Most organizations have acknowledged that social media belongs to contemporary communications strategies. The available data of social network usage has made success measurement in this field increasingly popular, so many businesses and also political organizations have started assessing their performance with social media analytics. Although we consider this a welcomed development to shed some light on what social media marketing can really do, we observe that these analyses often only scratch on the surface. They can lack substance or even deliver questionable results. This sometimes happens due to limitations by the used tools such as limited data availability, or simply because of time constraints. In some cases that may even lead to sufficient results. But nonetheless, during our work on social media analytics both from an academic research and a business perspective, we argue that far too many studies and analysis don’t get the maximum out of the accessible data. In this short opinion paper, we therefore want to discuss the assessment of social media impact and help sharpen the analyses to deliver better results. This is why we came up with five ideas for aspects we have identified that often get misinterpreted or missed out on in social media analyses. Some of these claims may be rather obvious and well examined, others not so much. We ourselves don’t claim that all analyses that we worked on are already perfect – more the contrary. There are many reasons why analyses are not as sophisticated as we’d like them to be. Our aim is to give an impulse for a more lively discussion on how to create better social media analyses for the benefit of all people working on related projects and subsequently to everyone who relies on our analyses.
1. Identify Specific Platform Characteristics When talking about social media analysis, we can usually boil it down to the more well known platforms like Facebook, Youtube, Instagram, Twitter and Snapchat. As a result, many analyses focus on those networks. Among them Facebook, still the most popular social platform globally, and Twitter, due to its wide usage in the US and status as an elite network in Europe, have attracted the most attention of data analysts and researchers. Experience shows that when people talk about social media communication, especially in certain fields, Facebook and Twitter as the two elder networks are often mentioned in the same sentence. And of course, as in many cases organizations are active on both networks, data analysts are sometimes forced to bring together the data for the two networks. Nevertheless the different networks have to be regarded as independent platforms. Their internal structures and logics as well as their sociodemographic characteristics make this differentiation necessary
When we compare Facebook and Twitter, one glaring structural difference is that there is a separation of private accounts and pages on Facebook, while this distinction doesn’t exist on Twitter. The microblogging service offers the same kind of profiles for every purpose and also, most of the profiles are public. The flow of communication between all users on Twitter is thus much more unified than on Facebook. This has relevant implications for the usage and subsequent performance assessment of each channel. The socio-demographic composition of the users is relevant in the discussion of what role a network might play in society. Looking at the German market, Facebook is the largest network by far concerning the monthly active users, with Twitter lagging behind significantly. A representative survey shows that 51% of all Germans use Facebook once a week while the number for Twitter is only 11%. Facebook is thus far closer to represent the German society as a whole than Twitter, which acts more and more like a niche network for the media and politics. This however doesn’t mean that Twitter is not relevant. On the contrary, due to its socio-demographic composition it plays an important role in the agenda-setting for different channels, including legacy media.
2. Get a Deeper Understanding of the Metrics While analyzing social media data, some metrics can be misleading. One prime example is a metric that many people use for success or impact measurement – engagement. Many social media marketers’ dreams seem to be to engage people in a dialogue, trying to bring them closer to the brand or product. In other words, bring them to interact with your content. But what these interactions really mean has to be thoroughly discussed especially when doing cross network analyses. Observing the total number of all interactions, often referred to as engagement, is a good starting point to see whether some analyzed accounts or figures stick out. However, we always have to take the different interaction types and their distribution into account. The term engagement can also mean rather different things when it comes to specific platforms. Let’s again use the example of Facebook and Twitter. On Facebook, engagement usually encompasses all interactions, meaning likes, comments and shares including the reaction emojis that are supposed to show different emotions and are actually an extension of the like button. On Twitter, engagement means all retweets, mentions and likes. You already see that this is different – also defined by the specific structure of each network. That being said, let’s focus on Facebook for the sake of simplicity. The point that metrics have to be carefully evaluated, however, counts for all networks.
Likes are the type of interaction that is most easily used, it’s just one small click. Compared to a comment, where people have to put at least some thoughts into what they want to say, the hurdle to click on the like button is quite low. Same holds true for reactions, although they include additional information for our analysis. Sharing content on the other hand needs some willingness to provide people within the own friendship group with content. So even if we only look at the Facebook universe, we have to carefully consider the different components that make up the metric of engagement, what they mean and how they are distributed. When it comes to the distribution, we want to point out to the problems when using aggregate data. The results can be driven by single events or constant news flow. Due to the logic of social media (virality) sometimes a single post or tweet can be the cause for the largest part of a profile’s engagement. This gets totally faded out when you use data for a full year and calculate a monthly average. The solution to this is to use different aggregates of your metrics and combine them with additional information. When compiling a report you can, for example, include the engagement for the post with the highest engagement and, if available, the stats for unique users. Or, try more sophisticated measures like a standard deviation that gives information about the spread or variation in your data.
3. Use Benchmarks and Additional Data Sources It can often be observed that social media analyses, especially in general interest media, pick some rather obvious KPIs and then talk about these numbers. However, as in many other disciplines, it makes sense to use comparative approaches in social media analyses. Most analysts and researchers already compare their own brand’s or organization’s social media performance with other actors in their market or benchmark it with the own history. This already adds a valuable perspective as you get to know if you’re lagging behind your competitors or if you are an over performer. Data from outside of social networks can give valuable further insights as well. As a rather simple example, quintly has combined follower growth data in the US presidential elections of 2016 with news data. The result showed that peaks in follower growth often appear at the same time as politically relevant events like TV debates, conventions or terrorist attacks. Take a look at our in-depth analysis here. This is just one example that illustrates that it makes sense to look at social media performance data not only in isolation. Social networks exist among many other online and offline channels and is also influenced by “real world” events, so why not compare your exposure in social media with legacy media data? This can give you information about the spread of your content. Is it a topic in social media alone, and thus only reaching the audience on those channels? Or is your content spreading cross-media and thus reaching a potentially bigger audience?
4. Think About the Influence of Paid Content According to a 2014 report from Statista, spendings on social media ads are rising worldwide. Therefore more and more reach and engagement is not caused by organic but by paid activities. If we take Facebook as an example, it is of course possible to measure the impact split by paid vs. organic for own activities. However, we don’t get this information if we want to analyze pages that we don’t have admin rights for. How much impact social media content creates organically and how much comes from paid formats is simply not possible to find out with publicly available data. It is however possible to create estimates, for example with processes that include machine learning like we do at quintly with our Sponsored Posts Detection mechanism. It is possible to predict if content that was posted publicly to a Facebook page got boosted by paying for it afterwards. But although this approach is quite accurate and shows some part of the picture, it only offers a partial insight. So when talking about any social media impact benchmarking, we always have to consider that the split between paid and organic can only hardly be determined as of today. Accordingly, it needs to be discussed how we deal with that issue and if there are possible ways to overcome those limitations in the analysis. The good news is that Facebook has announced in Fall 2017 that there will be more transparency in the future especially for the highly debated political ads. You can read more about that here.
5. Focus on the Real Goal of Your Analysis Social networks offer an abundance of data and this can sometimes be quite overwhelming. We therefore suggest to always start with a clear question that you want to answer with the data. Don’t let the data lead the way, at least not always. Social media data can be used for different approaches. One example would be that you can look at the effect that social media content has on the audience when you want to assess the effectiveness of your content marketing strategy. In that case, the analysis would focus more on data directly related to posts or tweets, thinking much more of individual content performance. You would more directly analyze engagement on a per post level to get a clear view instead of the overall page level.
A very different case is an analysis of customer care activities in which the metrics would focus on the exchange with users, such as looking at how many questions are being asked on your social customer care channel, when they come in and when they are being answered. On the other hand, as social media are controlled communication by organizations, the data can be used to measure the positioning of an organization e.g. a company or a political party. So instead of looking at annual reports or party manifestos you can turn to the social media channels (i.e. Facebook) and measure the profile of a company or party. The combination of the two approaches finally gives you some hints as to how a profile is perceived by your audience. It lets you know if the main positioning fields are the ones your audience really cares about and, therefore, engages with. This of course includes additional analysis tools like manual content analysis,word mining or more sophisticated automated content analysis like topic models. To put it simply: Which KPIs you report and which analysis and methods you use really depends on what you want to know from the data
Source : Quintly is one of the leading social media analytics providers, helping businesses, the media and politics to assess the impact of their social media channels. It’s based in Cologne, Germany and The Research Institute for the Public Sphere and Society (fög) is an associated institute of the University of Zurich and specializes in media studies in the field of business communication and political communication.