Social media metrics dashboard

The definition of the most suitable metrics to be used for evaluating social media performance is a controversial issue among operators. As...

The definition of the most suitable metrics to be used for evaluating social media performance is a controversial issue among operators. As previously mentioned, the selection of data that a firm retains to be important depends on the type of business and on the firm’s features and objectives. However, the uncertainty about which are the most appropriate data to consider, generally caused by a not clear definition of social media marketing objectives, along with the large quantity of data available (data overload), can represent a barrier for the development of social media measurement
(Gillin 2009).

The development of social media implies a change of perspective about metrics: from more traditional web analytics, based on number of pages views and clicks, to more recent social media analytics, on the contrary more based on conversations that occur in the online community.

Within social media measurement, we can identify a framework of analysis composed by four different kinds of metrics (Lovett 2011): foundational metrics, business value metrics, outcome metrics (KPIs), and counting metrics.

Next subsections will examine more in-depth the meaning, the calculation, as well as the methods of employment of these metrics. They are not distinct measures but they should be selected, integrated, and combined for each category with a clear business objective.

Foundational Metrics

Foundational metrics are a group of measures that represent the basis for business value metrics and outcome metrics because they help to build specific key performance indicators for each goal. They are cross measures that persist across channels and marketing activities. This allows making comparisons among media and specific marketing actions.

Lovett (2011) identified five main foundational metrics: interaction, engagement, influence, advocacy, and impact.

Interaction assesses people that reply to “call to action” and participate to a marketing activity. It comes from number of views, comments, shares, etc. They can be measured for one single activity or more and for a specific media or across multiple channels.

Engagement evaluates the degree of people participation and involvement to the conversation. Therefore, it is different from interaction that quantifies only the replies to specific incentives. However, we notice sometimes confusion in the distinction between these two concepts that impede the development of shared measures among organizations.

Influence represents the power of one person to influence one or more others on a topic, a brand, etc. Also in this case the concept is interpreted in different ways by various businesses and social media. Therefore, it is difficult to find a unique and shared measure of influence.

Advocates represent people who are so engaged with a company, brand, etc. to become promoter. It is the result of the ability of the company to create an authentic dialog with customers. The measures of advocacy are generally connected with a combination of sentiment analysis and metrics that reflect commitment of people toward a company, a brand, or a product.

Impact measures the ability of a firm to reach its business objectives. Generally, the most employed metric among financial measures is Return on Investment (ROI). The impact of a social media activity can be measured assessing expected outcome against a specific goal. For example, if the starting objective of an advertising campaign on Facebook is to acquire a certain amount of new customers, a measure of impact is the result of total exposure of the marketing activity divided by total new customers’ acquisitions (Lovett 2011). However, ROI can measure short-term, single activities related to specific marketing actions on social media (i.e., a Facebook advertising campaign). It cannot be a measure able to give the total impact of the overall social media activity (Cosenza 2014, Mandelli and Accoto 2010). Indeed, some marketing activities on social media cannot be measured in terms of financial metrics related to the return of investments (e.g., crisis management by means of Twitter).

Business Value Metrics

Metrics that come from social media should be of value for various corporate departments that have their own business goals. Indeed, each department needs to measure the success or failure of specific social media initiatives. Among others, possible business value metrics can regard for example the impact on revenues and on customer satisfaction (Lovett 2011).

In the first case, the company could measure revenues generated by a specific social media marketing activity by considering online purchases. Coupons and special codes allow firms to monitor the effectiveness and conversion of a specific social media advertising campaign, offer or post. However, this means the development of prearranged measurement instruments design and attentive planning.

As mentioned earlier, social media have become instruments to develop customer care (Social Media Customer Care SMCC). The interactions generated on social media offer a large volume of data to be analyzed in order to monitor customer satisfaction and reasons of complaints. For example, online forms can be sent to customers or published on social media and be then automatically elaborated
to provide managers customer satisfaction indexes.

SMCC measures can also help companies to make decisions about future investments. For example, if a travel company, by means of measurement activity, realizes that a vast majority of customers uses social media like Twitter and Facebook for customer care, it could decide to move part of the budget from traditional call centers to social media (e.g., social media training for employees of the call center, additional employees for the SMCC staff).

According to the business goals of a company, social media metrics planning can provide a proper range of measures for each department that should be considered in combination to make successful decisions.

Outcome Metrics

Key Performance indicators (KPIs) measurement is based on the previous examined foundational and business value metrics (Lovett 2011). They measure the degree of success of a company in reaching a specific marketing objective. For these reasons, they can vary according to the type of business and to the kind of social media activity. KPIs metrics should be established in advance respect to the marketing activity to connect strategies and tactics in reaching a predetermined objective. If
properly defined, outcome and business metrics can track the results of social media marketing actions over time, generating a benchmark.

Obviously, a firm can have different business objectives to which match KPIs. Among others, hereafter we will identify some examples starting from the classification of Lovett (2011).

A firm could be interested in measuring brand exposure that is the visibility reached after a certain marketing action. Brand exposure on social media is more persistent and amplified than on traditional media. Therefore, it is difficult to be controlled by companies. A measure of exposure is “Reach” that is the result of a calculation: the total amount of unique visitors of a certain post plus number of
shares for each social media considered multiplied by the average number of friends of that social media:

Reach = unique visitors + (shares on FB X average number of FB friends) + . . .

Another KPI employed to measure exposure is Share of voice that measures the brand mentions on social media (Likes, videos, tweets, etc.) among a competitive set. This metric can be calculated for a group of social media or for each medium, in this last case allowing a comparison. Share of voice can be calculated as follows:

Share of voice is generally presented as a percentage and is measured on a given time period of time to allow historical comparisons. Both Reach and Share of voice should be assessed before and after the marketing initiative to have an actual measure of exposure increasing.

Another goal of the organization could be to create a dialog with social media users. A measure of a company’s ability to create dialog is represented by Audience engagement that can generate multiple conversations and virality about a topic. This depends on comments and sharing activity by users but it is also the result of Reach. Therefore, we can consider the following two measures that can help firms to evaluate their ability to create a dialog about a topic, product, brand, etc.

Audience engagement represents the number of people that participates to specific marketing initiatives or to the life of a corporate page. By monitoring audience engagement, the company can identify hot topics.

Multiplying Reach and Audience engagement, we obtain conversation volume that considers the number of unique visitors exposed to a specific brand (product, topic, etc.) conversation across one or more channels and their level of engagement.

Conversation Volume = Reach X Audience Engagement.

If calculated for each social media or for each marketing initiative, these last two metrics allow a comparison of their effectiveness. For example, on the basis of the results of Audience engagement and Conversation volume measurement, a travel company could decide to employ Facebook for some type of topics and activities and Twitter and Pinterest for others.

A third example of goal a company can pursue on social media is to generate interactions: for example, call users to action by means of social media (e.g., ask for their collaboration by means of a contest). Results of this action can be measured by means of the two KPIs: Interaction rate and Conversion rate. The first one measures how many people start the process of interaction and, the second one, how many respond to the call to action respect to the total amount of people exposed to it. For example, a post on Facebook could ask for customers’ participation (contest, opinion, etc.). Then, a user can reply to the company’s request of activity, participating to the contest or giving their opinion (conversion), or simply he or she can interact with the company by means of like and share options without actually responding to the call to action (interaction). Calculations of these two metrics are the following:

Other metrics can help specific departments to understand the effectiveness of their activities. For example, in the case of SMCC is crucial to understand the response rate and timing (hours, days, etc.), the effectiveness, as well as the satisfaction. These metrics can help to support business activities and decisions. Hereafter, some examples of rates are reported:

Customer satisfaction indexes can be referred to single services or social media. They generally ask customers to give a score to different services offered on a scale of values (e.g., from 1 to 10).

Some firms go beyond satisfaction and try to stimulate advocacy that means to transform engaged and satisfied customers into promoters of a product, a brand, or a company. In order to reach this objective, sometimes firms develop the so-called ambassador programs, dedicated to a selected part of consumers to whom personalized services are directed. For example, main international hotel chains have created exclusive clubs and loyalty programs (i.e., Starwood Ambassador Program). The development of social media allows their extension by means of more and more personalized offers. In other cases, ambassador programs can be connected with particular events (anniversaries, launches of new products, etc.).

Results of these marketing actions can be measured in different ways. For example, a possible metric can be “advocates’ activity rate” that measures the members or participants’ level of activity to the program or to a specific event. A calculation generally applied is the following:

Other metrics can be identified also to measure the results of crowdsourcing. Companies may ask social media users to provide ideas to develop a specific product, an event, etc. sometimes supported by a contest. Measures in this sense can identify how many ideas are appropriate to the request and how many are the most appreciated by other participants.

Counting Metrics

Counting metrics refer to a large amount of data coming from single social media. They are called “raw data” because they come directly and automatically from the databases of social media without any elaboration (e.g., Likes, followers, number of visits, etc.).

Social media counting metrics are improving over time. They are increasingly available and can be exported by firms. Main social networks provide generally specific tools aimed at extracting all the available data allowing some sort of filtering functions and some kinds of elaboration (e.g., Facebook Insights). The continuous improvement of social media metrics tools represents definitely an advantage for firms even though sometimes this process may generate confusion in the comparing activity. For example, in the case of Facebook, the option “Like,” previously called “Fan,” was renamed by Facebook in 2010 “Like” in order to simplify and standardize the choices that people have on the social networking site. The meaning is the number of unique people who like the specific Facebook Page. However, the calculation remains the same and the word fan has continued to be
used by both Facebook and companies to create a sort of continuity. Moreover, in 2013, Facebook proceeded to a new restyling of Insights and some metrics were removed or split:

“People Talking about this” was the number of unique people who have created a story about the Facebook Page. A user could create a “story” when liked a page, commented on, or shared the page post, answered a question the company have asked, responded to an event, mentioned the Page, tags the Page in a photo and finally checked into or recommends the Place posted on the company Page.
This metric was split into separate elements: Page Likes, People Engaged, Page tags and mentions, Page check-ins, and other interactions on a Page (Facebook for business, October 2013).

“Friends of fan” was the maximum potentiality of exposure possible to be reached. It occurred when all the fans would have shared contents with others.

Despite, the discontinuity and the possible temporary confusion created with the past, the new version of Facebook Insights provides more punctual information and very valuable functions and opportunities

Therefore, counting metrics cannot be “the one and only” data used by managers. However, they represent an important record, to be interpreted combined with other measures in order to make business decisions.

Sometimes the large quantity of data that can be extracted from social media metrics tools may worry managers and hinder the process of data interpretation.

Sentiment Analysis

The literature on the topic identifies two kinds of analyses for the assessment of conversations and opinions: sentiment analysis and opinion mining. The two terms are generally used as synonyms, even though they had a specific and parallel evolution (Pang and Lee 2008). Since 2001, the expression sentiment analysis started to be used to define specific applications aimed at classifying reviews on the basis of their polarity (either positive or negative). In 2003, the term opinion mining
was used the first time to define tools that “process a set of search results for a given item, generating a list of product attributes (quality, features, etc.) and aggregating opinions about each of them (poor, mixed, good)” (Dave et al. 2003). Given the two definitions and the similarity of tasks, nowadays these expressions are generally referred to a more broadly concept that includes the “computational treatment of opinion, sentiment, and subjectivity in text” (Pang and Lee 2008).

Sentiment analysis has become particularly employed in marketing measurement activities. The term “sentiment” refers to the polarity of the opinion toward a firm, a brand, or a product. It can be positive, neutral, or negative (Lovett 2011):

positive sentiment identifies advocates and communities where the brand/ product is well accepted;

neutral sentiment allows the firm to understand in which communities conversations about a topic should be reinforced;

negative sentiment helps companies to understand where a direct support is needed to improve dialog.

Sentiment analysis can be both manual (human) and/or automatic. Manual or human sentiment analysis requires the action of a human element into the analysis. It is more accurate than automatic procedures but obviously very time consuming.

The advantage offered by human sentiment analysis is the opportunity to identify and interpret the true sentiment expressed also through abbreviations, sarcasm, emoticons, slang etc. On the contrary, automatic sentiment analysis employs complex algorithms that process text strings and determines the overall sentiment (either positive, negative, or neutral). They are less expensive and time-consuming than human analysis but they are unable to differentiate between subtle languages
nuances.

Automatic sentiment analysis can be operated by means of a wide range of tools, from more sophisticated and expensive (e.g., Radian 6, Sysomos, Alterian SM2, Buzztracker) to low cost solutions that can provide simpler services (e.g., Google Alerts, Social mention). The first type allows companies to monitor sentiment and conversations for multiple social media (Ceron et al. 2014a, b). The company has generally a reporting on a different base: real-time, daily, or period reports. Moreover, the firm can set specific rules to be warned in case of particular events. For example, if the conversation about a specific brand is generally 100, the system will warn the company in case of an exceptional increase. On the other hand, the second type of tool, low cost instruments, can be particularly helpful for small-medium size companies that are in the first steps of development of a social media approach.

A common solution increasingly employed by firms is a combination of the two practices (both automatic and human) in order to maximize efficiency and accuracy. A first automatic system analyzes text strings to identify positive/negative sentiment and then, in a second step, a human analysis occurs. New tools are developing to support this new mixed approach. For example, a study of Schmunk et al. (2013) propose a novel approach for automatically extracting and analyzing travel customer reviews that divides the mining task of sentiment analysis into different steps: recognition of property, recognition of sentiment, and recognition of subjectivity.

Sentiment analyses are frequently used in the travel sector to monitor online reputation and conversations. Some investigations employed automated methods. For example, Kasper and Vela (2012) developed a specific tool (BESOHOT) in the hotel sector to analyze and classify textual content of hotel customers’ reviews. Other studies focused on hotel reviews are those of Banic et al. (2013) and Aureli et al. (2013). Some other investigations employed a mixed approach both qualitative (text mining and expert judgment) and quantitative approaches (correspondence analysis) in order to content-analyze the narrative and visual information on a sample of websites (Choi et al. 2007). Stevenson and Hamill (2012) undertook a social media monitoring exercise among ten most visited cities in the world.

Independently from the tool used (human or automatic), what is really important for the effectiveness of sentiment analysis is to define in advance main research questions and then create a set of key phrases/words related to research questions.