As the social media platform grows and the industry changes over time, the details of Facebook analytics has come under heavy scrutiny. In response to this criticism, Facebook has confirmed that it will be making several corrections to its analytical metrics in its effort to improve accuracy and reduce errors and misleading metrics. Any unresolved issues will remain in the company's pipeline to be corrected as soon as possible.
Third-party viewing for better Facebook Analytics transparency.
Facebook will also be responding to user feedback regarding the need for greater transparency and will be partnering with third-parties to develop independent verification options. A Measurement Council comprising members from other businesses and measurement agencies will be formed to gather expert insight from other areas of the industry for the purpose of reviewing and advising on future metrics changes.
In addition, Facebook will also be setting up a specialized blog that will detail all updates or changes made to its social media platform.
Incorporation of third-party measurement features.
As part of its collaborative efforts with third-party measurement developers, Moat and Integral Ad Science, Facebook will enable third-party measurement and verification for display ads and video ads. This will allow all platforms, including website link ads and canvas ads, to be viewed and scrutinized using independent analytical tools.
Organic counts will also be analyzed using independent measurement tools. Under the new changes, Organic and Paid impressions will only be included in metric calculations if the post or ad actually shows up on a user’s screen. Facebook predicts that the changes will cause reported Organic Reach to drop approximately 20%. Users will be notified of the change via the Facebook interface and in the new blog.
Also, publishers will now be able to rely on third-party verification for video metrics through Nielsen’s Digital Content Ratings (DCR).
Facebook page video watch time (Corrected).
The average viewing time of Facebook page videos was being over-reported by 60 to 80 percent. This was because the metrics was ignoring views that were below 3 seconds.
Facebook has implemented two replacement metrics to address the issue. The first one is the original metric where the total watch time is divided by the number of times the video was displayed automatically or clicking.
The second new metric is the “unique average watch time”. It works by adding the duration of the single longest view time to the video’s total watch time, which is counted as one display time. The total amount of watch time is then divided by the number of times the video was displayed.
The rate at which videos were watched to completion were under-reported (Corrected).
Discovered mid-November 2016, third-parties have informed Facebook that video completion rates were being misreported due to mismatches in audio and video syncing. The mismatches are caused by hardware and software incompatibilities across different systems. For example, audio may last longer or shorter than the video itself.
Facebook will be making changes as to how its metrics measure videos that are watched to completion. The change is expected to cause a 35% increase in positive reporting.
Over-reporting of Organic Reach (Corrected).
An error in the Page Insights overview was causing the total Organic Reach of the 7-day and 28-day period summaries to be reported as unique visitors. Facebook says that corrections will result in a 33% decrease in the 7-day summary and 55% decrease in the 28-day summary. Paid Reach results and data in other fields will not be affected by the change.
Overstating time spent on articles (Corrected).
This error was causing the over-reporting of the average time users spent on articles by about 7-8% as the metrics were based only on the average time spent. The metrics have now been corrected to calculate the total time spent on an article divided by the total views.
Analytics over-reporting Apps referrals (Not yet corrected).
Ideally, Facebook Analytics should only calculate the number of clicks that an app or website receives through “referrals”. However, it is instead showing all clicks made on content in posts, including photos and videos. Referrals make up a large portion of all clicks on apps and their websites – about 70% of clicks on app posts and 30% of clicks on Facebook page content. Facebook says that it is still trying to identify the cause of the problem.
Interest lists being removed.
Interest lists, a feature that let users group related content for viewing, will be removed due to low popularity. Profiles that were created with this feature will be affected negatively, and Facebook says that a loss of about 5% of its followers can be expected.
More intuitive interface for ads metrics, with expanded descriptive names and calculations.
Facebook says that it will overhaul its naming conventions. After undergoing a full review, the company has decided that it needed to make its metric definitions more intuitive and easily understood by advertisers and users of its platform and metric. Examples of changes to naming convention include “View Content” to “Website View of Content” and “Video Views” to “3-second Video Views”.
All metric definitions and methods for their calculation will also be openly available for viewing. The added transparency will be useful for advertisers who are planning to take advantage of a new reporting interface and customize their reports with metrics that they need to fulfill their campaign goals.
How the Industry has reacted to Facebook Analytics errors.
It is not known how marketers and others heavily invested into the social media platform will receive the news of the errors. It is clear, however, that their demand for greater transparency through third-party verification and clearer definition of its metric parameters show that trusts may have been negatively affected as a result.
Nevertheless, Facebook’s acknowledgement of these errors is a promising development and shows that a workable relationship is easier than ever. Its new policy toward third-party measurement options, clear announcement updates and willingness to implement corrective action when necessary will create a self-regulating system that its users will be able to rely on for years to come.