Marketing analytics is not a static discipline that, once implemented, never changes. The reasons for doing marketing analytics don’t change much, nor do the core principles that drive analytics work. However, the approaches, techniques, and technologies involved in the practice of analytics are constantly changing. There is a marketing analytics frontier, and those who venture there are discovering game-changing insights that yield significant competitive advantages. This chapter explores some of the trails that the marketing analytics pioneers are currently blazing.

Change in the marketing analytics discipline is actually occurring at a dizzying pace. In general, this change has come in the form of innovations that are providing business and customer insights never imagined possible just a few years ago. The most cutting-edge marketing analytics practices are delivering more precise information at a greater level of detail than ever before, and this information is reaching marketers almost in real time. With this information, marketers are able to understand their customers at the deepest possible level, accurately predicting events like purchases and defections. Marketers are able to optimize their channel and media mix based on data, not intuition. The frontier of marketing analytics promises to transform marketing into an undisputed revenue center, not just a producer of campaigns and enabler of sales.

While there are many changes taking place on the marketing analytics frontier, this chapter focuses on three for which marketers can’t afford to claim ignorance: big data, predictive analytics, and advanced attribution. The rest of this chapter presents an overview of what these things are and why marketers should care about them.

Big Data

Understanding customers and what motivates their behavior has been a quest of merchants since commerce began. Marketing owns this mission, and marketers have traditionally studied their customers through a handful of research techniques, such as past transaction analysis, observation, surveys, and focus groups. Many organizations maintain some sort of master customer information file, often part of a larger data warehouse, that marketers can reference to study customers. This data, however, contains a finite set of information that exists in varying states of currency and accuracy. Although useful, this data gives marketers a limited understanding of customer behavior. For many, this state of customer data is their current reality: old and incomplete.

The modern consumer drops plenty of digital breadcrumbs, and if the marketer can follow them, these would reveal previously hidden patterns, preferences, and signals about their intent. Consider what is typical behavior for a consumer on any given day: they communicate via mobile phone, review and post to social media, check email, browse the Internet, make purchases, download music, and read reviews of products and merchants. Consumers who do these things create an enormous data trail that marketers can leverage, and exploiting this data is at the intersection of marketing and big data.

Big data is conceptually easy to understand. It simply is a set of data so large and varied that it defies processing and analysis through the use of standard database query tools. There is no specific size designation after which a set of data becomes “big.” In this case, “big” refers more to the complexity of analyzing it, rather than the size. Although it is also “big” in terms of size, as the sheer volume is often very large. What makes it large and complex is that it comes from several sources, and it usually isn’t a single, discrete set of data. It’s better understood as a collection of related data from many sources.

Historically, marketers have worked with data that is structured, and therefore easy to access and analyze, using tools as simple as a spreadsheet. In fact, a good way to think about structured data is as any form that will fit into the rows and columns of a spreadsheet. Sales transaction data is a good example of structured data. Almost all of the data that marketers have traditionally worked with in the analytics process is structured. Structured data certainly belongs in this discussion of big data, as it is one of the two broad classes of data that fit under the big data umbrella.

The other class of data, and the one that presents the most challenges, is unstructured data. Some examples of unstructured data include digital images, online videos, blog post comments, online product reviews, chat logs, webinars, location, sensor and GPS data, to name a few. A vast quantity of this type of data exists, and it is growing at an astonishing rate. Unstructured data contains information marketers can use to better understand customers and their sentiments, particularly if they can match this data to preexisting, structured customer data.

How can marketers access and use this data to their advantage? This application dilemma is the challenge and opportunity, because the hype and promise of big data is not about the data itself. The data is of little value without tools and applications for exploiting it. Big data essentially represents a large reservoir of potential energy. Transforming that potential into kinetic energy, to unlock the value it contains, requires analytics.

Standing in the way of big data applications that provide marketing benefits are the three V’s of big data, characteristics described by Doug Laney in a 2001 META Group research note: volume, velocity, and variety.² These help explain why big data requires new forms of processing to allow integration of insights extracted from big data with marketing and corporate decision making.

• Volume: The sheer amount of data that is becoming accessible to marketers, which is quite large. The opportunity of big data lies in having the capacity to process the large volume of data that exists.
• Velocity: The rate data is created or otherwise becomes available. For example, every second, on average, about 6,000 tweets go through Twitter.³ The challenge of velocity is having the capacity to absorb the data at the increasingly rapid speeds at which it arrives.
• Variety: The diversity of structured but mostly unstructured data types and sources. Big data implies the ability to simultaneously take multiple sources, such as video files, images, documents, text, call center notes, and geo-location data, and form a composite, actionable view of that data.

The application of big data analytics isn’t limited to marketing, but within the marketing department, it is driving more intelligent decisions that increase marketing effectiveness, improving return on investment. Some of the ways big data is proving valuable for marketers include:

• Greater information transparency. Information and insights that are normally locked with disparate types of data become accessible and usable for marketers. For example, by exploiting big data, marketers can more precisely segment customers, which leads to better matching of products and services to their needs.
• Personalization. Through big data analysis, marketers can develop a deeper and more sophisticated understanding of individual customers and their buying patterns. With this understanding it becomes possible to develop and deploy highly personalized marketing approaches that better allocate marketing budgets and resources and achieve a higher ROI.
• Predictive modeling. Build accurate models that help organizations understand how customers will respond to certain actions, such as a price change or new terms and conditions.
• Sentiment analysis. Because big data analytics considers data from many different sources, such as social media, it is possible to discern how customers or the public feel about a company, product, or service.
• Behavioral analysis. Using big data to analyze trends helps marketers understand how customers are responding to market conditions, sales, coupons, competitive pressures, or myriad situations as they are occurring. This analysis lets marketing refine strategies and develop messaging that resonates with customers.

These are just a few examples of how marketers can exploit big data. These benefits reveal why there is such interest in big data, and as marketers race to take advantage of analytics, they are encountering some challenges, many of them related to the three V’s identified earlier. Because there is so much data for marketers to consider and use, properly storing and managing it is a challenge. Big data analytics requires different software, tools, and skills to perform, and many marketing organizations discover they don’t have them on the team, let alone anywhere else in the organization. As organizations move to exploit big data, many are finding they have to add a technical skill set to the marketing team. Big data is taking marketing to a place it has never been before, but the promised results are causing increasing numbers of marketers to venture into this strange, new territory.

The people challenge for marketers who wish to make use of big data is a major hurdle. Big data analytics requires people with skills who have never before inhabited the marketing department. The ideal blend of skills for a big data marketing team member includes:

• Mathematical: the skills required to build, understand, and work with what are often complex data sets and models. This is the theoretical aspect of big data analytics. These skills are most often associated with a scientist.
• Technical: the technical proficiency to use software and tools to retrieve, analyze, manage, and store the data from a wide variety of sources. These skills are most often associated with someone in the IT department.
• Insightful: the ability to interpret the output of big data analytics and discern the “big picture” implications, translating them into actions marketing should take. This ability is most often associated with a researcher.
• Communications: the talent and skill of sharing results with the people who need to hear them and inspiring them to take the appropriate actions as a result. This ability is most often associated with a visionary leader.

A quick scan of this necessary skills list produces a realization: not only do people who have these skills not work in the marketing department, few people anywhere have all of these skills. These skills represent a rare combination that is not often found packaged in one person. It’s possible to find the mathematical and technical skills in one person, or the insight and communications skills together, but rarely all four. This people challenge is the greatest one for most marketers seeking to use big data. What many organizations are doing is outsourcing the mathematical and technical skills while working to develop them internally.

Another major area of challenge when marketing goes to use big data analytics is of course the data itself. Harmonizing the many data sources is difficult, because organizations generally have the data marketing wants to use in different formats, structures, and databases. Creating a single, harmonized view of it all that makes it easy for marketing to use is difficult if not impossible. On top of these challenges are concerns and issues with security and privacy, often because some of the data is stored in the cloud or in systems with vulnerabilities. In the rush to take advantage of big data, some organizations may not give enough thought to keeping it secure, and breaches are damaging, both monetarily and reputationally.

Although these risks and challenges seem daunting, they shouldn’t dissuade marketers from testing the waters of big data. All of the risks are manageable, and none of the challenges are impossible to overcome. The opportunities and advantages that big data can provide are simply too great to ignore. Marketers who want to move forward with big data should follow this action plan:

1. Start with a vision. Having a vision is the starting point for successfully leveraging big data analytics. A vision provides direction for the effort, and a vision that is more specific is better. An example of a good vision is understanding which customers are most likely to defect. Ideally, the vision you cast for big data analytics should align with specific business objectives.
2. Get and build skills. Most marketing teams don’t have the in-house skills to pursue a big data analytics initiative. Before going very far down the big data path, assess your skills and determine where deficiencies exist, putting a plan in place to acquire them. Those deficiencies are usually technical skills, which are needed in the early going to design and implement an architecture for big data to support the analytics work. Use consultants and outside services to bridge the skills gap at first.
3. Identify a pilot project. Prove the value of big data by using it to solve a real marketing problem, even on a small scale. A pilot project helps prove the business case for a longer-term investment in big data analytics and also creates the necessary internal support. It brings into clear focus the skills needed to move forward, helps define the right metrics for measuring future success, and clarifies the architectural elements needed to progress further.
4. Design a big data architecture. Based on your vision and goals for big data analytics, map out the necessary architecture for achieving them. The scope of your architecture will include system design, data flows, timing, and data models necessary to support big data analytics in your organization. Evaluate the alternative big data tools and platforms, make recommendations, and acquire them.
5. Present the business case for big data. The experience of a successful pilot project, a skills assessment, and an architectural plan are the necessary ingredients for creating a business case that should win easy approval for stepping up the investment big data analytics.

Big data often gets its start in the business intelligence unit of a company. Companies are using big data analytics to understand and engage customers in a way that inspires greater loyalty. They can make better, data-driven decisions, seizing the opportunity to improve marketing ROI, develop new opportunities, tailor promotions to specific target markets and help optimize marketing’s strategy.

Predictive Analytics

Any conversation about big data and marketing often leads to the topic of predictive analytics. Although these are not interchangeable terms or processes, big data and predictive analytics go hand in hand. Predictive analytics “is the use of statistics, machine learning, data mining, and modeling to analyze current and historical facts to make predictions about future events.”⁴ Although it’s still analytics, it’s a very different kind because it is forward-looking. All other types of marketing analytics tell us what has already happened, perhaps very recently, but still looking backward. Predictive analytics give marketers the ability to know what is going to happen with amazing accuracy. The results that some are getting through the use of predictive analytics seems almost magical.

Most of the automation and analytics that marketers use, while very helpful, provides a historical view of lead generation process performance. In fact, marketers have the ability to understand with great precision how campaigns have performed and where the bottlenecks are in the marketing funnel. What this technology can’t do is serve as a crystal ball that shows the conversion future of leads in the funnel. Marketers find themselves trying to make the most sense of the historical analytics data—looking in the rearview mirror—to squeeze the last drop of performance out of campaigns and lead-generation processes. What predictive analytics does is give sales and marketing that crystal ball: accurate predictions about which leads in the funnel will buy and when.

Predictive analytics is best thought of as “predictive intelligence” that comes alongside marketers as they invent creative ways to connect with prospects and customers through various content, promotions, and offers. It helps marketers understand the propensity to buy for the leads in the funnel, separating the wheat from the chaff from a sales perspective. Predictive analytics forecasts how well customers will respond to those efforts to influence them. It looks at the likely positive and negative outcomes of doing that, showing marketing who is least likely to respond to an offer and who is most likely to buy or respond to an offer.

Predictive analytics, therefore, works on both the expense and revenue sides of the profit equation. By knowing who is least likely to respond to an offer, marketing can save budget dollars by eliminating those prospects in the database from future direct mail or telephone campaigns, perhaps opting to reach out via some less expensive channel. By contrast, when predictive analytics helps marketers understand who is most likely to respond, marketing can increase revenue by presenting offers to those prospects are most likely to accept.

From the organization’s perspective, predictive analytics helps drive revenue. From marketing’s perspective, it helps marketing sustain higher levels of performance. Without predictive analytics, marketing is resigned to a trial-and-error approach to its campaigns, constantly trying to determine the best blend of campaign and media mix parameters to squeeze some incremental improvements out of its efforts. Predictive analytics can shortcut this trial and error process by enabling accurate predictions of how likely, how much, what, and when a prospect will buy.

A technology that is a powerful adjunct to predictive analytics is personalization, the dynamic customization of the digital content that is presented to visitors based on known characteristics and preferences. Consumers experience this on a regular basis when they visit sites like Amazon.com or Netflix. These companies use predictive analytics to understand the interests of their visitors and then suggest other things to view or buy. A spokesperson for Netflix described how the company employs predictive analytics: “We monitor what you watch, how often you watch things. Does a movie have a happy ending, what’s the level of romance, what’s the level of violence, is it a cerebral kind of movie or is it light and funny?”⁵ So far, business-to-consumer marketers have done the best job of using predictive analytics, and business-to-business marketers are learning how by observing them.

The place predictive analytics begins is with the data captured in an organization’s Customer Relationship Management (CRM) and marketing automation systems. This data provides marketers with a digital profile of a prospect based on their interactions with the company’s digital outreach: web visits, email opens, and so on. This is where big data enters the picture: this digital profile is combined with publicly available data from a wide variety of sources, for example, social media posts. Comparisons are made to the characteristics of successful sales transactions, and predictions are made. Although this process is conceptually simple, predictive analytics is about adding more sources of data and analyzing all of it to better predict which prospects are signaling an intent to buy.

Given the capabilities of predictive analytics, it’s not hard to understand why marketers are embracing it. It has, however, largely been the domain of large enterprises, because it takes some resources and expertise to put the infrastructure in place. Until recently, if a marketing team didn’t have the role of “data scientist” on its organization chart, then the chances were slim that it was doing predictive analytics. Now, however, a number of solution providers are offering cloud-based services to help make predictive analytics accessible to the masses. In short, it is becoming more mainstream and easier to do.

Now is the time for marketers to understand the capabilities and benefits of predictive analytics. To embrace the intelligence it provides seems like a no-brainer, but it’s not for everyone. For example, if a company’s products are not bought via a considered purchasing process—one that advances through the stages of need awareness, discovery, consideration, and decision—then predictive analytics is not a great fit. In other words, for impulse-purchased products, predictive analytics won’t provide much value. When a company’s solutions are acquired via a considered purchase process, there are still readiness issues to consider. Chief among them is the cultural readiness necessary to commit to and get success from predictive analytics. Analytics-averse organizations are far from ready to take the plunge; a data-driven orientation is a prerequisite for success.

Advanced Attribution

An ongoing challenge for marketers is proving how their efforts are responsible for generating business results. Marketers intuitively understand that their work bears some fruit, but the degree to which it does is often hazy because it is hard to know with precision how campaigns, advertising, and promotions blend together to produce results. If marketing ran a single campaign using a single media channel, then the task of attributing business results to these efforts is simple. But modern marketing takes place in a complex arena, with an intricate customer journey that is largely hidden from the marketer. It occurs across multiple devices and through mixed media types delivered through multiple channels. How can marketing know with any certainty which part of that media stream produced the result? Marketers know that some part of the stream of media it produces has influence, but which part? The knowledge of how to accurately attribute results to efforts would provide a powerful means of optimizing marketing efforts.

Back in the pre-Internet days, marketers could drop a direct mail piece to prospects and wait for responses. It was easy to match responses to the mailing and determine the conversion rate. The analytics to determine cost per customer acquisition and the ROI for the campaign were also simple. It wasn’t difficult to know what to attribute campaign conversions to, because the direct mail piece was the only element in the media stream that could produce it.

Today, the Internet is the primary marketing platform, and the number of online channels available to marketers continues to proliferate. Marketers use many of them simultaneously, and the more they use, the more difficult it is to get clarity about which components are most responsible for conversions and which ones are not pulling their weight. What is known is that the collective influence of the marketing media stream produces conversions. But which part did the best job of creating awareness? Which was best at triggering the purchase? The fallback position for many marketers has been simply to assign full conversion credit to the last piece of media in the stream, no matter how diverse the media stream was. This “last click” method of attribution, however, is imprecise.

Marketers need to know much more about how all their media perform so they can optimize the whole stream. Simply attributing credit for conversion to the last clicked element in the media steam is a wholly inaccurate approach, because the volume of media in use by most marketers is substantial and ­varied. The longer the media stream and the more varied it is, the less likely that last touch attribution accurately reports which part was most responsible for a conversion. The variety of channels makes accurate attribution difficult, and further complicating the attribution dilemma is the way customers consume content across multiple devices. A sales transaction may start online through a computer but end up being completed on a mobile device.

To understand this challenge, consider the scenario in which a consumer sees paid search media on her computer at work, leading her to visit a seller’s website, where she downloads a free buyer’s guide. The next day, that consumer receives an email from the seller with a discount offer, and as she is riding the train home from work, she uses her mobile phone to click on the offer, investigate it, and ultimately make a purchase. In this scenario, which element of the media stream gets credit for the conversion? The downloaded buyer’s guide might have been highly influential, but if last-click attribution is in use, the email will receive credit for the conversion. This leads to erroneous decision making about which part of the media stream to optimize or where to invest further.

Marketers need a holistic method that considers everything the consumer or buyer sees, the entire media stream from the customer’s perspective, accounting for all channels and devices. With this information, marketers can accurately understand the influence of each media stream element in a conversion. Getting this view of media performance is necessary because a customer conversion is driven by all that a customer sees, not just one part of it. The great news for modern marketers is that there is a way to get this kind of information about their campaigns, and it’s called advanced or data-driven attribution.

Advanced attribution is an approach for properly and precisely allocating fractional conversion credit across the entire media stream that customers are exposed to on their buying journeys. Marketers can understand how their content and media choices are helping produce conversions, and the extent to which each component in the media stream contributes to producing conversions. This level of attribution is something that marketers have longed wished for, but never thought possible. Advanced attribution is no longer in the realm of science fiction or wishful thinking: it’s here!

By using advanced attribution, marketers can know which components of their campaigns generated conversions, and even how long before the conversion event those components influenced the conversion. This information makes it much easier to distinguish between media elements that created awareness, which ones promoted interest, and which ones were most influential in the conversion. By using advanced attribution, marketers gain tremendous clarity on what’s going on with their campaigns. This information allows them to make smarter decisions about which elements of the media stream are most effective at the various stages in the buying cycle, enabling more effective use of them to get better response and conversion rates.

The capabilities of advanced attribution allow marketing to answer a series of questions that once instilled fear:

• How does each marketing channel fractionally or incrementally contribute to the results marketing produces?
• If marketing gets an increased budget for campaigns or promotions, where is the best place to spend it?
• If marketing gets a decrease in budget for campaigns or promotion, where will cuts least affect the outcomes?
• How do differences in the key campaign variables—creative, devices, audiences, and tactics—produce different outcomes?
• How do these key campaign variables interact or work together to produce the outcomes?
• What is the optimal blend of tactics and media channels to achieve the best possible business result?

For the marketing team using advanced attribution, when their executives want to know how much the group needs to spend to generate a needed amount of revenue, the answer is no longer an educated guess.

So far, the discussion of advanced attribution has assumed that everything marketing is doing is in the digital realm. In practical application, marketing campaigns frequently blend digital and traditional (offline) media to generate conversions. For example, a retailer might exploit several digital channels in a campaign or promotion, in addition to sending coupons through the mail, which are tracked by the retailer when customers redeem them. The coupon redemption data is added to what is also known about how those customers were influenced by the digital media to which they were exposed. In this way, the retailer gets a complete picture of how the channels worked to influence customers. The scope of advanced attribution isn’t limited to just digital media, but can include traditional media, too, as in this example. It can provide a very holistic view of everything that influences conversion.

The competitive implications of advanced attribution are enormous. For markets where the key players rely heavily on digital media, advanced attribution has the ability to tilt the playing field in favor of those who use it. Those who don’t use advanced attribution are caught in an ongoing effort of straining to see incremental improvements in media performance, because they’re making media decisions that don’t always work. Those marketers that do use advanced attribution have clarity about conversion credit, allowing them to make smarter media choices with proven results. Advanced attribution provides a new window of understanding about where conversion credit truly belongs. It puts a powerful lever in the hands of marketing for generating sales lift from every channel in use.

The three types of analytics presented in this chapter are currently the frontier of marketing analytics, but this won’t be the case for long. All of these solutions are in current use, sometimes in combination. The increased use of big data, predictive analytics, and advanced attribution by marketers is moving awareness of their benefits up, while costs and ease of use are getting better. Already these approaches are well within reach of medium and even small companies. They are no longer the exclusive domain of large, well-funded enterprise marketing teams. The advantages from using these—and most innovative solutions and approaches—comes with being agile and deploying them quickly. In other words, the longer it takes to exploit them, the more mainstream these approaches become, and the less differentiation they provide.

Today’s frontier is tomorrow’s civilized territory. So it is with big data, predictive analytics, and advanced attribution: there will come a day when “everyone’s doing it.” Marketers looking to manage their performance to higher levels and transform marketing into a true revenue center should always be scanning the horizon, looking for the next innovation to reveal itself.

__________________________________

¹See http://siliconangle.com/blog/2012/06/15/geoffrey-moore-discusses-big-data/.

²Douglas Laney, “3D Management: Controlling Data Volume, Velocity and Variety,” February 6, 2001, META Group. http://blogs.gartner.com/doug-laney/files/2012/01/ad949-3D-Data-Management-Controlling-Data-Volume-Velocity-and-Variety.pdf.

³See http://www.internetlivestats.com/twitter-statistics/.

⁴Mick Hollison, “Love or Hate It, Why Predictive Analytics Is the Next Big Thing,” Inc. Magazine, September 17, 2014. http://www.inc.com/mick-hollison/love-or-hate-it-why-predictive-analytics-is-the-next-big-thing.html.

⁵Louis Gudema, “How Predictive Analytics for B2B Sales and Marketing Can Offer Huge Returns,” Econsultancy blog, February 19, 2014. https://econsultancy.com/blog/64364-how-predictive-analytics-for-b2b-sales-and-marketing-can-offer-huge-returns/.