From Big Data to Big Marketing: Seven Essentials - FICO

0 downloads 261 Views 878KB Size Report
Big Data into best marketing decision—what we're calling “Big Marketing.” ... Keys to making Big Data analytic ins
»» insights

From Big Data to Big Marketing: Seven Essentials Best practices for performing data-driven personalized marketing at an immense scale

Number 63—November 2012 The ability to draw deep customer insights from Big Data and bring them rapidly into operational decision making is transforming the discipline of marketing. Today marketers have the opportunity to understand customers at a depth that harkens back to the past of neighborhood stores and local banks. To have a deep, relevant and constantly improving dialogue with these customers. And to do so at an immense scale. Intelligent consumer devices, ubiquitous broadband networks and social media are all contributing to the explosion of Big Data. Standardized hardware components and service-based software architectures are enabling this data to be brought together and analyzed in massively distributed, parallel and virtual ways that bring down the cost of high-performance computing. As a result, Big Data techniques—used thus far in very specialized settings—are affordable and implementable for an increasing number of companies.

Big Marketing is increasing campaign effectiveness by 10 to 20 times

Leveraging new technologies, marketers can now make relevant personalized offers to individual customers in the manner and at the moment most likely to elicit profitable responses. Every offer can also fully balance the customer’s shopping propensities with the multiple objectives and constraints of the business (product, brand, enterprise) and its trade promotion partners. This white paper examines seven essentials for transforming the potential of Big Data into best marketing decision—what we’re calling “Big Marketing.” We’ll cover:

•  Why now is the moment of opportunity for Big Marketing. •  How Big Data techniques help overcome data silo obstacles to make customer-level decisions. •  Creating intelligent behavioral triggers that outperform simple business rules. •  Getting answers to “million dollar” marketing questions that increase ROI. •  Keys to making Big Data analytic insights immediately useful in operations. •  How to get a head start with configurable prebuilt applications.

www.fico.com

Make every decision countTM

From Big Data to Big Marketing: Seven Essentials

»» insights »» The Need for Big Marketing Now

“Big data can deliver higher margins and productivity ... Marketing levers can affect 10 to 30 percent of operating margin.” —Big data: The next frontier for innovation, competition, and productivity McKinsey Global Institute, May 2011

“Technology promises fundamentally to alter the economics of low-margin banking staples such as processing payments ... Rather than merely generating an instruction to move money that might be worth a few small coins, the information that comes with such a payment might open up new sales and advertising opportunities that could be worth hundreds of times as much.” —Retail renaissance The Economist, May 2012

Big Marketing techniques and technologies are delivering concrete business value for companies across multiple industries. These leaps in performance—like marketing campaigns that are 10-to20-times more effective than traditional sales offers—come from using analytic-driven insights to generate immensely more intelligent 1-to-1 offers and messages on a grand scale. Such offers are not only more relevant to the consumer but are also optimized to achieve maximum response, conversion and take-up rates. Marketers need this new source of propulsion now. While many have long used information technology and data analysis for competitive advantage, most companies are not yet performing the data-driven customer-centric decisioning needed to do effective personalized marketing, especially at scale. Marketing is still largely operations-centric, with customers receiving offers based more on regularly scheduled promotional programs than on their own individual needs, behaviors and timeframes. Today successful marketers need to do both, but few companies have the capability. That’s a problem since marketers are challenged to break through the wall of noise created by the 16,000-plus marketing messages lobbed at the average consumer per day.1 It’s a problem in the retail industry, for instance, with the sector’s share of consumer spending continuing to decline, and profitability under pressure not only from price-sensitive customers but suppliers as well. It’s a problem too for CPG companies, which can’t succeed by simply keeping a larger share of the profit pie. In fact, according to a 2012 survey by the Grocery Manufacturers Association, McKinsey & Company and Nielsen, “the most profitable CPG companies were five times as likely to collaborate with retailers, tapping into their own consumer knowledge to help retailers reach new audiences, while working with retailers on mutually beneficial pricing and category strategies.”2 The continued dominance of operations-centric marketing over customer-centric approaches is a problem in retail banking too, as financial institutions struggle to retain valuable customers and increase profitability through cross-selling. Meanwhile, internet access via smartphones and other technological and social changes are about to cause an upheaval in retail banking. A recent series of articles in The Economist warns that “these changes will give bank customers more clout, allowing people effortlessly to find the best deals, mainly at the expense of banks’ profits.” They will “overturn” existing financial relationships and “undermine the old model of retail banking.”3 But the articles also point to “big opportunity” for banks that “tap into new sources of revenue by mining their enormous troves of customer data.” To turn this potential into measurable business value, chief marketing officers and other executives must keep their eye on the end game. The end game is the ability to make the best marketing decisions, at the level of individual customers, on the immense scale of the enterprise. It’s the ability to attract, retain and grow their best customers through reminders, rewards and recommendations of value. And it’s the ability to execute these decisions, with utmost reliability

1

“ Precision Marketing: Maximizing Revenues Through Relevance,” p1, by Sandra Zoratti and Lee Gallagher, published by Kogan Page, June 2012

2

2 012 Customer and Management Channel Survey, Grocery Manufacturers Association, McKinsey & Company and Nielsen

3

 “Counter revolution” and “Retail renaissance,” The Economist, May 19, 2012

www.fico.com

page 2

From Big Data to Big Marketing: Seven Essentials

»» insights and consistency, across all channels, products and brand engagements.

“39% of marketers say they can’t turn their data into actionable insight.” —Marketing ROI in the Era of Big Data: The 2012 BRITE/NYAMA Marketing in Transition Study

“A significant barrier to succeeding with big data will be the ability to ask the right questions, and use the right technologies to get the answers.” —Hype Cycle for Big Data Gartner, July 2012

Why is now the time to “go big”? Because Big Data technology and methods have evolved beyond the point where early adopters must stitch together their own solutions from tools and custom software code. Today, companies can get a head start with configurable Big Marketing applications, implemented on-premises or delivered through a cloud-based SaaS (software as a service). These prebuilt applications can be delivering best decisions to operational systems in as little as 120 days.

Figure 1: How Big Data impacts marketing Big Data enables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . personalized marketing. . . . . . . . . . . . . . . . . . at an immense scale. 1. Understand

2. Predict

1. Understand Build more complete, timely view of customer from larger quantity of diverse data 1. Understand 2. Predict 3. Decide

1. Understand 2. Predict 3. Decide 4. Answer 5. Engage

1. 2.Understand Predict 3. Decide 4. Answer 5. Engage 6. 7. Learn Experiment

3. 4. Decide Answer 5. Engage 6. Learn 7. Experiment

5. 6. Engage Learn 7. Experiment

2. Predict 4. Answer

2. Predict Find behavior patterns not evident in smaller, homogeneous data sets 2. Predict

4. Answer

3. Decide Scientifically identify the best actions from trillions of possibilities 6. Learn

2. Predict

4. Answer 6. Learn

4. Answer Explore value-creating questions that balance many complex factors 4. Answer 6. Learn

5. Engage Drive multichannel customer interactions from Big Data insights 6. Learn

6. Learn Adapt faster to customer responses and market change 7. Experiment

7. Experiment Create competitive advantage by pushing the edges of business-as-usual

www.fico.com

page 3

From Big Data to Big Marketing: Seven Essentials

»» insights »» Big Marketing Essentials 1. Understand

Here are seven essentials for turning Big Data into best marketing decisions.

2. Predict

1. Develop a “back to the future” understanding of individual customers The phrase “back to the future” has been used to describe Big Data techniques that are beginning to enable companies to understand their customers in an “old” way. Local businesses used to know their customers’ needs, attitudes and propensities to buy. The corner butcher knew which cuts buyers favored and how much they generally spent on meat. Bankers knew which additional financial products might be useful to each depositor.

Understand 3. Decide

Build more complete, timely view of customer from larger quantity of diverse data

5. Engage

Today analytics is being applied to Big Data in ways that make it feasible for businesses to get to know individual customers again. Only now this knowledge comes through identifying behavior patterns found in the vast quantities of bits and bytes from store point-of-sale (POS) devices, and online and mobile digital activity.

4. Answer The constant stream of data from such activity—abundant, timely and varied—provides the opportunity for marketers to understand their customers more fully. While respecting consumer privacy,4 companies may be able to develop this more complete picture not only from what customers say (in online comments, ratings and profiles, blog posts, tweets, etc.) but also from what they do (data from in-store and online purchases, mobile payments, web searching and browsing, etc.). They can employ a variety of analytic technologies for capturing, cleaning and transforming this diverse data into useful insights. Their conversations with customers can be guided by more complex decision strategies that take this richer view into account. In fact, Big Data techniques can help companies rise above key obstacles to customer-centric marketing posed by traditional channel- and product-specific data management. FICO has helped a CPG company, for example, use such techniques to understand and elevate engagement for more than 11 million consumers across more than a dozen brands. In the past, nobody at the company6.had a complete understanding of how customers were interacting Learn with all of its brands. Now all marketers can benefit from cross-brand insights and cross-pollinate across customer contact points (e.g., social, online, email, mobile and point-of-sale devices). Similarly, retail banks have much to gain from understanding patterns of behavior across products. For example, if a customer is shifting more grocery purchases from a debit card to a credit card, is that a sign of growing risk, a signal of an effective rewards-based credit card promotion or, given everything else known about the customer, an opportunity to offer overdraft protection or expanded credit? By rounding out data from across the enterprise with the expanding range of external data, banks gain even sharper insights into customer lifestyle, lifestage and spending.

7. Experiment

The key to building this cohesive view of the customer from internal and external sources is the ability to handle data that hasn’t gone through the normalization and structuring processes required in traditional relational databases. Alternative approaches to data management (e.g., newSQL, noSQL) thus play an important role.5 They expand the types of data that can be 4

I t is important that marketers be aware that leveraging more diverse data may have privacy implications. A thorough understanding of and compliance with applicable privacy laws are essential to ensuring continued consumer trust.

5

 ewSQL is a class of RDBMS (relational database management systems) that provides scalable performance while N maintaining key principles of traditional single-node RDBMS. NoSQL is a broad class of database management systems that do not adhere to the RDBMS model and generally do not use SQL (structured query language) for data retrieval and manipulation. These include graph and key-value-store databases.

www.fico.com

page 4

From Big Data to Big Marketing: Seven Essentials

»» insights analyzed to include semistructured data, such as web logs that reveal browsing activity, and unstructured data, such as text from Facebook pages, tweets and online product reviews. Other important technologies include high-velocity data stream processing for filtering and capturing incoming data on the fly. The faster the process of capturing useful data and turning it into actionable insights, the more responsive and relevant the conversation with the customer can be. Apache Hadoop, an open-source software framework, is being used to write MapReduce programs that analyze heterogeneous data streams, extracting value from data that would otherwise be difficult to mine.6

“The big-data explosion is driving a shift away from gut-based decision making ... in today’s volatile business environment, judgment built from past experience is increasingly unreliable. “And yet a recent CEB study of nearly 800 marketers at Fortune 1000 companies found the vast majority of marketers still rely too much on intuition.” —“Marketers Flunk the Big Data Test” HBR Blog Network, August 2012, Harvard Business Review

Recent technology developments make it possible to do this high-performance computing in massively parallel processes on distributed architectures of inexpensive, standardized hardware and open source software—bringing down the cost of working with Big Data. Virtual infrastructures, available through cloud-based SaaS providers, are also lowering the entry barriers and reducing startup times. These services enable companies to take advantage of the latest advances in nontraditional data management and processing technologies, while avoiding the time and cost of building them into enterprise infrastructures.

2. Predict

2. Find the real predictors of customer purchasing and other propensities What motivates customers to take action is often the result of a complex, subtle interplay of factors. As companies shift from traditional methods of sending out offers based on operationsdriven programs to generating individualized offers based on customer propensities and behavior, it’s important to avoid using simplistic behavioral triggers based solely on empirical evidence and judgment. Is it effective, for instance, for a bank to create a business rule that says “whenever customers with acceptable risk scores spend more than $400 on tires, send them an offer for an auto loan?” Probably not. The behavior patterns that accurately predict whether a consumer is likely to purchase an automobile in the near future are more complex.

Predict

4. Answer

Find behavior patterns not evident in smaller, homogeneous data sets

The expanding Big Data universe makes it possible to discover behavioral patterns that may not be evident in smaller, more homogenous, more fixed data sets. Model builders have the opportunity to find new attributes highly predictive of certain behaviors, as well as combine them into unique predictors for their company—a potential source of competitive advantage. Many such elements may be incorporated into powerful predictive models.

6

 ap-Reduce is an efficient means of finding, filtering and transforming data over massively distributed file M systems. This efficiency comes from distributing the tasks out to clusters of processors (map), then consolidating their results into one central data list (reduce).

www.fico.com

6. Learn

page 5

From Big Data to Big Marketing: Seven Essentials

»» insights 3. Decide

4. Answer

3. Scientifically balance all the factors in complex marketing decisions Companies can get a head start with Big Marketing by using predictive models with tried-andtrue business rule libraries for common marketing decisions, such as up-selling and cross-selling strategies. Yet, with the opportunity to take an expanding range of data and analytic predictions into account, decision processes can grow quite complex. Companies that rely on simple business rules alone may soon find it difficult to manage or even fully comprehend how they are making decisions. Use of data-driven techniques for modeling and optimizing decision processes is essential for effectively bringing Big Marketing intelligence into operations.

Decide

5. Engageidentify the Scientifically best actions from trillions of possibilities

Action-effect models are one of these key techniques. They predict how likely the customer is to respond to a particular action within a particular time period (e.g., is this customer likely to purchase if offered a 20% discount on lawn mowers valid for 30 days?) and what the effect 6. Learn would be on KPIs (e.g., response rate, sales, interest income).

The most sophisticated Big Marketing projects generally encompass dozens, even thousands, of models employed simultaneously. Working with a leading retailer, for example, FICO is using automated techniques to generate propensity models at the product sub-category level for every offer available in the retailer’s loyalty program. These propensities, along with investment goals and eligibility criteria for each offer, are input into a mathematical optimization to match offers to consumers in a manner that How big is Big Marketing? maximizes overall response. For one retail client, FICO is analyzing 10 million transactions/day

using thousands of predictive models. There’s plenty of room for these numbers to grow, as FICO’s use of Big Data analytics in other industries shows. In credit card fraud detection, for instance, FICO analyzes some 7. Experiment 24 billion payment transactions/month, performing tens of thousands of analytic calculations in a fraction of a second to spot suspicious behavior. The National Football League uses FICO analytics to develop game schedules that maximize television ratings while accommodating the needs of teams and fans. To find the best schedule, trillions of scheduling permutations, including some 20,000 variables and 50,000 constraints, are considered.

Optimization scientifically balances what the company is trying to accomplish with real-world constraints under which the business and its partners operate. Let’s say the objective of a campaign is to achieve the highest incremental sales that have a positive margin. Additional constraints might require the campaign to target a maximum of 2 million customers and stay within the combined budgets of Supplier A, which has a promotional budget of $5,000, and Supplier B, which has a budget of $10,000. Only customers who have not purchased the item in the past 120 days will be targeted, and customers whose local store does not carry the promotional items will be excluded.

With easy-to-use simulation tools, marketers can modify one or more constraints to see how the optimal decision strategy shifts, exploring trade-offs and choosing the best operating point for their business. In the case of the retailer loyalty program described above, the result of optimization and simulation is a frequently refreshed selection—from literally trillions of possibilities—of a unique set of relevant offers for each customer in the program. This has driven response rates as high as 30%. FICO is using a similar approach to help a US super-regional bank grow its credit card portfolio and increase penetration of its deposit household base. By leveraging a wealth of data with Big Marketing analytic techniques, the bank succeeded in getting the right products to the right customers—those most likely to respond and use the credit product in a manner that generates profits. Results in the first 12 months include a 15% increase in cross-sell penetration and significantly higher usage, driving average balances up by 8%.  

www.fico.com

page 6

From Big Data to Big Marketing: Seven Essentials

»» insights 4. Answer

4. Answer “million dollar” marketing questions

1. Understand

Decision modeling, optimization and simulation can also be used to answer difficult questions, such as “With a trade partner that wants to push out a million discounted offers for milk, what percentage of that would be most profitable for us to promote to customers?” 2. Predict The larger and more varied the data set being analyzed, the more complex the range of questions that can be answered in a statistically reliable manner. These answers can be extremely valuable. In Gartner’s CEO Survey 2012, 100% of respondents could identify a piece of information that, if they had it, would allow them to run their businesses better.7 “Customer intelligence” was the most sought-after category of information, above information on competitors and even on sales.

Answer 6. Learn

Explore value-creating questions that balance many complex factors

3. Decide

Using Big Marketing techniques, companies can find answers to questions they believe will give them an edge. For example, FICO is helping one retailer explore trade-offs such as: “Will 10% off be more effective than free shipping, or than simply telling customers that the desired item is in stock and will be reserved for a short period of time?” and “Is offering 12 months of interest-free credit necessary, or will 6 months be nearly as enticing?” We’ve helped an Asia-Pacific lender increase usage of an installment loan product attached to a 4. Answer credit card by figuring out “Which customers are likely to increase their loan utilization if offered a 30% temporary discount, and which customers would be sufficiently incented with just a 20% or even 10% offer?” The technique is helping the lender better understand the subtle relationships between line increase, utilization and delinquency in a market where consumers have traditionally been reluctant to take on debt. Moreover, Big Marketing analytic techniques can help retail banks answer questions that connect operations-level customer decisions with executive-level direction. Instead of limiting themselves to isolated account-level questions like “Should I offer this customer a line of credit, and if so, what should the limit be?”, they can answer questions that capture the real-world complexity of retail banking decisions: “In light of our portfolio goals and this customer’s current and future value to the bank, risk profile and expected reaction, what is the most profitable offer?”

5. Engage

6. Learn

5. Engage customers in individualized multichannel relationships Once a decision strategy is chosen, it must be output in an immediately executable form. Generally this is in the form of a “decision tree” of business rules. But unlike rules that have been authored by marketers based on empirical evidence and judgment, these rules are packed with data-driven customer intelligence. As a result, when they trigger a response to a customer behavior, that response has been shaped by deep customer insight. Decision trees can be imported directly into rules-driven operational systems. Or they can be implemented into a business rules management system (BRMS), which simultaneously performs decisioning as a service for any number of operational systems—including real-time dynamic generation of optimized treatments during interactions with customers.

7. Experiment Engage Drive multichannel customer interactions from Big Data insights

Two FICO clients illustrate how this might play out in the end game of delivering offers and orchestrating interactions with customers. One is a CPG company that drives interactions across all

7

“CEO Survey 2012: The Year of Living Hesitantly,” Gartner, Inc., March 2012

www.fico.com

page 7

From Big Data to Big Marketing: Seven Essentials

»» insights of its many brand websites from a consistent set of promotion and interaction rules. Another is a large retailer whose loyalty program members can access their individualized offers from a home computer, via a mobile app or at an in-store kiosk. Figure 2 depicts an example of a cross-channel Big Marketing campaign.

2. Predict

Figure 2: Cross-channel Big Marketing in action @ Email 1: Renovation tips

Opts in

Telemarketing Call Planning: Bathroom remodel

4. Answer

Budget: $12,000–$20,000

Home improvement financing with platinum card

@

@

Email 2: How to hire a contractor (ask for profile info)

Email 3: Design tips (enticement for in-store consultation

Visits website Registers Gets appliance coupon

Candidate for: Store credit card upgrade

Books appointment online

Text reminder: with special platinum card rate—”ask at consultation”

Consultation Discuss remodeling needs, credit card

Cross-Channel Data, Campaign Management and Analytics

6. Learn

6. Learn from what is happening in the market now Because Big Marketing decisioning is based not only on historical data, but also on data being collected frequently—in some cases, via constant streams from many sources—it’s responsive to change. Learning from and adapting to changing behaviors and market conditions happen at both the back end and front end of decision processes. At the back end, the broad range of data being analyzed and the emphasis on transactions means that predictive data elements are frequently refreshed. The retailers discussed in this paper are generally collecting daily customer-level response data from transactions and generating new propensity scores for customers on a frequent basis. In addition, automation is used to update predictive models every 90 days to factor in changes in aggregate consumer behavior.

Learn Adapt faster to customer responses and market change

At the front end, companies use analytic learning loops to rapidly measure how customers are responding and to adjust decision strategies where necessary while campaigns are still underway. One retailer, for instance, has used this process to refine a lapsed-product offer strategy that is now driving unprecedented response rates.

www.fico.com

page 8

5. Engage

6. Learn

From Big Data to Big Marketing: Seven Essentials

»» insights Learning is swift since there’s no need to wait for all of the data to come in to determine which strategies are working and which aren’t. By comparing early actual results on KPI metrics against simulated results, marketers can extrapolate longer-term outcomes. With the ability to modify business rules without need for IT assistance, they can change decision strategies in minutes.

7. Experiment

7. Launch forward-looking experiments to get ahead of competitors Variances between simulated and actual results may be opportunities not only for correction, but for experimentation. Using analytic learning loops in conjunction with systematic experimental design, companies may discover opportunities that are not evident to competitors and gain forwardlooking insights into how customer behavior is evolving.

Experiment Create competive advantage by pushing the edges of business-as-usual

The quantity and variety of data used in Big Marketing increases the range of experiments companies can conduct in a short amount of time while still producing statistically significant results. In addition to testing variations on well-performing decision strategies, companies should test strategies that are beyond the edges of business-as-usual and organizational comfort zones. This edge-probing deliberately introduces controlled variation into the production data, thereby expanding what can be learned from it. Moreover, bold front-end experiments like these will increasingly be supported by new approaches to back-end exploratory analysis. Emerging methods in the era of Big Data involve analyzing larger amounts of data, initially, in a rough or “good enough” way, then iteratively working toward narrower, more precise results. Such methods can supply unexpected predictive attributes to be used in unconventional decision strategies for edge-of-envelope experiments in the production environment. As the consulting company Ovum states, “Big Data is a change of mindset regarding the art of the possible.”8

“How will corporate marketing functions and activities need to evolve if large-scale experimentation is possible and how are business processes likely to change? ... Using data to analyze variability in performance—that which either occurs naturally or is generated by controlled experiments—and to understand its root causes can enable leaders to manage performance to higher levels.” —Big data: The next frontier for innovation, competition, and productivity McKinsey Global Institute, May 2011

8

“What is Big Data? The Reality for Analytics,” Ovum, June 2011

www.fico.com

page 9

From Big Data to Big Marketing: Seven Essentials

»» insights Get a head start on Big Marketing Marketers across industries can get a head start on Big Marketing through FICO’s configurable prebuilt applications, available as cloud-based services or for on-premise installation. Both products can support Hadoop Map-Reduce, as well as more traditional SQL analytics. FICO® Analytic Offer Manager identifies the right offer for the right customer at the right time. It employs a number of Big Data techniques for analyzing a complex mix of client data, both structured and unstructured, from multiple sources to form a cohesive view of the customer. It also leverages Big Data functional and scale-out concepts by partitioning and pipelining event data to deliver efficient offline and real-time predictions. The product includes FICO’s industry-leading predictive analytics, such as propensity scores and uplift scores, as well as worldclass optimization tools for assigning offers to individuals while meeting business goals. FICO® Customer Dialogue Manager enables marketers to design, execute and manage precisely timed campaigns that engage customers across all channels, including social media. (New channels can also be brought into the mix as they emerge.) The product leverages Big Data technologies, including columnar databases and advanced shared-nothing replicated storage, to provide the high-velocity insights that enable relevant conversations with customers. Marketers can apply these insights to generate individualized, interactive dialogues with customers in a coordinated fashion through on-demand or automatically recurring campaigns, as well as event-triggered campaigns based on hundreds of types of events.

»» Conclusion

With the advent of Big Marketing, we’re headed “back to the future”—a future that will finally make operations customer-centric. While marketers have been working toward this for years, today’s Big Data techniques will help them rise above the obstacles and realize the immense benefits. Learn more: •  Visit the FICO Labs Blog for the latest ideas and developments in the era of Big Data. •  D  ownload Insights white papers, which regularly cover analytic innovations and best practices in marketing.

The Insights white paper series provides briefings on research findings and product development directions from FICO. To subscribe, go to www.fico.com/insights. For more information

North America toll-free +1 888 342 6336

International +44 (0) 207 940 8718

email [email protected]

web www.fico.com

FICO and “Make every decision count” are trademarks or registered trademarks of Fair Isaac Corporation in the United States and in other countries. Other product and company names herein may be trademarks of their respective owners. © 2012 Fair Isaac Corporation. All rights reserved. 2922WP 11/12 PDF