It’s a frustrating truth: Sometimes a data set just can’t tell the story you want or think you should be hearing.
On Thursday, a story in Recode relied on new data to suggest that Facebook’s change to its News Feed to de-emphasize commercial posts in favor of those by individuals may indeed be causing people to spend less time on the social network — just as CEO Mark Zuckerberg had said could happen when he announced the coming rollout on January 11.
The canary in the coal mine? Shrinking ad impression growth and rising growth in cost for those impressions since the change in January and February. The story was picked up by multiple outlets.
The data showing changes in CPMs (the average cost per 1,000 ad impressions) and impressions in January and February come from a data set shared by AdStage, a cross-channel campaign reporting and automation platform and Facebook marketing partner.
From the Recode story:
The general thinking is that less time spent on Facebook typically equates to fewer ads seen on Facebook.
That seems to be the case, at least based on early data. The number of ad impressions Facebook delivered in News Feed in January was down year over year, and in February ad impressions were up, but at a much smaller rate than previous months
Facebook CPMs — the cost of a thousand ad impressions — was up 122 percent year over year. In February, CPMs were up 77 percent. It marked the two highest year-over-year jumps in ad prices for Facebook over the past 14 months.
So, impressions went down in January and back up in February compared to the previous year. And CPM growth spiked in January and remained high in February compared to 2017.
I asked AdStage if they’d share their data with me. They did.
The problems & more data
Looking at the data, I cannot see how it can tell a correlation-causation story that falling impressions and rising CPMs are signals that people are spending less time on Facebook. Maybe the News Feed change is causing people to spend less time on Facebook. Maybe it isn’t. But this data doesn’t help us figure that out.
There’s also a timing issue. The change was announced almost halfway through January and wasn’t fully rolled out yet. Wouldn’t we see more impact in February than in January?
Making some kind of early verdict on the News Feed change was also not AdStage’s intention with this data. The company understands it represents a relatively small segment of advertisers and ad dollars running through Facebook, and there are lots of caveats and other signals that can’t be parsed out to draw that kind of conclusion.
Furthermore, the impression and CPM changes are part of a trend, as the Recode piece acknowledges: “The algorithm change may just be exaggerating a trend we were already seeing.” Maybe the algorithm is exaggerating a trend, maybe not. This data can’t answer that one way or the other.
Here are the impression and CPM growth trend charts from AdStage:
You’ll likely notice that the relatively lower growth in impressions year over year started in August 2017. This reflects what Facebook had been warning about for about a year — that ad load in Facebook’s News Feed would max out, causing impression growth to stall and taper.
The impressions chart also shows there was negative impression growth (across AdStage customers) in January, but positive, if slower, growth regained in February.
Looking at the CPM chart, growth did spike (122 percent) in January but then fell back in line closer to what was observed in October 2017 (77 percent vs. 72 percent year-over-year growth, respectively).
So, yes, data from AdStage customers in January definitely looks funky compared to the other months. Is that because more people were spending less time on Facebook? How could we possibly infer that from these charts, particularly knowing it represents a very small slice of advertising activity on the network and that the News Feed update likely wasn’t fully rolled out for most of January?
The bigger picture
Let’s zoom out and look at what Facebook has been saying about trends across its advertiser base (emphasis added).
At the end of Q3 2017, Facebook reported a rise in overall ad prices. On the company’s quarterly earnings call, Facebook CFO Dave Wehner said of the reasons for the price increase, “One is just the auction dynamic … as supply growth has slowed, then there’s more competition, and you’re seeing prices increase as demand continues to grow. But I think what is important here is we’ve been getting better and better at targeting as we optimize for real business results for advertisers, and we’re better at converting the signals that we get from those advertisers into finding the right ad spots for them.”
At a Morgan Stanley conference last week, Wehner reiterated this:
“We have seen that the character of revenue growth has shifted more to price, so this past quarter it was a 43 percent increase in price year over year and a four percent increase in impressions year over year, and that’s an inverse of what we’ve seen in past years. So, it’s increasingly driven by price.”
Wehner also said this about the potential impact of the News Feed change:
“… the impact of the business is much more muted because we’re still seeing that there’s lots of post engagement. So when you’re taking away time from things like passive video, it doesn’t mean you’re not seeing as many posts in [News Feed]. So … I don’t think the impact on the business is really that profound relative to the impact on time.”
The takeaways from Wehner’s comments: Ad prices have been rising for many months. Impression growth has slowed for many months. Post engagement hasn’t dropped off with the change, and Wehner expects little impact on the ad business.
Zooming in on an engagement story
Now to zoom back in on the AdStage data.
Here’s what we can actually take from it: Engagement on performance ads from AdStage customers has been trending up compared to a year ago. On average, since October 2017, even as impression growth has slowed, clicks on ads to advertiser websites or apps have been growing.
Also, after four months of negative growth, AdStage reports click-through rates have been rising since November, increasing sharply in January (108 percent) and February (191 percent) year over year, on average, for its customers.
And while CPM growth spiked in January, the growth in the cost per click (CPC) to take users off of Facebook has been trending down year over year since October for AdStage customers.
In summary, performance advertising on Facebook has been getting more efficient — more clicks at a lower cost — for AdStage customers over the past four months compared to the previous year. Spend increased 111 percent year over year in February.
Taken together, these trends look much more attributable to ongoing pressure on ad load in the News Feed and improvements in ad targeting than January’s News Feed change announcement. “Overall, this signals to us that Facebook’s ad algorithm is really good at finding the best audience to serve ads to,” AdStage Director of Growth Marketing & Customer Acquisition JD Prater told me.
As to whether people are spending less time on Facebook now, I hope you’ll agree these charts can’t tell us that.