Editor’s Note: This is the fifth – and final – in a series of posts on “The Marketer’s Guide to Measuring Social Media Marketing Engagement.”
The previous posts in this series (Part 1, Part 2, Part 3, Part 4) explained how to define a standardized set of metrics and categories to lay the foundation for a solid social media marketing measurement strategy. This post will discuss three advanced measurement techniques which can be applied on top of such a foundation: sentiment analysis, unified engagement scores, and cross-network measurement.
Sentiment analysis involves determining the degree of positivity or negativity of a comment or reply, effective for assessing brand image. The wealth of data in social media has fueled new interest in automating sentiment analysis at scale across dozens of properties where hundreds or even thousands of comments are posted daily.
There are a number of companies today that perform sentiment analysis, either manually or using algorithmic classification. Depending on the service, you may either get a simple positive/neutral/negative classification or a numeric sentiment score. The sentiment may also be measured at different levels:
- Brand-level sentiment: mainly useful as a replacement for traditional brand surveys and allows you to measure your brand versus competitor brands at a high level.
- Product-level sentiment: more useful than brand-level since the additional granularity allows higher potential for insight generation compared to brand-level analysis.
- Topic-level sentiment: has potential to produce insights at a deeper level than even the product-level analysis. Examples may include distinction between product feature sentiment versus pricing sentiment or reactions towards individual promotions.
More detailed insights require more granular data and therefore more advanced or time-consuming analysis. The above examples require data on brand, product, or topic being collected by the sentiment analysis tool and a way of aggregating this post-level data according to relevant categories.
Regardless of the granularity at which sentiment data is collected, it is critical that sentiment be reported separately from engagement score. If these scores were combined – for instance, by weighing the value of comments based on sentiment – then one will likely end up with significant loss of information, such as ten positive comments being scored similarly to 100 neutral comments.
Unified Engagement Score
A unified engagement score that encompasses all types of engagements greatly simplifies analysis since it allows distillation of multiple KPIs into one.
The easiest way to create such a score is to assign weights on each type of engagement. The greater the value placed on a particular type of engagement, the greater the weight should be. For example, the formula below may represent engagement score on Facebook.
Many marketers value amplification metrics the most. This is due to amplification activity driving virality, with the expanded reach potentially translating to an increase in other types of engagements as well.
Applause, conversation, and other types of non-viral engagements also help with promoting conversation and retention so they should not be ignored entirely. They also help to increase the algorithmic ranking of posts, indirectly improving reach. How exactly to weigh them relative to one another depends on your social media strategy and objective, as outlined in an earlier post.
While being able to condense four metrics into one makes engagement analysis more manageable, this simplification is not without drawbacks. All aggregation results in loss of information. According to the Facebook formula shown above, for example, there may be no distinction between a post that had acquired 1000 likes and no comments versus one that had acquired 200 comments but no likes.
Unified engagement scores are therefore practical for measuring engagement trends at a high level, but detailed analysis for insight generation should always drill further down to individual metrics.
Even with unified engagement scores, the total number of metrics that need to be tracked across networks can still be significant. For instance, tracking four KPIs (organic/viral engagement, paid engagement, engagement rate, and cost per engagement) across five different platforms (Facebook, Twitter, YouTube, Instagram, and Google+) results in a total of 4 x 5 = 20 metrics that need to be reported on. Cross-network engagement scores help to simplify reporting further in cases where unified engagement scores are not enough.
In an earlier post, we grouped engagement metrics depending on the four classifications: conversation, applause, amplification, and consumption. By creating these metrics as formulas, it becomes possible to aggregate cross channel data. The simplest example is where metrics of similar types are summed:
Conversations = Facebook Comments + Twitter Replies + LinkedIn Comments + …
Applause = Facebook Likes + Twitter Favorites + LinkedIn Likes + …
Amplification = Facebook Shares + Twitter Retweets + …
In more complex cases, metrics for one channel may be weighed differently from another if the networks are thought to have different levels of marketing effectiveness.
Taking the aggregation one step further, it is also possible to aggregate across networks and across engagement types to create a single cross-network engagement score that incorporates all networks and all types of engagements, as in the figure below.
As a warning, the rule of more information being lost with more aggregation applies here as well. At this level of aggregation, there can be no actionable insight gleaned just from the metric itself. For instance, even if the cross-network engagement score goes up, the rise could have come from any network or any type of engagement so it is difficult to derive the appropriate action without drilling down into individual channels or metrics.
As marketing enters an era of accountability, there is greater pressure to show quantitative results of marketing investment. In order to maximize the value of social marketing measurement, it is necessary to define actionable and business-relevant metrics and categories, which together lay the foundation for a solid measurement framework. With such a framework in place, it becomes possible to generate insights that are accurate, actionable, and valuable to the larger business, further establishing social media as an integral component of marketing strategy.
Want to spend your time driving marketing performance, rather than gathering data?
Let us show you what we are up to.