REVOLUTION CASE STUDY [x+1] Completes Next-Generation POE; Its Origin Enterprise Data Management Platform for Automated, Big Data-Driven Marketing Optimization Revolution R Enterprise Tapped for High-Performance, Scalable Predictive Optimization Engine Company: Industry: Challenge: Solution: Results: [x+1], New York, NY www.xplusone.com Software and services for optimized digital marketing through multi-channel visitor experiences on personalized websites and real-time digital audience targeting [x+1] s need for real-time analytics, automated model updates, ability to include new data types and manage quickly-growing data volumes (without sacrificing performance) were not well-matched for existing closed platform analytics application Re-engineer entire analytics application with Revolution R Enterprise, leveraging RevoScaleR for Big Data Analytics, and a distributed computing platform for data management Company has achieved performance and scalability required for growth Background Every marketer wants to know which prospects to attract, where and how to find them and how to make their experiences as relevant as possible. New York based [x+1] s Origin Enterprise Data Management Platform (DMP) provides powerful Big Data analytics for its category leading clients such as JP Morgan Chase, Verizon and Intuit. [x+1] helps them accelerate customer acquisition, manage the customer purchase funnel, cross-sell / upsell and optimize audience targeting across touch points, including online media buying. 1
The [x+1] Origin Enterprise DMP uses a centralized decision engine called the Predictive Optimization Engine (POE ), which utilizes Revolution R Enterprise, and leverages both proprietary and third party data managed by its Big Data Framework. POE targets the best audiences across channels and individually tailors messages by site and visitor in real time so that marketing and media strategies are achieved. Advanced statistical models generated by POE direct and decision the three major touch-point components of the Origin DMP: the out-bound Origin Media DSP (demand side platform), the in-bound Origin Site personalization tool and the Origin DMP. The centralized decision engine determines in real-time for [x+1] s clients which audience segments to target and which messages or offers to show prospects and customers across display media, websites, and, via the Origin DMP, email, SMS, click to chat systems, and call center systems. For outbound media channels, POE determines the optimal price to pay for each user impression for each client offer across billions of bid requests per day, taking into account the client s channel scale goals and budget constraints. In concert is the creative decisioning capability in media that determines the optimal content asset to render. For the customer s own website visitors, the engine determines which offers, messages, experiences or advertisements, and other content to show to an individual visitor for a given site visit. POE continuously builds models to allow [x+1] s clients to automatically create a personalized experience for each site visitor, increase conversion rates and optimize marketing spend. This same optimization and intelligence is leveraged across other touch points using the Origin DMP, providing the first truly multi-channel optimization. How the [x+1] Predictive Optimization Engine Personalizes Experiences [x+1] s Origin Site tool automates the movement of customers and prospects through a process prescribed by marketing by using business logic based on advanced statistical modeling. Imagine that on a recent visit to your bank s website, you searched for mortgage rates. Upon your next visit, POE customizes your home page to include a banner that informs you about the company s new mortgage product. At your third visit, the bank s website homepage features a photo of a family standing in front of a lovely home with a For Sale sign that has been covered by a Sold sign. A prominent section of the site informs you that the bank is offering $500 towards closing costs for new mortgage applicants. You click the link, start the process, and POE site engine has done its job. 2
Challenge: Big Data Analytics In Real-Time! Marketing optimization is unquestionably a Big Data Analytics challenge, involving terabytes of data derived from internal sources (e.g. CRM systems, call center transcripts and company website activity) as well as third-party sources such as sociodemographic segmentation, interest/affinity information, behavioral data, search engines, etc. In [x+1] s case, the Origin DMP manages terabytes of data, with terabytes being added monthly. Optimizing the science of marketing in this Big Data age requires modern analytics that are turbo-charged by not choked by big data, meaning that they deliver very fast performance even at very large scale. Why do marketers need speed? The most successful marketers are generating unique website experiences and user interactions in real-time. They are constantly updating their online media bids across scores of properties based on ever-changing price and availability data. As in financial trading models, big data analytics must be done fast to seize opportunities that are here today and gone in just a few seconds. 3
We need a highperformance analytics infrastructure because marketing optimization is a lot like a financial trading. By watching the market constantly for data or market condition updates, we can now identify opportunities for our clients that would otherwise be lost Leon Zemel Chief Analytics Officer, [x+1] Why complicate marketing analytics with the challenges of scaling to big data? Simply, the results generated by Big Data analytics leapfrog traditional analytics for two primary reasons: 1. First, analysts can increase the number of attributes, or variables, considered in the model (for example use in-market segments from an exelate, BlueKai, Datalogix, or other 3rd party data providers). Moreover, new data types may contain never-before-considered attributes that when included in the model, can offer novel insight. 2. Second, expanding the amount of data against which a model is executed (for example from a three-week span to a three month span) allows marketers to uncover trends or outliers that might not have been apparent in a shorter window. Boosting analytics performance to match the opportunity presented by Big Data is often the catalyst to develop second-generation, big-data-enabled analytics applications. [x+1] s Chief Analytics Officer Leon Zemel opted to develop a nextgeneration POE in 2011 With more than a dozen third party data sources and scores of new data attributes becoming available, Zemel and the analytics team wanted to ensure that POE could scalably harness all available data sources, while extending the models time horizon and improving its self-directed functions. In other words, the team set out to build the next-generation of the real-time digital marketing environment. Further, [x+1] felt that the openness and scalability of the Revolution Analytics platform would remove barriers to innovation and experimentation, while providing superior flexibility to integrate with a growing set of digital execution channels (e.g., mobile) and partners (e.g., Facebook). In addition, Revolution R s economic model is perfectly suited to growing, innovative companies like [x+1]. [x+1] Migrates to Revolution R Enterprise Revolution Analytics consultants worked alongside [x+1] s team to adapt Revolution R s leading technology to X+1 s patented POE. [x+1] customers benefited from [x+1] s ability to incorporate its latest thinking about marketing optimization gleaned from years of successful customer deployments. POE is a big-data-enabled, leading edge marketing optimization engine. Using Revolution Analytics scalable, parallelized RevoScaleR algorithms, the speed of the underlying analytics engine has been turbocharged, allowing for greater efficiency, scalability, and time-to-market with actionable analytics. Zemel explains, We need a high-performance analytics infrastructure because marketing optimization is a lot like a financial trading. By watching the market constantly for data or market condition updates, we can now identify opportunities for our clients that would otherwise be lost. 4
[x+1] achieves critical scalability through its Big Data infrastructure, which manages terabytes of data comprised of thousands of client characteristics, third-party data and information about media availability, for example. The Revolution Analytics-powered POE creates a customized model for each of [x+1] s clients. Each client model analyzes their specific data in a way that has been accelerated through RevoScaleR, a key component of Revolution Analytics high-performance big data analytics capabilities. The resulting decision logic (e.g., what creative to show what audience, what bid price to place in the real-time media market) is then pushed into the production system to be executed in real time. A view is also provided to the marketer/agency so that decision rules are transparent. We wanted to move to a more open platform that would allow us to continue to innovate without sacrificing performance or scalability. Revolution Analytics met that requirement and gave us additional opportunities to offer new capabilities to our clients, explained Zemel. Results: High-Performance, Scalable Big Data Analytics Platform to Support [x+1] s Growth With next-generation POE, [x+1] can now: 1. Universally apply the POE engine more widely and more rapidly as volume and client demands rapidly increase; 2. Innovate due to the openness of the R-based platform; 3. Incorporate even more market and customer information in the POE model by handling bigger datasets with better, more granular variables (hundreds, if needed) and; 4. Offer new product capabilities to its clients faster than ever before. Looking forward, [x+1] also has the opportunity to offer new types of services to their clients. With Revolution Analytics RevoDeployR server, [x+1] could make each client s POE available to the client s own analysts through a web services interface, which will allow them provide even more transparency to their clients. X+1 s ability to store massive amounts of data and serve it to Revolution Analytics allows [x+1] to fine-tune and accelerate POE model development and deployment and automates website updates to convert visitors and optimize media spend for its clients. 5
About [x+1] Known as the [x+1] Origin Digital Marketing Hub, this unique combination of capabilities allows marketers to synchronize their messages to consumers across a broad range of channels: websites, display, landing pages, email, SMS, direct mail, chat, and call centers in real time, thus maximizing campaign performance and ROI in a consistent, repeatable, and measurable way. Top companies in financial services, automotive, retail, CPG, telecommunications, online services and travel have significantly increased the performance of their digital marketing investment by using [x+1]'s solutions and services. The company is headquartered in New York City, with offices in Connecticut, and Illinois. About Revolution Analytics Revolution Analytics is the leading commercial provider of software and services based on the open source R project for statistical computing. The company brings high performance, productivity and enterprise readiness to R, the most powerful statistics language in the world. The company s flagship Revolution R Enterprise product is designed to meet the production needs of large organizations in industries such as finance, life sciences, retail, manufacturing and media. Used by over two million analysts in academia and at cutting-edge companies such as Google, Bank of America and Acxiom, R has emerged as the standard of innovation in statistical analysis. Revolution Analytics is committed to fostering the continued growth of the R community through sponsorship of the Inside-R.org community site, funding worldwide R user groups and offering free licenses of Revolution R Enterprise to everyone in academia. Please visit us at www.revolutionanalytics.com 6