Measuring Social Media ROI

Traditional methods of measuring ROI are misplaced when it comes to gauging the impact of social media initiatives. A detailed look at some branding and satisfaction metrics and advanced attribution models that help marketers understand the true impact of social and how it can strengthen relationships between the brand and customers.

Originally written for and published by Huge Inc. 2014.

Introduction

Now more than ever, brands want to be “liked.” Facebook, Twitter, Instagram, TikTok and many other platforms allow brands to connect directly to consumers. Even traditionally B2B brands have embraced social media. Increasingly, though, marketers have questioned the ultimate value of these efforts, asking what is the value of being “liked”? Certainly, it is difficult to determine the return on investment for social media interactions. Major brands from Coca-Cola to Nissan to MasterCard have all admitted they don’t know if their social media investment impacts their bottom lines. Because it is difficult to determine “social ROI,” general dissatisfaction with social as a marketing channel has grown.

By thinking about the problem purely in terms of financial returns — on what are usually relatively small investments to begin with — marketers shortchange the ultimate value of social and potentially miss out on opportunities to meet many different kinds of important business objectives.

Key Findings:

• The real value of social media initiatives is in developing and strengthening a relationship between a brand and its consumers. Traditional ROI measurements and attempts to analyze direct impact on sales are futile and misplaced.
• Marketers should measure social initiatives in terms of branding and satisfaction metrics like Net Promoter Scores, which measure both satisfaction and virality.
• Advanced attribution models, like Agent-Based Modeling (ABM), are the best ways to predict how social media efforts might drive offline user activity that meets business objectives like increasing sales or decreasing call center volume.

Brands can’t measure social media success without knowing their business objective.

Early on, too many brands rushed in to building a social media presence but lacked a cohesive strategy that aligned with overall business goals. Social was and for many still is a silo, but it’s critical to align social media KPIs with specific business objectives. Brands should establish goals before the start of a social media campaign and determine what kind of tracking measures they need to implement upfront. Goals may include generating revenue, reducing customer service costs, shifting brand sentiment, improving operational efficiency, cultivating customer relationships or gleaning insight into target markets.

These objectives are not mutually exclusive. In fact, a social media program can serve more than one goal. While increasing revenue and reducing cost ultimately align with financial goals, it doesn’t mean that social media can’t also assist in meeting PR, research or marketing objectives. Zappos has used Twitter to conduct market research for new lines of business. Comcast has a dedicated customer service Twitter account (@comcastcares) in addition to its main Twitter handle used for branding and company news. DKNY deftly avoided a PR crisis by addressing a potential threat involving copyrighted images used in a storefront by quickly posting an explanation and remedies on its Facebook account. Dell has found success with its @DellOutlet account in driving sales for refurbished computers, building to nearly 1.5 million followers and $6.5 million in revenue less than two years after launching the account in 2007. Companies such as Ford and Starbucks are getting better and better at integrating social media into multiple departments and business functions.

Getting beyond “ROI” to measure the true benefits of social media.

The only thing certain about social “ROI” is the “O”. First, the investment made in social is typically relatively low, even when paid social media efforts — i.e. sponsored stories — are included. It is easy to ask a question like “does Facebook drive in-store sales?” But not only is it difficult to answer, (as we elaborate on later), it is also the wrong question to ask. “ROI” is the wrong term to use. Instead, marketers should accept that calculating a specific financial return is not immediately realistic and focus on measuring the overall impact a brand’s social presence has on its relationship with customers. Marketers should ask three basic questions as they evaluate this impact of a social media campaign on the customer relationship:

• How many? How many impressions appeared in market? How many people were reached?
• Who? Who were these people? Did they fit the target?
• Did it work? This is the most difficult to answer. By answering the first two questions, however, it is possible to get much closer to answering the third.

But while these familiar questions apply to social media, the third one — did it work? — is complicated by two unique attributes of social as a platform: (1) the communication in social is two-way, i.e. users and brands talking to each other, making it more like a customer service model than a traditional marketing one; (2) in its ideal state, social media is viral. Of course, the first two questions — who and how many? — can be answered by looking at metrics from the platform itself or third-party analytics services. But where to find a metric that helps answer the third, given social’s unique nature?

Brand metrics, like Net Promoter Score, are the best means to determine whether social media efforts work to impact the customer relationship.

Understanding how consumers who follow a brand on social media feel about that brand and how that is different for those that don’t is really what will reveal if an organization’s social media efforts are working.

While we do not recommend Net Promoter Score for standard brand tracking, we believe that it is the KPI that best measures the impact of the true benefits of social media. NPS is a measure that identifies how likely users are to recommend a brand to others. This measurement captures both the customer service aspect of social (Zappos, for example, has always had an extremely high NPS due to its well-known customer service), as well as its potential for virality. The uniqueness of this metric, and the ability to access it from many syndicated studies make it a strong option for social media branding analytics.

This is not to say other metrics have no importance. For an e-commerce brand, tracking conversion and optimizing to conversion is clearly important. For an ad-supported site, knowing what types of content drives the highest levels of time on the site is central. Both of these cases, however, involve optimizing metrics for near-term success, not KPIs used to measure the overall health of an initiative.

Advanced attribution models, such as Agent-Based Modeling (ABM) are the best way to determine whether social media efforts drive sales or decrease call center volume.

Finding the connection between social media efforts and offline metrics are much more difficult to establish. Increasing sales and decreasing call center volume can be goals of social media marketing. Measuring the impact of social against these business objectives is best achieved by using a variety of modeling approaches, such as media and market mix modeling. These approaches are hampered by the fact that social as a channel encompasses both the prompts from brands that might drive users to act as well as users’ actions themselves (whether prompted or not), making it difficult to establish causality. In addition, paid television media can easily overwhelm the impact of social initiatives. Further, social media is frequently “new” for a business, meaning the historical trends required for traditional modeling are probably non-existent. Finally, social also requires quick optimization, which is a weakness of models that can only be delivered three months after a campaign is over.

Because of these challenges, we recommend using Agent-Based Modeling (ABM) to best read the impact of social on offline behavior. ABM is a system of dynamic simulations that relies on creating rules for the ways that “agents” operate in a virtual space. The agents are then replicated to match the target population. They are exposed to stimuli, in this case varying types of social media interactions. Offline behaviors are also fed into the model. Through historical simulations the model can be built to replicate the past. As a result it can be used to predict the future of how social will impact users, and therefore what the offline outcomes are likely to be.

A high-wire act: Balancing paid and owned content to stoke virality and drive ROI.

While most brands have moved beyond the “build it and they will come” mentality, they must balance multiple, mutually reinforcing tactics to optimize and monetize the social graph.

• Acquiring fans with paid media: Simply creating a Facebook page or Twitter account is not enough. To build an audience, paid media is still an integral component to developing reach. On Facebook, even with a significant fan base, brands still need to spend in order to guarantee their content is seen. Likewise, Twitter’s Promoted Tweets expand a brand’s visibility with potential audiences. Like other digital media campaigns, optimization for social media is continuous.
• Engaging users and going viral: In addition to paid media, brands need to keep users engaged through owned media as well. This requires not only a stream of compelling, shareable content. It also means keeping users involved with the brand through continuous interaction. An extension of this is more elusive: positioning the brand’s social presence to maximize serendipity in real time and achieve true virality.

A thirsty Senator and Poland Spring’s missed viral opportunity. While no amount of preplanning can guarantee viral success, brands should position themselves to seize serendipitous opportunities as they arise. During his response to the 2013 State of the Union Address, Senator Marco Rubio became parched and famously reached for an off-camera bottle of Poland Spring. Social media erupted, yet the brand’s Twitter account had been idle since 2011, and it didn’t acknowledge the buzz until a day later on its Facebook page. Commentators contrasted the delayed response to Oreo’s now legendary “You can still dunk in the dark” tweet during the blackout at the Super Bowl the week before.

Balancing acquisition and engagement to avoid alienating fans: While it is tempting to “buy fans” with paid media and then consider the resulting large number of followers “success,” the reality is that brands often lose the interest and loyalty of followers immediately by not following up with engaging owned content. By analyzing Facebook and Twitter data for 50 randomly selected brands across, retail, travel and CPG data we found a negative relationship between change in positive brand sentiment and rapid fan growth. In other words, those brands that grew fastest tended to have fans and followers that progressively liked them least. It is inherently true that as a brand gains followers beyond a core of diehard fans, its overall popularity will become diluted. But our hypothesis is that some of that falloff in brand sentiment is due to brands’ failure to keep new followers engaged with compelling owned media. By allocating resources to sharing that content and positioning for viral opportunities, brands can preempt some of that inevitable loss of sentiment.

Tying results back to business objectives: In terms of driving ROI, marketers need to map their social media goals to specific metrics. For instance, an airline may use Twitter as a customer service channel so an ROI framework should assess reduced call center volume as a key metric. For an automotive brand, the metric measured may be increased brand favorability and loyalty rather than net sales, assessed via surveys. The final stage in the framework discussed above is demonstrating how paid, owned and viral efforts supported the original business objectives. MagentaSectionBreakBar800

Three factors drive social media investments.

While there may not be a single formula for social media success, there are three major dynamics a brand should consider before choosing to invest in a given platform:

• Channel: The sheer scale of Twitter and Facebook has made them essential platforms for any major brand. Investing in a presence on second-tier social channels, though, should not be automatic. Marketers should consider the strengths and weaknesses of each platform and how they affect overall social media goals. For instance, Facebook may not be a great awareness driver without a substantial investment in building a fan base but can be a strong platform for deepening fan relationships. For some of these channels having a paid presence is key in order to grow a fan base.
• Industry: The ways users interact with brands differ from category to category. Consumers usually behave differently when dealing with Citibank than they do with American Airlines, for example. Additionally, industry regulation can affect how a potential relationship between a user and brand is structured. For instance, Twitter added an age verification feature for liquor companies to weed out minors. For publicly traded companies, the SEC has provided recent guidance on how to disclose material information to investors. A brand’s position relative to its competitive set may also affect its social media strategy. For instance, a brand that is number two in its category may lack awareness but have an extremely passionate fan base. In that situation, perhaps the goal is deepening the relationship to these fans, and encouraging them to evangelize on behalf of the brand.
• Objective: As mentioned previously, the social media strategy should align with the overall business objectives. For example, a company with a reputation issue may focus its social media efforts towards shifting brand perception. Awareness may already be high, but if sentiment is negative, social media can help change that. Brands can also have multiple objectives, and need to keep that in mind when measuring impact.

Not every social media platform is right for every brand.

We ranked social media sites across several criteria: the ability to target by demographics, lifestyle and location; the diversity of content forms available; and their reach.

This framework is merely a tool to guide social media investment. Audience demographics and usage should also impact a brand’s decision on which social media platforms to develop a presence on and how. For instance, while Instagram scores relatively low for its ability to slice and dice its users by demographic, its audience is overwhelmingly female, (at 68 percent) and young, with 89 percent of users 35 year old or younger according to mobile measurement company AppData. It is also popular among African-Americans, Latinos and urban residents, according to the Pew Research Center’s Internet & American Life Project. For brands seeking to reach these demographics, building an Instagram following may be more critical than Vine or Tumblr.

Controlled experimentation can help inform investment in social media.

Unlike other forms of digital media, testing a social media platform is more complicated. While in most other forms of digital media, a brand can control its presence, turning it on and off, or testing with a small subset of subscribers, in social media a brand can’t invest in building a presence, growing a fan base and developing content only to abandon the platform if it isn’t working. “Once you have a presence and a community, you can’t just shut it down,” warns Senior Product Strategist Tyler Starrine.

While orchestrating a perfect controlled experiment in social media is impossible, smart marketers understand the need to spend their budgets on the most impactful channels and platforms. Experimentation requires analytic expertise to do well — statistical skills and experience in research design and scientific method–as well as executive level support. It also requires understanding of the environmental variables in play on different social media platforms for not just your brand but also competitors and the overall industry ecosystem. With those pieces in place, it’s possible to design a controlled experiment that allows a brand to expose audiences to a media mix, measure variances in outcome and calculate the incremental effect of a particular social media strategy.

A simple approach for this analysis is to implement a pre/post study. The starting point looks at the NPS for customers who are also fans of a brand, as well as customers who are not fans. Comparing the change in both groups over time allows a brand to understand how fan interaction with the brand on a given social channel is impacting NPS. Additionally, comparing the difference between those two groups against other competitors in the space can help the brand benchmark its growth against natural growth in the marketplace. While this framework doesn’t provide a specific “ROI metric,” it does measure the impact of social media efforts on the customer relationship, allowing the brand to better track successes and correct mistakes.

The chart above analyzes data from LoudDoor to better understand the impact that being a fan of a brand has on its customers. Using LoudDoor data we calculated the NPS for customers of a brand who were Facebook fans over the course of a month. We then looked at customers who were not fans and tracked the same metric and compared the results. We analyzed data from the overall population, as well as two segments, “Soccer Moms” and “Millenials,” focusing specifically in the automotive category.

Some key insights emerged from this data. First, there is enough change in consumer mindset over the course of a month to use this as a means to determine the value of social media activity. Further experimentation is required to determine the value of social media interactions. Second, specific industries differ from the norm. The changes we observed in the automotive category were much more pronounced than those across all industries. Finally, demographic segments differ greatly from each other in their responses, telling us that just measuring the overall impact of a social campaign may hide important variations among different groups.

Conclusion.

Getting to social ROI isn’t easy. The number of platforms, the diversity of metrics, the lack of currency, and social’s hybrid role as content channel, PR platform and CRM tool makes it difficult to come to a single solution. This is further complicated by the fact that most platforms have limited (though improving) analytics tools that are not designed to speak to each other. By moving beyond traditional ways of understanding and measuring “ROI” and comparing outcomes against specific business objectives, brands can tap into the real value of social. Using a combination of brand tracking measures, such as Net Promoter Scores, and applying advanced attribution techniques, such as Agent Based Modeling, can help a brand optimize its efforts through the social channel.

Further, brands must understand which platforms are right for a specific goal and also how to balance paid media aimed at acquisition and owned media aimed at engagement and virality. Experimenting with reducing and increasing paid media to understand its impact can help brands find this balance and has minimal risk.