Which Digital Channels are Best for your Marketing Strategy?
A predictive model that can help answer the question
In a world that is increasingly digital, personal interactions become an (ironically) fresh way to distinguish your brand from others. Throwing an event allows you to cut through our cluttered digital lives while also personifying your brand or product. In Spredfast’s case, we’ve doubled down on events to show our commitment to our customers. We’re more than a software company, we’re partners who want to see our customers succeed. We’re putting a face behind the calls, emails, and chats. Hosting events has allowed us to embed personal touch-points as an extension of our commitment to customers.
In preparation for Spredfast Summit 2016, we created an integrated marketing campaign, utilizing all of our digital channels to create demand for our event. Our goal was to drive awareness and conversions/event registrations amongst our key audience of current and potential customers.
The main challenge was determining the overall cost-effectiveness of our efforts—and which channels were most efficient in driving clicks that converted into registrations. We needed to confidently answer the classic advertising quote, “Half of the money I spend on advertising is wasted; the trouble is I don’t know which half.” We needed to continually report on the impact of our traffic due to the length of the campaign. Understanding that sponsored promotions on all networks help drive overall awareness, we needed to be able to distinguish the strengths and weakness of our media strategy, separating the networks that drive awareness from the networks that actually drive conversion.
Digital Channels to Get the Ultimate Goal
Our media strategy was to use Facebook, LinkedIn, Twitter, Instagram, and Google Adwords to drive traffic to the Summit landing page. E-mail was a key channel in driving registrations with our established audience. We also employed a variety of retargeting campaigns, custom audiences built from email lists, and custom audiences based on job titles and companies for social. On Google Adwords, we targeted relevant keywords.
With Spredfast’s established audience, email was a key channel for driving registration.
Our initial solution for conversion rate optimization was examining sources that lead to registration button clicks. Since this is a paid event, we used a secure third-party vendor for transactions; therefore, we were not able see attribution by the channel for completed transactions. Although registration button clicks may be enough for most people to determine where to allocate their budget, we were interested in taking this a step further. Let’s examine the overall scope of the integrated marketing campaign and investigate the impact it is having, in addition to registrations/conversions.
Going Deeper: The Social Media Performance Model
We deferred deep analysis to Dr. Gary Wilcox, Professor at the University of Texas. Wilcox is a recognized expert on social media, advertising research, and branding whose opinions and work have been featured in such national media as The New York Times, Los Angeles Times, U.S. News and World Report, The Huffington Post and AdWeek, among others. In 2011, he began the development of the social media performance model to measure important relationships between marketing efforts and brand key performance indicators (KPIs).
With an understanding of these measurement issues, we used the social media performance model (SMPM) using data provided through Spredfast. This analysis procedure enables improved insights using data science. In this case, the improved insights are to assist conversion rate optimization. In the context of a business-to-consumer brand, executed across organic social media, e-mails and social spend across four platforms (Facebook, Twitter, AdWords and Instagram), the model was used to provide predictive insights. As new insights are derived from the findings, the SMPM enables data-driven strategy that can be used to influence business outcomes achieved not only through organic social media but also other forms of marketing efforts.
Adaptation of the Social Media Performance Model
Previous research using the SMPM examined a wide variety of brands from nonprofits, to B2C and B2B brands. KPIs for this project with Spredfast included website sessions, registrations, and revenue. Website landing page sessions and registrations were obtained from the company’s Google Analytics with revenue calculated based on varying registrations fees. Organic social media vehicles included Facebook, Twitter and Instagram and were collected and tabulated by Spredfast. Paid social was reported in dollars spent by each social vehicle. Email was measured by number of landing page sessions originating from email. All data were reported on a daily basis.
The SMPM utilizes a time-series analysis approach to examine the relationship of the predictor variables on the outcomes variables. In the past, use of measures of website traffic has been found to be useful for modeling campaign effectiveness, especially website visitors/sessions as an accurate predictors of sales. Three analyses were completed – one for website sessions, one for registrations, and one for revenue. Significant predictors for the sessions series are listed in the table below.
For website landing page sessions, two paid social channels exhibited statistically significant relationships – Facebook and Twitter. In addition, Google AdWords and email-driven sessions were significantly related to website landing page sessions. In other words, for this particular Spredfast event, Facebook and Twitter—as well as AdWords and email—stood out in terms of driving substantial website traffic.
Facebook and Twitter stood out in terms of driving substantial website traffic.
The models also provide evidence of the relative importance of the predictor variables in each of the regression series. For example, a 1 per cent increase in Facebook spend was associated with a .83 per cent increase in website sessions, whereas a 1 per cent increase in AdWord spend was associated with only a 0.32 per cent increase in website sessions.
When examining the registration and revenue data, only one individual predictor variable was significantly related – email sessions. The results of the revenue/registration models mirrored each other with only minor differences in fit statistics. However, these final models only explained approximately 25% of the variation in registration and revenue data, indicating that other marketing activities such as social selling or personal emails, not present in the digital model played an important role impacting registrations and revenue.
Know Where You’re Spending Effectively
The SMPM enables data-driven strategy that is beneficial as we analyze our performance during long campaigns and determine future budget allocations. This method enables marketers a way to analyze their data with more confidence. For brands investing in not only social media but other digital activities as well, understanding the importance of these performance indicators is crucial to making the smartest use of your marketing communication budget. Because we’ve seen that email is a significant driver in registrations, we are able to validate our hunch that email is an effective channel to allocate resources to. Since we’ve seen ads on Facebook and Twitter along with Google AdWords have noteworthy impact on website sessions, we’ll likely tweak our organic and paid social scale for upcoming events like Smart Social New York and Smart Social London. Ultimately, the ability to prioritize promotional tactics with proven outcomes may be the most useful business result of them all, allowing any brand to make more effective digital marketing decisions.