The importance and relevance of social shares and social media in our online marketing efforts must be one of the more frustrating subjects out there for marketers, since we’re still not able to prove how much they weigh in ranking our websites, promoting our brands, or making us visible online.
Time and time again studies by Searchmetrics or MOZ have tried finding correlations between social signals and search engine rankings and every time, actual positives were hard to confirm. Some signs are there, but it’s still a longshot to declare their exact value. Here is one of the takeaways from Searchmetrics’ study for 2015: “Top ranking URLs have more social signals – this factor increases exponentially in the top places.”
Yet the conclusion remains largely the same as in previous years:
“The question of how social signals directly affect rankings remains. As noted in our analysis, higher-ranked URLs have more social cues such as Likes, Tweets and +1s than those sites further down the ranks, but Google has continually emphasized that it is not using social signals as a direct ranking factor.”
We can tell social signals (shares, tweets, retweets, pins, likes, +1s, etc.) weigh in popularizing a piece of content, a website or blog, and definitely help with brand awareness, lead generation or customer engagement, but how much do they actually help your site?
Taking Social Media Personally
As a small, in-house marketing department here at Advanced Web Ranking, we do our best to keep tabs on every aspect of our online presence and are constantly trying to improve on all fronts. Lately, we’ve been tackling precisely the subject of social media and its impact on our rankings, lead generation, brand awareness, and so forth.
We have been looking at our own metrics and past performance, trying to come up with a long-term, goal-oriented plan that would focus our social media presence. Whereas in the past we only used social for awareness, the promotion of our products, and support, we decided it was time to revisit our overall approach and take more premeditated steps.
This led to a series of difficult questions that we needed to answer in order to inform our social media strategy:
- are we using all the right social channels to increase brand awareness and promote our products? And will they return high value?
- is our social media strategy aligned with our overall business and marketing goals?
- does our content support those goals while providing value to our target audience?
- are we using the right metrics to measure progress and success? And more importantly, what metrics would we be interested in according to the goals we want to achieve?
- what difference can we make with a better defined plan?
So it was time for us to come up with solutions and lay a new foundation for our approach to social media. For that we wanted to explore the ins and outs of social media marketing and gain a wide perspective on the subject, preferably from the best sources available. We wanted to know which metrics mattered the most when measuring the success of social shares and any campaigns across social, and find some best practices for producing content that attracts lots of shares.
Social Shares ≠ Links
According to Searchmetrics’ study results, the higher up in rankings an URL is, the more social signals it has. But would that translate to more inbound links as well?
That’s what this study by BuzzSumo and Moz tried to find out. Their research consisted of analyzing over 1 million articles to see whether there is any correlation between the number of social shares and links that one piece of content can get. As many of you who’ve read the Moz article by now know, the results were rather disappointing – no correlation, let alone causation, between social shares and links pointing to an article.
This means that most articles that had a great number of shares didn’t necessarily have an increased number of links pointing to them, or vice-versa. You can read more about their findings in the link above, but what they basically discovered was that most content out there is not worthy of sharing and gets even fewer or no links. What was interesting however is that they pinpointed which types of content are more likely to be shared, and drew some insights on how content format and length can attract social sharing and/or links.
Content which is supported by research, or editorial journalism get both a high number of shares, and an increased number of links pointing to them. It was also confirmed once again that long form content (over 1,000 words) gets shared more often than short content. As for content format, list posts drive both shares and links, but so do videos or quizzes.
So if our goal with social media was to attract social shares and links, we would need to focus on a very specific type of content and even then, success wouldn’t be guaranteed. As with most cases, writing with your audience and their interests and needs in mind is always the best invisible “ranking factor” that might just surprise you and get you closer to your goals.
Still, this only brought up even more questions.
Say we managed to produce the type of content readers love: what metrics would really matter in assessing the performance of a piece? How could we really tell it made the impact we needed and wanted?
Social Shares ≠ Readability
So we dug into it some more, and found data that proved… we don’t usually use the right data when measuring the success of a content piece!
This time, statements came from Chartbeat, which, in their own studies, wanted to see if there is any correlation between social sharing and the readability of a text. Again, it seems that URLs with the biggest number of shares weren’t those that had been read/consumed in their entirety, but rather those which were consumed in a ratio of about 30%.
In this very informative article, Farhad Manjoo crunches on data Chartbeat revealed about the readability of Slate articles – which might not be a point of reference for all types of websites, blogs and ezines, but which does show a trend. The majority of Slate readers consume about 50% of an article, but scroll depth gets higher if the article contains photos. Yet about 5% of readers don’t scroll at all, they just leave the page without interacting.
For both Slate and other websites, Chartbeat found that there is little correlation between Twitter shares and amount of article consumed.
“Both at Slate and across the Web, articles that get a lot of tweets don’t necessarily get read very deeply. Articles that get read deeply aren’t necessarily generating a lot of tweets.” said Manjoo.
The best explanation for Chartbeat’s findings came from the company’s own CEO, Tony Haile, in a piece he wrote for TIME a couple of years ago. Here, he goes in depth on how time spent on page doesn’t exceed 15 seconds for up to more than half of all readers. In the article, Haile debunks several myths about how readers interact with content online. Among them, he also states their data shows we don’t necessarily read what we click on, and articles who get the most shares aren’t necessarily the ones we’ve read in their entirety.
“There’s a very weak relationship between scroll depth and sharing” – Tony Haile, Chartbeat CEO
They also included some possible explanations for these findings:
- clicks from social media are more likely to come from mobile devices, where readers typically spend less time on the page
- readers have particular preferences for the links they click on – studies show people are more likely to share stories that are happy or nostalgic (rather than crime-related, for example)
- it’s possible that some highly retweeted stories contain all the information necessary in the headline, so there’s less impulse to click
And, let’s face it: sometimes we share content because it was created or shared by a thought leader, a brand, an influencer – this is a great way of initiating interaction with them, getting their attention, and positioning ourselves better in front of our own audience. Marketers know what this is about all too well. They not only practice it, but preach it in industry articles or courses too.
Whether that should bring any comfort or not is beside the point, since our goal as content marketers is to produce pieces that attract all the success metrics possible: time on page, shares, likes, links, entity mentions, reach, unique views, and so on. But there is still disappointment in knowing social media signals can be a hit-or-miss affair – either you’re an already reputable brand with a huge community that keeps amplifying your content and advocates on your behalf, or you need to work really hard on understanding what your readers and prospective readers will be interested in and producing that.
But what’s the value loss here if you still get your shares?
The real problem is that you don’t get readers to engage with your brand. You don’t get to tell your story, send your message, leave your mark on their memory. You miss out on opportunities of converting them into long-term customers. It is also a matter of common sense, but of study as well, that the more time a reader spends with your content, the better their understanding and recollection of what you present will be.
And as Manjoo’s article above points out, some companies like Upworthy, Medium and of course Chartbeat themselves, have wisened up to the online user’s reading patterns and habits and, with the help of Chartbeat tools and data, are now measuring relatively new, but much more valuable metrics, namely “attention minutes” or “engagement time”. This kind of measurement allows them to keep their eye on the real ball, on the qualitative aspect of how readers interact with their content, which in turn serves them useful insights into improving what they produce and how they produce it – including placement of share buttons or ads on the page!
It bares mentioning here that Chartbeat data was referring to Twitter shares specifically, but like Adrienne Jeffries mentions in her excellent post, the same behavior might apply to Facebook as well. Her article is particularly insightful on how Upworthy has changed the way they measure the success of content with readers by tracking attention-focused metrics.
Now this is all fine and extremely interesting, but it still didn’t make it clear what the real value of social shares is in an overall marketing strategy. The more we read on the subject, the farther back in time our research went. And things only seemed to worsen when we found HubSpot’s study on social shares versus clicks.
Social Shares ≠ Clicks
To further prove we’re tracking the wrong metrics when it comes to social media and audience engagement, HubSpot conducted their own studies to see whether there is any correlation between social shares and clicks.
In an article published as far back as 2012 on Social Times, Shea Bennett made another disappointing claim:
“HubSpot’s Dan Zarrella looked at more than 2.7 million tweets that contained links and found that there is almost no correlation between retweets and clicks. Indeed, 16 percent of the analysed tweets generated more retweets than clicks, which leads us to only one conclusion – many Twitter users will blindly retweet something without even looking at it.”
What HubSpot’s study basically said was that a lot of people retweet content without even clicking on the links they share, which aligns with Chartbeat’s findings that a large percentage of people don’t even read the content they share – especially when it comes from a source that’s already considered authoritative. But should that be reason enough for content creators to stop producing quality content just because they know people will share it anyway?
Not at all; as I mentioned before, having readers entirely skip reading your posts is a missed opportunity to interact with them, engage them, and staying top of mind. I won’t even get into what that means for paid advertising.
So here was another piece of evidence pointing to the fact that social shares aren’t directly related to valuable, quality metrics that might indicate the success of a post. Remember that both Chartbeat and HubSpot focused solely on Twitter shares, but it’s pretty safe to assume users would have a similar behavior on other social media platforms like Facebook or Google+.
What about those who won’t use Chartbeat? How can we measure engagement with our content, or determine what type of content works and what doesn’t?
One way to do it is the way good marketers always have: by taking an in-depth look at all of your metrics, across all channels, including social. So it’s not just about likes and tweets (by now Twitter has stopped showing tweet counts on its share buttons), but about engagement metrics such as comments, retweets, favorites, or clicks. Applying social listening is also an important asset – pay attention to what people are saying, when and where they’re saying it, and try to get a feel of the general sentiments associated with their interactions. All these, combined with the usual analytics data like CTR, time on page versus bounce rate, returning visitors, and very clearly set goals and KPIs can help a marketer’s assessment of the performance of their campaigns.
It might mean the process remains a complicated one, but it can still be very useful in indicating those pieces of content that are truly successful with readers. And by comparing data over time, and individual pieces of content between them, or at a higher level, campaigns themselves, you can gain insights for each particular situation or project you work on.
Social Shares vs. Organic Traffic
Thinking further about this, the question of whether URLs with lots of social shares are more visible in search results and vice versa came up.
The idea that URLs that get lots of social shares will also get more traffic seemed very probable, imagining that content Google deems as qualitative should most times overlap with the content that gets most appreciated by people via social media.
We wanted to find some of our own answers and test out that idea in a short, preliminary study.
We decided to start with just a small batch of URLs, and see whether some initial data would show something relevant enough to justify research on a bigger scale. We were cautiously optimistic about finding a consistent correlation (not causation) between social sharing and organic traffic.
So we got together with one of our devs and laid out what we wanted to try. The task seemed simple enough to begin with: we wanted to track a batch of 10,000 URLs and see if we get any hint at the above-mentioned correlation.
For this initial study, we took advantage of one of our most recent projects – AWR Cloud Research, where we track over 30 million keywords in Google with a search depth of 20 results.
We selected the top 100k domains by traffic from the AWR Cloud Research Database for an initial research. For each domain we also got the top 10 pages by traffic.
We retrieved the social shares data for every domain and computed the domain traffic as follows:
SUM (CTR x SearchVolume)
To keep the results contained, we removed URLs that had no, or very little traffic, and no, or very few keywords to rank for.
We only measured URLs with keywords ⪴ 10 and traffic ⪴ 1000.
Across our approximately ten thousand URLs, the correlations were rather weak, and only seemed to become even weaker as we limited URLs to those with a specific amount of visits (clicks). For URLs with traffic lower than one billion visits, the Pearson correlation dropped to an underwhelming 0.144 and toggled around the same values for URLs with visits between one hundred thousand and one billion.
The Pearson (P) correlation coefficient = 0.342
Adding social share values equal to, or larger than 1,000 didn’t make much of a difference either, except in decreasing the correlation even further. All this resonated pretty badly, especially when we consider Tony Haile’s statements that there is a correlation between the number of tweets and the volume of traffic that goes to an URL. Of course, tweets don’t represent all social shares, and all shares are not equal, but I expected the data would have uncovered something similar in this study.
We deemed the results of this short study inconclusive, and in the end were not convinced of their validity. All signs seem to point to the fact that social shares should have a noticeable influence over traffic, and we were interested in determining just how big the correlation is.
The Marketing Endgame
Because we got no pertinent results to speak of with this initial phase of the study, we decided to move on to plan B and cover a larger, more relevant number of domains and URLs in a future study. The results only fueled our curiosity even more and determined us to start a new, more extensive and complex analysis.
Taking into account all the information we’ve stumbled on, the contradictions and confusing elements, I’d be tempted to surmise that social shares have little to no effect over search engine rankings. Except experience and ranking factors’ studies keep showing something else.
Perhaps we’ll not find out very soon just how much social signals factor in rankings, but continuing with studies like these might help us draw some useful insights into how we can use social more precisely in order to achieve the marketing goals we set.
More and more marketers agree that SEO and marketing strategies should aim for one and the same thing: providing value and quality to your visitors, thus turning them into long-time customers and brand advocates. So I believe the approach should be a holistic one – creating content with clickable headlines that retain user engagement by providing something useful beyond clickbait tactics.
As for our internal marketing efforts at Advanced Web Ranking, we concluded that all we can do is rely on social media as much as the next brand. We’ll continue to measure growth and track metrics separately, depending on what goals we’re trying to achieve. I still believe social presence is extremely important for a company’s visibility and raising brand awareness.
Over to you!
We’ve already started gathering some of the data, but could really use your input on this. What would you test if you started a social shares’ study today? What particular aspects of social shares and signals interest you, and what would help you make better assessments of your social campaigns and presence on social media?
Hopefully, an open discussion about this might help us expand the aim of a future study and analyze data in a way that will empower as many online marketers as possible. I look forward to your comments, answers, and questions!