Should embedded Twitter cards really be considered social traffic?

The Tribune published on Aug. 19, 2016, a fun story after one of our travel writers was given what appeared to be a copy of Colonel Sanders’ original recipe, a super-secret blend of 11 herbs and spices that KFC has taken great lengths to keep that way. We even whipped it up in our test kitchen.

We launched, promoted, shared, tweeted etc.

The piece did well, but not quite as well as I had predicted it would. Initially.

Then, a couple days later, it started to make the rounds and many publications started reporting on our reporting, creating their own versions of the story, linking back to us.

Usually, an inline text link to the original source doesn’t drive a lot of audience. It’s fair and perfunctory, but doesn’t amount to much traffic. Here’s an Associated Press folo on our KFC story on NBC News. As you can see here on Chartbeat, it’s referring about 9 realtime visitors.

That’s fairly typical of a link. Sure it’s there, but no one really clicks on them.

So three days after it was published, on Aug. 22, we started seeing some traffic to the story from Fox News and from Twitter. The latter was nothing huge, but it was noteworthy. As you know, Twitter is generally realtime. Moz pegged the lifespan of a Tweet at 18 minutes. I’ve found that estimate a little short, but you don’t usually see three-day old stories start to get new life on Twitter without some help. And we weren’t tweeting it anymore.

We couldn’t initially pin it down, even with CrowdTangle. When we followed the other traffic sources, we found where it originated — Fox News. Fox had aggregated our story and embedded a Twitter card from a business tweet.

I assume they just wanted the visual. But that was referring about 50–60 realtime concurrents (said Chartbeat), which is fairly strong for Twitter, and it was outperforming the inline link in the story at that time.

On Aug. 25, we saw an even larger bump after the New York Times also aggregated it.

The NYT took the same approach — an inline text near the top and an embedded Twitter card lower.

Similar results as well, with a lot of traffic coming to the Tribune’s site under the social/Twitter umbrella. Here on Aug. 25 you see the results of two tweets sent on Aug. 19 — the one from business on Fox News and one from the main Chicago Tribune Twitter account on the New York Times — still driving a lot of traffic. And that’s audience that Omniture, Google Analytics and Chartbeat (to name a few) would classify as social/Twitter. Which it is, technically. But really, at this point, it’s just a fancy link with some share tools more than it is a truly social experience. Most of the interactions were simple click-thrus. People weren’t commenting, favoriting or quoting anymore.

Here’s the data from Twitter Analytics, showing nearly 24K click thrus from one tweet.

Twitter Analytics report for Aug. 19, 2016.

Here’s the total from three Twitter cards, again, via Twitter Analytics.

In fact, those three Twitter cards to the same link drove the fourth-most click thrus since in 2016.

So if you view Twitter cards through that lens of a glorified link, this traffic is actually a lot less social than common analytics software would indicate. Most of that later traffic (seen in the chart below) came from an audience that first had to visit another site and then click on a link — either in text or in a Twitter card. Is that Twitter card traffic truly “social”?

Scale only via Adobe Analytics to protect the innocent.

It would be nice if embedded Twitter cards could be tagged to give a better indicator of source rather than appearing like a purely social source being quoted and RT’ed. Just give me the option. Consider that if neither Fox nor the NYT had included inline links, it would have been even harder to track this traffic.

As a side note, the referred traffic tended to stick with the story. It wasn’t garbage drive-by readers/viewers.

Chartbeat Daily Report for Aug. 22, 2016.
Chartbeat Daily Report for Aug. 25, 2016.

Anything over a minute is a fairly decent engaged time, and on those two days both averaged 1 minute and 18 seconds. Very respectable.

It’s also worth noting that the inline links did drive traffic in this case, which is very atypical. In fact on the NYT, the link near the top drove three times the traffic as the Twitter card embedded lower.

Chartbeat Editorial Dashboard showing 438 concurrents from and 152 from one CT tweet.

So if you’re aggregating any Tribune content, hey, feel free to include a tweet. Worked well here. It just would be nice if I could separate embedded Twitter card traffic from standard Twitter traffic when I look back to analyze the audience data. The same would apply to any of these embedded semi-social experiences.

It also would be nice if Twitter Analytics (or CrowdTangle) would allow you to express the engagements with a single tweet over time (by hour/day) by type (RT, CTR) and not just as a total. That would highlight these cases more easily.