While social media usage generates vast quantities of data that advertisers and marketers can use for customer profiling and targeting, many brand marketers still rely on expensive and often dated survey-based market research. That’s because, in short, Facebook can tell you who and what a user likes, but it can’t tell you why.
The data on which marketers’ models rest are still largely proxies for the emotional and attitudinal roots of that behavior. Understanding those aspects of consumer behavior is still largely the province of focus groups and small-scale, in-person research studies that are often difficult to project up to a mass market, making predicting and shaping consumer behavior still more art than science.
Brand marketers are facing the following:
- Most of the audience targeting and programmatic sales platforms available to advertisers today were not built with brand marketers in mind. They’re very good at identifying and targeting consumers who might be in the market for a new car or an airline ticket, based on behavioral cues, but brand marketers – unlike direct marketers – aren’t necessarily looking for that sort of immediate sales conversion.
- For brand marketers to gain maximum value from digital platforms, they will increasingly need to be able to automate the sort of qualitative audience insights on which brand campaigns are based in order to develop effective programmatic media-buying plans.
- Companies like Resonate, Choicestream, and Cubeyou offer tools that, to varying degrees, integrate traditional market research techniques with modern social media and digital tracking and profiling to enable marketers and advertisers of all types to go well beyond demographic targeting. Marketers crafting campaigns based on attitudes and lifestyles as well as demographics – especially those interested in buying digital advertising media programmatically – should evaluate these tools for adoption.
Image courtesy of PJPhoto69/iStock.
Campaigning: from politics to branding
While the promise of one-to-one marketing may never arrive – or prove less than cost effective for all but the most expensive products and services – digital marketing technologies have proven extremely effective in direct marketing, classified ads, couponing, and the like. However, they have yet to have a huge influence on brand marketing. Part of the problem is due to customer profiling limitations. Marketers spend a great deal of effort trying to translate social media and behavioral data cues into intelligent guesses at basic demographics like gender, age, and income level.
Understanding customer attitudes, values, and lifestyles still depends on traditional – not to say “old-fashioned” – market research techniques like phone surveys and focus groups. That is starting to change, with the adoption of real-time data analytics and programmatic ad buying and selling, where marketers buy audiences rather than media properties via highly automated processes and ad networks or exchanges.
That kind of automation will likely become more critical to brand marketers as they refocus more of their efforts from traditional TV channels to digital platforms. As discussed in a previous Gigaom Research report, Online Video Courts TV Dollars, brand marketers increasingly are looking to leverage digital tools like programmatic ad placement and dynamic ad insertion to improve the targeting of their messaging as well as the ROI on their spending. As more premium video inventory has become available online, brand marketers are beginning to shift ad dollars out of their TV budgets and into digital.
One change agent is Resonate, a startup based in the Washington, DC, suburb of Reston, VA, which launched in 2008 and cut its teeth helping campaigns target potential voters. At the time, however, its founders were not looking for careers in politics. The two principals, CEO Bryan Gernert and COO Andy Hunn, each had backgrounds in technology, not policy issues, having both previously worked at Cybertrust, an e-commerce and information security firm acquired by Verizon Business in 2007. Some members of the startup’s board of directors, however, had backgrounds in politics and issue advocacy, and were looking for ways to target voters at scale based directly on their values, attitudes, and beliefs rather than on rough approximations derived from demographics or online and offline behavior. So Resonate set out to build values-based tools that could collect data at a large enough scale to be used for targeting.
The company tried a number of different approaches, and has continued to refine its collection methodology over time, but today sits on an enormous database that enables Resonate to map observed behavior to a rich matrix of consumer values and attitudes that shape that behavior, and to predict (or forecast) behavior based on those values.
The data come from three principal sources:
- Anonymous behavioral data: Resonate uses a pixel placed on web pages to track browsing habits and now claims to have data on roughly 90 percent of the online U.S. population, collecting it at a rate of roughly 2 terabytes of data a day.
- In-depth surveys: The company recruits 15,000 to 20,000 individuals per month, or roughly 200,000 per year, to participate in in-depth (roughly 30 minute) online surveys that ask detailed, attitudinal questions about wide range of political and marketing-oriented topics. About 80 percent of the survey stays the same from month to month, allowing for longitudinal analysis, while the rest is made of rotating groups of questions on specific topic areas and client-specific items.
- Household data: demographics, voter registration, shopping behavior, etc., mostly purchased from third-parties.
All of that is then combined to produce a taxonomy of over 4,500 attributes, which can be combined in to analyze consumer behavior and attitudes in detail. Categories range from personal values (e.g. family and friends, security and well-being, personal happiness, sense of community) to political ideology and issues (e.g. energy policy, gun rights, taxes, entitlement programs, LGBT rights) to product preferences and lifestyles (e.g. daily fitness routine, car-pooling, green engagement, daily prayer) as well as biographical data, and data on media usage and retail purchase behavior.
Though many of the insights into consumer values provided by Resonate are not new, and have been the subject of marketing research for years, the combination of granularity, scale, and depth in the data Resonate collects allows it conduct what amounts to qualitative research using quantitative methods.
The amount of behavioral data it collects, and has been accumulating over years, means Resonate can identify the users of nearly any brand’s products. It then develops a behavioral and demographic profile of those users and can quickly map those attributes to a set of values and attitudes those users are likely to have in common. Today, Resonate’s business is split roughly 50-50 between political work and working with brands.
Tapping social media
The major online ad players, including Google, Facebook, AOL, and Yahoo, are all actively courting brand marketers and have enormous data sets of their own to bring to bear on providing qualitative audience insights.
Like Resonate, Facebook is thinking about politics as well. It is mining its users’ public posts to try to divine their political sentiments and is now making those data available to ABC News and BuzzFeed to help guide their electoral coverage. While the behavioral data Facebook has access to lacks the qualitative richness of what Resonate collects through its long-form surveys, the sheer volume of data churned out by the social networks’ 1 billion-plus users could potentially allow Facebook to surface useful insights into their value and attitudes. As BuzzFeed editor-in-chief Ben Smith put it in a blog post, while sentiment analysis is tricky – it’s not good at recognizing sarcasm, for instance – Facebook has access to an unprecedented “sample size” of natural language to pore over algorithmically.
For now, Facebook is focusing on potential editorial uses for its political sentiment analysis. Changes to the Facebook API announced in April will make it harder for apps to access a user’s friends. While the changes affect all categories of apps among the most directly impacted will be those developed by political campaigns.
The fire hose of data pouring out of social media networks is also enabling marketing tech entrepreneurs to surface qualitative insights using quantitative methods. Silicon Valley-based Cubeyou tracks the interactions of 250 million Facebook users with over 55 million products, services, brands and places daily. It then uses a predictive algorithm to convert those tracking data into detailed customer types, including hobbies, lifestyle, demographics, psychographics, and personality types.
The company compares its approach with traditional telephone survey-based market research firms like MRI – which also got its start in political campaigning. Customer profiles that once took months of in-person surveys and analysis to develop, Cubeyou claims can now be derived in near-real time and mapped to their media usage and preferences to help marketers shape and target their messaging efficiently. While Cubeyou’s data set is huge, it is limited to Facebook users and their interactions with brand pages on Facebook. The company says that focus is to avoid duplicating user profiles in its audience. Customers use Cubeyou more to qualify and quantify audience profiles, and translate that into the media planning, whether that’s online or traditional media.
Supporting programmatic buying and selling
In contrast to Cubeyou’s approach to media planning, Boston-based Choicestream’s demand-side platform is geared to optimizing online campaigns using fully automated programmatic techniques. After building a base of technologies and expertise for direct marketing and re-targeting audiences based on behavioral data, it’s increasingly focusing on the branding-oriented top of the classic marketing funnel. Choicestream has expanded its offerings to support ad creative, as well as media planning and buying, and is beginning to dabble in programmatic “native advertising” that blends content and promotion.
Choicestream’s proprietary modeling engine uses machine learning to scan and segment in real time the 50-60 million daily users its network covers. Its customers use short polls that are often embedded in an ad unit to tease out attitudinal information, especially for B2B branding. Its B2B customers tend to be in telecommunications, networking, logistics, and recruiting. Some budget up to 20 percent of a campaign’s spending to fine-tune creative and messaging dynamically.
Choicestream’s infrastructure is integrated with Facebook’s FBX platform, and can support ads within the Facebook news feed or on the right rail. Its execs concede that it’s not always easy to work with the Facebook plumbing, and Facebook inventory using Choicestream varies in performance effectiveness: Travel, for instance, far outperforms retail.
Key takeaways
Profiling and ad-targeting techniques based on customer attitudes and values – on top of traditional demographic measures – will be critical to online brand marketing.
Large-scale data gathering and new ways of thinking about qualitative research and targeting are beginning to mature. They are leveraging social media with old-school market research techniques, and are being optimized to support programmatic ad buying.
Much of the pioneering work has been done by startups with roots in politics. But as brands increasingly look to automate the profiling and media buying process, major online ad platforms with roots in search and social networking are also well positioned to capitalize on the trend.
About Paul Sweeting
Paul Sweeting is a Gigaom Research analyst, and is the founder of Concurrent Media Strategies, a Washington, D.C.-based consulting and editorial services firm specializing in digital media technology and policy issues. In 2007 he developed and launched Content Agenda, a website owned by Reed Business Information, the publisher of Daily Variety, Broadcasting & Cable, Video Business, Publishers Weekly, and other media-related properties. He left RBI in 2009 and launched the Media Wonk blog, which examined the impact of digital technology on the way cultural products are created, communicated, and perceived, both in commercial terms and as a cultural and political phenomenon. In 2010 he launched Concurrent Media Strategies and the Current Media website, which incorporated the Media Wonk blog.
About Gigaom Research
Gigaom Research gives you insider access to expert industry insights on emerging markets. Focused on delivering highly relevant and timely research to the people who need it most, our analysis, reports, and original research come from the most respected voices in the industry. Whether you’re beginning to learn about a new market or are an industry insider, Gigaom Research addresses the need for relevant, illuminating insights into the industry’s most dynamic markets.
Visit us at: research.gigaom.com.
