Facebook 'fesses up to fudged ad metrics … again
Social Network™ messed up counts of the stuff advertisers and publishers love
Facebook has offered the advertising world an An Update on Metrics and Reporting that's needed because it's been counting things wrong, again.
Sharp-minded Reg readers will probably recall that in September 2016 The Social Network™ sheepishly admitted that it had overstated average video viewing times by up to 80 per cent for around two years. The company patched things up, apologised, promised to do its very best not to have something like this happen again.
This brings us to today and the new update, in which Facebook admits “We’ve uncovered a bug in Page Insights. On one of our Pages dashboards, one summary number showing 7-day or 28-day organic page reach was miscalculated as a simple sum of daily reach instead of de-duplicating repeat visitors over those periods.”
This is kind of a big deal because plenty of businesses spend up big on Facebook pages. They'll need to recalculate return on investment now.
The company also admits that for Instant Articles, its fast-to-load feature designed to get eyeballs in front of publishers' content, “We’ve determined that the average time spent per article had been over-reported by 7-8% on average since August of last year.”
Another glitch saw the site “undercount the metric 'video watches at 100%',” albeit under limited circumstances. A fourth counted the wrong clicks when assessing traffic to Facebook apps, again resulting in an inflated number reaching users and Facebook's analytics service.
Facebook's solution to its self-created mess is more third-party auditing and a new blog explaining its metrics.
The MetricsFYI blog is billed as “a new channel for regular information on metrics enhancements”. Which rather sounds like Facebook has more changes planned.
It is tempting to assume those new changes will also be responses to errors, but the first instalment of the blog announces the retirement of a feature called “Interest lists” that's been axed due to “low consumer usage”. ®