SERVICE INSIGHT EDUCATION SERVICES FIELD SERVICES MANAGED SERVICES PROFESSIONAL SERVICES SERVICE REVENUE GENERATION SUPPORT SERVICES TSIA-SI-14-006 February 4, 2014 TSIA Consumption Analytics Framework by Jeremy DalleTezze 1. INTRODUCTION How often is the “A-word” spoken in your organization today? Analytics has been a buzzword on the minds and lips of business leaders for the past decade. If searches on Google reveal anything about business psyche, then the trend lines in Figure 1 indicate that analytics overtook outsourcing in 2007, and then intelligence in April of 2008 in the marketplace of business ideas. If we look to where this 1 interest began, we see in Figure 2 that developing countries such as India are just as likely to be trendsetters as the west in an interconnected global economy. Figure 1: Business Buzzword Web Searches All websearch data is provided by Google. These graphs can be explored here: google trends dashboard. 1 These figures imply that on a relative basis, a bigger percentage of Google queries in India were made for analytics than in any other country in 2006. India is home to many technology firms, and many innovative individuals gifted and trained in the sciences; this may be the cause of their leadership. 2 Figure 2: "Analytics" Web Searches Global Trends Jan – Jun 2006 Jan – Jun 2008 Jul – Dec 2013 Dark blue indicates relatively higher search volume. We all know that analytics is a topic of interest for business leaders. Now, let us define it, and then use the remainder of the article to explain why it is important to technology service providers, and how you can cultivate this capability to achieve a competitive advantage. 2. DEFINING CONSUMPTION ANALYTICS Put simply, business analytics is the organizational process of improving outcomes by leveraging data. This process involves people, technology, math, statistics, data, and a lot of work. Given this generic framework, experts then provide more narrow definitions for their particular fields, as seen in the long 2 list of common analytics prefixes in Figure 3. These specific terms and more precise definitions typically cover the nature of the data, the type of analysis, and the overarching goal of the organization. Figure 3: Common Analytics Prefixes 2 Created here: http://www.wordle.net/. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 3 Consumption analytics was introduced in the book B4B: How Technology and Big Data Are 3 Reinventing the Customer-Supplier Relationship. The authors argued that analytics was a required capability for companies to move from a product-focused service provider to an outcome-focused service provider. In this context, and as illustrated in Figure 4, a success science team, with clear corporate objectives, could utilize analytic techniques to leverage their growing data inventories to help their customers achieve specific outcomes. The overall goal of consumption analytics is to improve your customer’s consumption of technology products and services by describing and inferring insights from a large variety of data streams. Figure 4: Success Science and Consumption Analytics Figure 5 provides specific objectives in this paradigm. The first objective, Service Efficiency, is being pursued by many of TSIA’s members. As an example, a PS member routinely assesses the distribution of utilization across its employees in an effort to make their professional service projects more predictable as they balance profitability and customer satisfaction. In another example, a support services member has identified leading indicators of CSAT and call volume to enable a proactive approach to making customer support more efficient. Customer Adoption objectives aim to understand how customers, and their employees, are adopting your products and services in an effort to facilitate healthy and productive consumption paths. While this objective does require that your organization has access to usage information, it does not require a XaaS platform. Specific examples from the membership range from benchmarking customers across adoption usage to suggesting best practices, training, and product features in a data-driven format. 3 Wood, J.B., Todd Hewlin, and Thomas Lah. 2013. B4B: How Technology and Big Data Are Reinventing the Customer-Supplier Relationship. San Deigo: Point B, Inc. Available for purchase at Amazon. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 4 Lastly, companies that pursue Customer Business Outcome objectives are able to develop datavalidated success plans for specific business goals of their customers. A few members have developed offerings where they first assess the customer’s complete environment and approach to achieving common business results, and then create a detailed plan of adopting products and services to achieve the most profitable outcomes. As they progress through their plan, tracking is done to monitor success, and in some cases trigger payments for services. An even smaller set of companies is in the process of automating large chunks of this activity. Figure 5: Consumption Analytic Objectives 3. TACTICAL FRAMEWORK FOR ANALYTIC PROJECTS AND PROCESSES With the overall success science methodology in place, and leaders agreeing on the right combination of objectives to pursue, a tactical approach is needed to complete analytic projects and develop consumption analytic capabilities. To that end, we have created the five-step framework shown in Figure 6. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 5 Figure 6: Tactical Analytics Framework a. Question Inventory • Compile and prioritize questions to answer that drive the largest ROI, plan for implementation. b. Data Inventory • Gather and connect required data streams to address highest priority questions. c. Descriptive Analytics • Answer “what and when” questions related to the prioritized objectives. d. Predictive Analytics • Answer “how and why” questions related to the prioritized objectives. e. Decision Making • Integrate analytics into both your and your customer’s business processes via benchmarking, forecasts, experiments, and simulation. a. Question Inventory With the proliferation of data, we could utilize analytics for every business outcome and question what we have, but we need to start small and aim for the largest, lowest hanging fruit. In order to identify these targets, and to make sure that the organization can effectively roll out new data-informed processes in the last step, one must assimilate a cross-functional team of leaders to compile this inventory and set the appropriate strategy and plans. At the minimum, you need executive leadership, customer representatives, IT managers, and senior analysts in the room. b. Data Inventory During the question inventory process, your IT and analytics’ employees will have a long list of metrics related to the high-priority outcomes and processes. This list will require the breaking down of data silos in your organization. Several of our members are in the process of a multi-year enterprise data integration process, but this does not prohibit them from connecting the data points now to uncover insights. There is a plethora of affordable connectors and analytic SaaS platforms that can connect 4 these in a matter of hours or days. Providing your IT and analytics employees with professional 4 We are not suggesting that the multiyear full-data integration can be replaced by a few days of work; rather, analytic platforms provided by companies such as MicroStrategy, Tableau, and Ducksboard can be leveraged to 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 6 development time to learn about the latest big data innovations may be warranted, as well as seeking a specialized vendor. If your data contains millions upon millions of rows, sampling can overcome any processing speed issues, or you could leverage computing-as-a-service platforms such as Amazon’s Elastic Compute Cloud (EC2). Tech issues related to “big data” are no longer valid excuses for delaying your analytics progress, although relationships with your customers may prove a significant challenge for Customer Business Outcome objectives (see Section 5: Consumption Analytics Challenges and Emerging Practices for more on this issue). c. Descriptive Analytics Most BI tools provide the core descriptive analysis of an organization’s selected KPIs, such as the normalization of variables, reporting of statistics, visualizations, and alerts. The goal is to create that single version of truth that employees can access at any time to quickly get a pulse of the related business outcome’s metrics. More important than the technical aspects of these dashboards are the 5 data quality and employee understanding of the presented information. Do your employees really know what each metric represents? Do they understand how the reported statistics may or may not be influenced by outliers? An investment into the soft skills around utilizing dashboards may have a larger return than the initial tech spend. d. Predictive Analytics After the first three steps, even if you pick a very specific business outcome to explore, the number of candidate metrics in your dashboard may be incredibly high. Now, we can leverage the tools of predictive analytics to cut through the clutter to identify the most valuable information to track, study, and utilize. Figure 7 shows a simple process to follow when conducting analytic projects and building data-driven tools and processes. The success plays are the organizational actions that you can take in an attempt to influence a certain business outcome. For example, your company can provide weekly emails with linked training videos in an attempt to increase end-user capabilities with your software. They could also hold live webinars, send instructional guides, email short case studies, and perform many other actions as well (hence, the many rectangles and squares in the diagram). Here is where predictive analytics can help: it can identify which KPIs impact your outcome (the two bright blue circles), and which success plays impact that smaller set of KPIs (the three bright blue rectangles). immediately connect your data, conduct basic analysis, and share interactive dashboards with colleagues as you wait for your enterprise data solution. If your analysts are proficient programmers in languages such as R, Matlab, or Python, they can perform in-depth analysis on such connected data sets using both open-source and paid connectors or drivers. These algorithms can also be connected to your dashboards to increase the insights, for example, R is now connected to Tableau and MicroStrategy, and MATLAB, SAS, S+, and R are connected to TIBCO Spotfire. 5 Data quality is seen as the biggest barrier to successful analytics and BI initiatives in both 2013 and 2014, according to a survey cited in the “State of Analytics” in Information Week, November 25, 2013. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 7 This investigative and validating process should become part of a continual process improvement in all of your analytic activities. Figure 7: Data Validated Success Plays Success Plays KPIs Outcomes Levers to Influence Performance Performance Indicators Related to Outcomes Outcomes of Your Business and/or Customers e. Decision Making Having leveraged the data to (1) reveal a set of KPIs that influence your outcome of interest, and (2) identify success plays to improve those KPIs, now you must leverage these insights by incorporating them into your decision making throughout the entire organization. Each level of employee must have 6 access to the same version of truth, although it will likely be summarized differently for each level. As employees adjust behavior in this analytics roll out, your organization must track the KPIs and 7 outcomes to assess the impact. As a part of this assessment, steps “a” through “d” must be repeated, thus it is imperative that the technical steps in “b” through “d” are done in a scalable way so that future labor requirements are minimal. 4. TSIA CASE STUDY: HOW TO PROACTIVELY IMPACT ACCOUNT RENEWAL In the fall of 2013, TSIA assembled a consumption analytics taskforce. Our meetings brought together leaders from Marketing, Research, IT, and Member Success. As itemized in Figure 8, our crossfunctional group followed the five tactical steps covered earlier in this report. We selected account renewal for our first project because of its importance to our business, but also due to practical issues: 6 See the report by Paul Martin (Brocade), entitled “The Metrics-Driven Organization,” for a great example of how an organization incorporates analytics into decision making at all levels. 7 This is another example of the PIMO construct: plan, implement, monitor, optimize. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 8 (1) our MSR team, who would be most affected by this project, already had data-informed processes in place, and (2) we had the necessary data to effectively model renewal. Figure 8: Analytic Steps to Improve Renewal a. Question Inventory • Focused on a "customer adoption" objective: account renewal. Created prioritized list of other questions for future work. b. Data Inventory • Connected data from different sources. Left harder-toreach sources on prioritized list for future work. c. Descriptive Analytics • Reported basic statsitics on over 90 variables, including simple correlations to renewal. d. Predictive Analytics • Identified 12 nuanced KPIs that influence renewal, and 8 success plays that affect the top 4 KPIs. e. Decision Making • Renewal forecasts are included in Member Success reports, KPI dashboard is provided to members as a free benchmarking service, success plays are tracked. Like most projects of this type, the data inventory process provided the most difficult technical challenges, though extra time was committed to ensure that future iterations would be less manual. Instead of loading all of the hypothesized KPIs into a dashboard for the organization to view, we built 8 several models to cut through the clutter, effectively tying specific KPIs to renewal for different types of accounts. As of the date of this report’s publication, we have incorporated the insights and summarized data streams into our member success process. As an added benefit, we are leveraging the dashboards to provide free benchmarking services to our members as part of periodic member engagements (success plays). 8 We used a mixture of clustering algorithms to explore different types of accounts, but ended up using a segmentation scheme that was based on a single attribute already emphasized by sales and member success employees. Within each segment, we built logistic and decision tree models to identify the success plays and KPIs. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 9 5. CONSUMPTION ANALYTIC CHALLENGES AND EMERGING PRACTICES In future reports, presentations, and case studies, we will provide more details and best practices 9 regarding the steps covered in this framework article. Based on member feedback thus far, the most challenging issues are getting access to customer data and effectively monetizing analytic outputs around customer adoption and customer business outcome objectives. Below are some emerging practices and early thoughts around these issues. Getting Access to Customer Data10 Customer Adoption objectives require customer usage data from your products and services. As indicated earlier, while this may be a technical challenge for on-premise software providers, it is not insurmountable. The more delicate issue will be convincing your customers that they need to give you access to data so that you can provide them with better products and services. This can be a Catch-22 situation, as your initial value proposition will have no evidence, but you can easily refer to other cases. During the conclusion of his keynote address at TSW Service Transformations 2013, Andreas Weigend, former chief data scientist from Amazon.com, provided the following quote on the emerging mindset of consumers everywhere: “Give data to get data.” All of your customers have this mindset in other parts of their lives and business; you just need to show them your plan for how your products and services will improve over time with this data, just as other products in their lives have improved with their data. Customer Business Outcome objectives require more customer data from activities not directly connected to your products and services. As an example, consider Bret Barczak (GE) and Tim Kottak’s (GE Healthcare) presentation, “The Application of Data and Analytics to Solidify Your Company’s Value Propositions.” GE Healthcare is in the business of improving several of their customers’ business outcomes, including operational efficiency. Given their penetration into hospital hardware needs, much of the required data is captured as usage data; other outcomes require more data and information, such as how the hospital arranges all of its hardware, employees, and patients in the hospital. To convince hospitals of why they should share this information with GE, their value proposition contains (1) glaring statistics such as “nurses spend 21 minutes per shift searching for lost equipment,” and (2) the obvious B4B fact: since GE Healthcare has a huge customer base that is trying to achieve the same goals, GE Healthcare is in the best position to help them achieve those goals! 9 The author presented the basics of this framework at TSW Service Transformations 2013 and has been working with a small advisory group of TSIA members on the cutting edge of analytics in services. 10 For a complementary and more detailed discussion of this topic, see chapter 9, “Pivot 3: The Data Handshake,” of B4B: How Technology and Big Data Are Reinventing the Customer-Supplier Relationship, J.B. Wood, Todd Hewlin, and Thomas Lah. 2013. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 10 Monetizing Analytics Nearly all of our members have been using analytics to achieve Service Efficiency objectives. Monetization was not a major issue in these instances, as the efficiency gains decreased costs, thus improving the bottom line. Customer adoption analytics can have a similar indirect effect on the bottom 11 line if your organization has effectively aligned itself for “Land + Expand Selling” : your analytics can leverage customer usage data to make upselling and cross-selling more efficient. But creating the new suggested adoption services in B4B (i.e., adoption planning, consumption monitoring, and consumption optimization) will not be trivial, and the initial variable costs of providing the services will be rather high. Outcome service offers, and the associated analytics, will require an ever-bigger, upfront risky investment from the supplier. In a decreasing revenue environment where customers are demanding the shift from CapEx to OpEx contracts, investments into non-revenue-generating activities are tough to justify. This is where leadership needs to provide the long-term success plan that may cause public investors to cringe. The bottom line may worsen in the short run, as immediate monetization will likely have a small price tag, if at all. Several of our members making this shift have adopted tactics to minimize this impact, as shown in Table 1. As their Customer Adoption and Customer Business Outcome analytics mature and improve, along with their associated premium services and new selling capabilities, their bottom lines will improve greatly. Table 1: Observations from Companies Making the Shift 12 Tactic Explanation Start with Your Best Customer(s) Gaining access to new customer data and creating monetizable services go hand and hand. If you start with your most successful customers, you can create the best case studies validating your value propositions when you begin to scale these activities across your customer base. Tiered Pricing for Benchmarking As soon as you study your customers’ usage data, creating a benchmarking capability is straightforward. Basic benchmarking can be given to your customer base almost immediately. Once they begin to realize the value, monetizing more sophisticated benchmarking and adoption services will be an easier discussion. 11 See chapters 7 and 8 of B4B: How Technology and Big Data Are Reinventing the Customer-Supplier Relationship, by J.B. Wood, Todd Hewlin, and Thomas Lah. 2013. 12 These tactics were derived from several conversations from a group of TSIA members over the fall of 2013. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 11 12 Tactic Explanation Customer Business Outcome Objectives in Managed Services If your organization is growing a managed services division, this is the best place to grow your analytic capabilities for your customer business outcome objectives. The charter of managed services, and the nature of their business and conversations with customers, optimizes the odds of success for gaining access to data and leveraging it for success. 6. BECOMING A DATA-DRIVEN TECH SUPPLIER The strategic changes and investments required by the calls to action in B4B are considerable, overwhelming even, for the majority of our members. As this report on consumption analytics is being written, the topic is still a green-field throughout enterprise IT service providers. But, do not mistake this report as a theoretical abstraction of things that may never come to fruition—this is a framework to help you catch up to where certain IT tech service providers already are, where many will be soon, and where the vast number of data-driven consumer companies have already been. Table 2 provides a few short examples of such companies, while TSIA publications and presentations scheduled for the remainder of 2014 will provide more thorough case studies. Table 3 provides a short list of vendors that can provide you with certain consumption analytic capabilities. These lists are not comprehensive; rather, our main intent is to provoke your thoughts about how companies are leveraging analytics. Table 2: Companies with Maturing Consumption Analytics Capabilities Industry Company Description Finance SigFig SaaS solution that connects an individual’s investments into one platform. Updates on the entire portfolio, and how they align with the user’s ultimate outcomes regarding growth, risk, etc. are dynamically and continually updated. This quote from the CEO articulates this datadriven tech supplier’s position to other investment models: “They don't scale well, so a person who is trying to manage money for a couple hundred people, they can't look at everyone's account all the time...if you could replace that human with a machine—which has been done in a lot of other industries to great success—you really can build a better, 13 more scalable, lower-cost solution.” 13 From “Can Robots Manage Your Money Better than You? Startups Say Yes,” Steve Henn, Dec 30, 2013, NPR blog. 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 12 Industry Company Description Customer Service Zendesk With over 35,000 customers, 24 million tickets, and 200 million users, Zendesk leverages its unified database of insights to provide better customer support, build better products, and educate customers. Its dashboards assess churn risk (low/med/high) and their benchmarking services enable customers (and Zendesk) to compare their customer support outcomes (CSAT, efficiency, scales) by industry, tenure, 14 feature, use, and much more. Human Capital Management Success Factors Human management software that aligns people data (Success Plays) and business data (KPIs and Outcomes). They have developed a system that ties disparate data together to connect it to specific 15 outcomes for a range of customer types and employee levels. Table 3: Consumption Analytics Vendors Company Description Gainsight SaaS solution that bundles a business customer’s data, both from within and outside of salesforce, to proactively monitor consumption data. Their analytics offer leading indicators of both churn and cross-/upselling opportunities. Scout Analytics “Scout is a recurring revenue management solution designed to maximize customer value and accelerate growth in revenue and profits. Scout is the first solution that transforms usage data into predictive analytics and automations that help you engage your customers proactively and at scale. By providing ongoing insight into customer usage data, Scout helps companies ensure customer success, increase and automate add-on and upgrade sales, optimize rate plans, and maximize trial 16 conversions.” SATMAP Their software optimizes support by pairing agents with callers based on personality. Data is continually collected on callers and agents, and their predictive models and algorithms are consistently updated based on outcomes. ® 14 Read more in Sam Boonin’s presentation, “How Zendesk Uses Data to Make Customers Happier.” Read more in Omid Razavi’s presentation, “Cloud Services That Scale and Delight Customers.” 16 This quote was found under “The Solution,” on 1/23/2014, from http://scoutanalytics.com/overview.php. 15 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com 13 Over the next decade, we believe that the successful companies in our membership will become datadriven tech suppliers. Most, if not all, decisions will be data informed, and the great majority of services will be scalable with as much automation as possible to leave service professionals as much time for high-quality interactions with customers as possible. As the business model in Figure 9 evolves, TSIA will be reporting on the best practices, most common KPIs, and levels of success. Figure 9: The B4B Data-Driven Tech Supplier 17065 Camino San Bernardo, Ste. 200 | San Diego, CA 92127 | Tel. 1.858.674.5491 | Fax. 1.858.430.3571 © Technology Services Industry Association | www.tsia.com