoption include technical barriers to implementation, perceived security concerns, and level of effort required for implementation. Many of these barriers will be removed over time, which should spur continued adoption of Log in with Facebook by the IR 500.
Facebook is by far the dominant Social Login option chosen by these online retailers. All IR 500 sites that have implemented a Social Login solution include Log in with Facebook. Furthermore, 77% of these sites only offer Log in with Facebook. Sociable Labs believes that online retailers will continue to prioritize Log in with Facebook over other forms of Social Login (Twitter, Yahoo, Google, etc.) as Facebook continues to advance its offerings around Facebook Platform.
The majority of IR 500 sites that offer forms of Social Login other than Facebook have done so by utilizing a third-party software vendor such as Gigya or Janrain. However, only 17% of the IR 500 that have implemented Social Login have chosen to do so via a vendor. The majority of the IR 500 that have implemented Log in with Facebook have done so independent of third-party software, which is why Sociable Labs believes Facebook will remain the primary, and only, social login option offered by most online retailers.
Sociable Labs also found that a large percentage of IR 500 sites that have implemented Log in with Facebook have done so incorrectly, or are not leveraging best practices. A full 30% of those who have implemented Log in with Facebook have done it incorrectly by requiring the user to create a site password in tandem with Log in with Facebook. Exactly half of those who have implemented Log in with Facebook have yet to offer it during the checkout process, missing a big opportunity to generate more usage and provide a consistent user experience. None of the IR 500 that have implemented Log in with Facebook are taking advantage of one of the primary benefits of Log in with Facebook — Persistent Login—to automatically login users on return visits.
You can download the full study here: http://www.sociablelabs.com/download-login-study
There has been a lot talk in the blogosphere about reported declines in utilization of so-called "Social Reader" Facebook applications, like those from the Guardian and the Washington Post. The dip in active users has led to a lot of conjecture about OpenGraph publishing being a passing fad and one that consumers are outright rejecting. Much of this conjecture is absolutely false, while some is just misleading. This is how we see the facts:
1. Reports of broad consumer rejection of Facebook seamless sharing are absolutely false. More and more applications are growing and thriving using seamless sharing (aka OpenGraph Publishing), such as SocialCam, which grew to 40M active FB users in the course of two weeks (http://www.appdata.com/apps/facebook/150768931647055-socialcam
2. Some Social Readers that are Facebook canvas apps, have seen declines over the last month. However, websites using seamless sharing, such as Yahoo News, have not seen any significant decline (http://www.appdata.com/apps/facebook/194699337231859-yahoo-social-bar
). We have confirmed this directly with many companies using OpenGraph publishing from their websites. The below graph for Yahoo shows that daily monthly active usage is still in the 40M zone, ranging between 38M to 42M.
It is important to draw a distinction between canvas apps and onsite integration. Onsite integration has a continuous source of organic traffic already, and hence is buffered from changes to Facebook newsfeed algorithms. Canvas apps like Washington Post Social Reader are going to be more affected by these types of changes given their sole reliance on Facebook traffic.
3. The drop-off in usage for social readers can be attributed directly to changes to the Facebook newsfeed algorithm. The drop-off starting April 10 looks dramatic because of a substantial rise in activity end of March and early April. Looking at the graphs holistically over a longer time horizon shows the effects of the Facebook newsfeed changes as rather minimal, relative to the total usage. This graph from Techcrunch and AppData, show that the Guardian's daily active user count went from 200,000 to 150,000, a 25% drop, but much more reasonable than a drop from 600,000 to 150,000, if you don't count the temporary peak:
It has been our experience that Facebook continuously fine tunes the algorithms, but the amount of referral traffic remains remarkably consistent over time. We believe that Facebook will strike the appropriate balance.
Thus, our conclusions are as follows:
1. Companies evaluating seamless sharing initiatives should recognize that the overall business benefit is still substantial, despite recent gyrations to the Facebook newsfeed algorithm.
2. Consumer usage is still strong as ever.
3. Companies that choose only to deploy Facebook canvas applications should be cognizant of the risks of having Facebook as the sole source of traffic to the application, which is why onsite integration is the preferred choice. Onsite integration yields a more consistent source of traffic using the OpenGraph APIs, and hence is less dependent on the Facebook newsfeed as the sole source of usage.