What I Learned Writing Best Practices in Data Management

Today data is like water: free-flowing, highly available, and pervasive. As the cost of storing and collecting data decreases, more of it becomes available to marketers looking to optimize the way they acquire new customers and activate existing ones. In the right hands, data can be the key to understanding audiences, developing the right marketing messages, optimizing campaigns, and creating long-term customers. In the wrong hands, data can contribute to distraction, poor decision-making, and customer alienation. Over the past several weeks, I asked over thirty of the world’s leading digital data practitioners what marketers should be thinking about when it comes to developing a data management strategy. The result was the newly available Best Practices in Data Management report. A few big themes emerged from my research, which I thought I would share:

Welcome to the First Party

Digital marketing evolves quickly but, for those of us working as digital marketers or publishers for the past 10 years, we have seen distinct waves of transformation impact the way we use data for audience targeting. Early on, audience data was owned by publishers, who leveraged that data to control pricing for premium audiences. The Network Era quickly supplanted this paradigm by leveraging tag data to understand publishers’ audiences better than the sites themselves. Buying targeted remnant inventory at scale created new efficiencies and easy paychecks for publishers, who found themselves completely disintermediated. The DSP Era (which we are still in) continued that trend, by completely separating audiences from media, and giving even more control to the demand side. Today, the “DMP Era” promises a new world where publishers and advertisers can activate their first party data, and use it for remarketing, lookalike modeling, and analytics.

The ubiquity of third party data (available to all, and often applied to the same exact inventory) makes activating first party data more valuable than ever. Doing so effectively means regaining a level of control over audience targeting for publishers, and being able to leverage CRM data for retargeting and lookalike modeling for the demand side, as well as a deeper level of analytics for both sides. If there has been one huge takeaway from my conversations with all of the stakeholders in the data-driven marketing game, it is that getting control and flexibility around the use of your own first-party data is the key to success. As a marketer, if you are buying more segments than you are creating, you are losing.

The New Computing Paradigm

In order to successfully activate all of the data your company can leverage for success takes a lot of work, and a lot of advanced technology. Whether you are a publisher trying to score audiences in milliseconds in order to increase advertising yield, or an advertiser attempting to deliver a customized banner ad to a prospect in real-time, you need to store an incredible amount of data and (more importantly) be able to access it at blazing speeds. In the past, having that capability meant building your own enormous technology “stack” and maintaining it.  Today, the availability of cloud-based computing and distributed computing solutions like Hadoop has created a brand new paradigm or what former Microsoft executive and current RareCrowds CEO Eric Picard likes to call the “4th Wave.”

“Being a Wave 4 company implicitly means that you are able to leverage the existing sunk cost of these companies’ investment,” says Picard. That means building apps on top of AppNexus’ extensible platform, leveraging Hadoop to process 10 billion daily transactions without owning a server (as Bizo does), or simply hosting portions of your data in Amazon’s cloud to gain speed and efficiency. As digital marketing becomes more data intensive, knowing how to leverage existing systems to get to scale will become a necessity. If you are not taking advantage of this new technology paradigm, it means you are using resources for IT rather than IP. These days, winning means applying your intellectual property to available technology—not who has the biggest internal stack.

Social Data is Ascendant

One of the most interesting aspects of data management is how it is impacting traditional notions of CRM. In the past, digital marketing seemed to end below the funnel. Once the customer was driven through the marketing funnel and purchased, she went into the CRM database, to be targeted later by more traditional marketing channels (e-mail, direct mail). Now, the emergence of data-rich social platforms had actually created a dynamic in which the funnel continues.

Once in the customer database (CRM), the post-purchase journey starts with a commitment beyond the sale, when a consumer joins an e-mail list, “friends” a company’s page, follows a company’s Twitter account, or signs up for special offers on the company’s site. The next step is an expression of social interest, when the consumer agrees to make public his like for a company or brand by “friending” a company’s page, following a company’s Twitter account. Beyond the “like” is true social activation, wherein the consumer actively (not passively) recommends the product or service, through commenting, sharing, or other active social behaviors. The final step is having the consumer sell on your behalf (directly via affiliate programs or, in the softer sense, as a “brand ambassador”).  This dynamic is why Salesforce has acquired Radian6 and Buddy Media.

For digital marketers, going beyond the funnel and activating consumers through social platforms means understanding their stated preferences, affinities, and that of their social graph. Most companies already do this with existing platforms. They real key is tying this data back into your other data inputs to create a 360 degree user view. That’s where data science and management platforms come in. If you are not ingesting rich social data and using it to continually segment, target, expand, and understand your customers, you are behind the curve.

[This post originally appeared on the EConsultancy blog. Get the paper here.]

Can you Buy “Brand?”

SreeUnderstanding Social Affinity Data

Marketers are increasingly turning to social platform data to understand their customers, and tapping into their social graphs to reach more of them. Facebook “likes” and Twitter “follows” are religiously captured and analyzed, and audience models are created—all in the service of trying to scale the most powerful type of marketing of all: Word-of-mouth.  With CRM players (like Salesforce, who recently acquired Buddy Media and Radian6) jumping into the game, digitally-derived social data is now an established part of traditional marketing.

But, are marketers actually finding real signals amid the noise of social data? In other words, if I “like” Lady Gaga, and you “like” Lady Gaga, and my ten year old daughter also “likes” Lady Gaga, then what is the value of knowing that? If I want to leverage social data to enrich my audience profiles, and try and get the fabled “360 degree” view of my customer, “likes” and “follows” may contribute more noise than insight. I recently sat down with Colligent’s Sree Nagarajan to discuss how brand marketers can go beyond the like, and get more value out of the sea of social data.

Colligent (“collectively intelligent,” if you like) goes beyond “likes” and actually measures what people do on social sites. In other words, if you merely “like” Lady Gaga, you are not measured, but if you post a Lady Gaga music video, you are. By scraping several hundred million Facebook profiles, and accessing the Twitter firehose of data, Nagarajan’s company looks at what people are socially passionate about—and matches it against other interests. For example, the data may reveal that 5% of Ford’s socially active fanbase is also wild about NASCAR. That’s great to know. The twist is that Colligent focuses on the folks who are nuts about NASCAR—and like Ford back. That’s called mutual engagement and, arguably, a more powerful signal.

Nagarajan’s focus on this type of data has many questioning the inherent value of targeting based on social media membership. “In any social network’s lifecycle, likes (or ‘follows’ or friends) start out as genuine signals of brand affinity. However as more and more like the page their audience gets increasingly diluted, making likes less of an indicator of brand’s true audience. True engagement as measured by comments, photo posts, re-tweets, hashes, etc. became much better indicators of brand affinity and engagement.”

Colligent data recently convinced Pepsi to choose Nicki Minaj as their spokesperson, since the data revealed a strong correlation between socially activated Pepsi and Minaj fans. Think about that for a second. For years, major brands have used softer, panel-based data (think “Q Score”) to decide what celebrities are most recognizable, and capture the right brand attributes. Now, getting hard metrics around the type of people who adore your brand are just a query away.  Digital marketers have been talking about “engagement” for years, and have developed a lexicon around measurement including “time spent” and “bounce rate.” Social affinity data goes deeper, measuring true engagement. For Nagarajan, “In order for the engagement to be truly effective, it needs to be measured from both sides (mutual engagement). The parallel is a real-world relationship. It’s not enough for me to like you, but you have to like me for us to have a relationship. Mapped to the brand affinity world, it’s not enough for Pepsi fans to engage with Nicki Minaj; enough Nicki fans have to engage with Pepsi (more than the population average on both sides) to make this relationship truly meaningful and thus actionable. When true engagement is married with such mutual engagement, the result is intelligence that filters the noise in social networks to surface meaningful relationships.”

So, what else can you learn from social affinity data? With so many actively engaged fans and followers, throwing off petabytes of daily data, these networks offer a virtual looking glass for measuring real world affinities. If you think about the typical Facebook profile, you can see that many of the page memberships are driven by factors that exist outside the social network itself. That makes the data applicable beyond digital:

  • Television: Media planners can buy the shows, networks, and radio stations that a brand’s fans are highly engaged with.
  • Public Relations: Flacks can direct coverage towards  the media outlets a brand’s fans are engaged with.
  • Sponsorships: Marketers can leverage affinity data to determine which celebrity should be a brand’s spokesperson.
  • Search: SEM directors can expand keyword lists for Google and Facebook buys using social affinity-suggested keywords.
  • Display: Discover what sites Ford’s socially activated consumers like, and buy those sites at the domain level to get performance lift on premium guaranteed inventory buys.

Are we entering into a world in which marketers are going to use this type of data to fundamentally change the way they approach media buying?  What does it mean to “buy brand?” Sree Nagarajan sees this type of data potentially transforming the way offline and online media planners begin their process. “Much of the audience selection options available in the market today are media based. Nielsen defines TV audience, Arbitron radio, ComScore digital sites, MRI magazines, etc. Brand marketers are forced to define their audiences in the way media measures audience: by demographics (e.g., 18-49 male),” remarks Sree.  “Now, for the first time, social data allows marketers to define audiences based on their own brand and category terms. Now, they can say ‘I want to buy TV shows watched by Pepsi and more generally, Carbonated Soft Drinks audience.’ This will truly make marketing brand-centric instead of media-centric. Imagine a world where brand and category GRPs can be purchased across media, rather than GRPs in a specific media.”

Look for this trend to continue, especially as company’s become more aggressive aligning their CRM databases with social data.

[This article originally appeared in ClickZ on 12/11/12]

Discover more on this topic and others by downloading my new whitepaper, Best Practices in Data Management