Business User Empowerment

Business User Empowerment through Collaborative Analytics CHI 2008 “Social Data Analysis” Workshop: Position Paper

Daniela Busse

Introduction

SAP Labs (Palo Alto) 3475 Deer Creek Rd Palo Alto, CA 94304 USA [email protected] Richard Hong SAP Labs (Palo Alto)

“Data analysis is traditionally thought of as something done by experts in isolation or in small groups. Social data analysis, however, suggests the possibility of massive collaboration in the discovery process, involving experts and non-experts alike.”

3420 Hillview Ave Palo Alto, CA 94304 USA [email protected]

Copyright is held by the author/owner(s). CHI 2008, April 5 – April 10, 2008, Florence, Italy ACM 1-xxxxxxxxxxxxxxxxxx.

To this day, the power of social data analysis (i.e. ‘collective analysis of data supported by social interaction’) is not yet tapped into successfully by enterprise software, despite evidence to the fact that collaborative analytics is a fact of life for many business users, and related requirements are being communicated by our customers loud and clear. In today’s enterprise software, ‘old school’ approaches to (heavy-weight) analytics abound: monolithic data warehouses are interfaced to via slow and cumbersome reporting tools, that deliver static, asynchronous, and

2

single user snapshots, a.k.a. business intelligence reports.

projects run at SAP Labs (Palo Alto), and attempt to identify areas of opportunity for social data analysis.

However, the real ‘intelligence’ that analytics could provide for business users is articulated in visions of future analytics capabilities, that:

Collaborative Analytics in the Enterprise




are embedded contextually in the everyday work activities of professional users (not isolated in a separate reporting environment)




represent dynamic and real-time windows into the enterprise data (not sealed-off snapshots)




allow an interaction with the data & related activities that is easily accessible and intuitive, i.e. easy to use for any level of user




are supported by tools that allow multifaceted collaboration on and around the data and the analytics results




eventually will make use of the “wisdom of the crowds” of professional users in socially, continuously creating & interpreting the data cloud available in the enterprise space for better business performance

In this position paper, the authors will briefly outline their thoughts, ideas & insights into two of these areas (collaboration support and embedded analytics), illustrate those with examples from past & current

In the current model of Business Intelligence, all to often specialized roles are created within companies (such as “business analysts” or, more informally, “our SQL reports guy”..) whose sole role is to be familiar with cumbersome reporting interfaces, and (hopefully) with having an overview of the data that’s available in the system for reporting use. These reporting users are then responsible for fielding questions from the user pool they service, and answer them through business reports in regular intervals or one-off occasions. The target recipients of these reports typically are the knowledge workers at that company, up to c-level users that often need a higher level of aggregation, and a perspective across the value chain (often focusing on the bottom-line impact of whatever operational process under investigation). Information pyramid Here is another level of detail on the different user roles in enterprise analytics, to annotate Fig 1 below. C-level (executives): 360 degree view of information because of the position on the pyramid. They can turn around 360 and look at the info at all direction but only within a short range so the level is high, not very detailed. Only one dimensional collaboration. Mid-level: Business analysts and information workers have overview of information in a specific domain and also relatively detailed information. They have two dimensional collaborations (vertical and horizontal).

3

O-level (operational officers and workers): These people in the lowest level of information, the original application transactional data, have two dimensional collaborations but they cannot see too far in both directions.

Financial Operational Steering: the “S&OP” Process For instance, during the ‘Sales and Operations’ process, in which different business units come together in regular meetings in order to share and compare their data, unearth & explain potential issues, and aim at coherence in data and interpretation across business units, is a good example of a business user activity that is critical to good business performance, has a rich use of social data analysis, but is currently completely unsupported by enterprise software tools. The work is done manually, in meetings and legwork, often painfully and laboriously, to make sure that overall the bottom line impact can be calculated by informed planactuals comparisons across the value chain. Overall, the finance department will need to be able identify how the company overall manages to achieve their earning estimates on a quarterly and yearly basis. We proposed a solution for this business activity (see Fig. 1) that highlights the collaborative needs of the financial controller in gathering and interpreting financial and operational data from various business units, and in following up on potential issues (i.e. exception handling).

Figure 1 – information Pyramid: Collaborative Analytics across User Roles

In today’s world of work, however, the expectation is there from information workers and c-levels alike, to be able to have these data & interpretations available at their fingertips, and being able to modify and manipulate them themselves, ad-hoc, and embedded in their current work activities.

We also think that this area, however, would be a great case study for social data analysis as discussed in this workshop: we envision the business units getting actively involved in maintaining their data, in the interpretation & comparison of it, and in the unearthing of issues that span multiple departments – rather than these activities being bundled in the hands of the financial controller, as is the case today.

4

Exploring opportunities for social data analysis in contexts such as this one will be one aim of the authors in the proposed workshop.

Figure 2 - Collaborative Analytics for Financial Controlling in the "S&OP" Process in Enterprise Companies

Embedded Analytics Embedded analytics is a new trend in business that promotes the concept doing analytics right in the business process context, or in many cases people’s social context. It is all about empowering the end users

to do analytics without any IT support because they do it right inside their familiar territory, their business application. Traditional analytics is tool-based centralized process where people have to a sophisticated tool at a fixed space. Also, the data used for analysis is based on historical data. This time and space constraints made business analytics the task for IT professionals and savvy business analysts. The new generation analytics, powered by real-time analytics and embedding, is increasingly becoming pervasive which breaks the time and space barriers. The social interactions are crossing not only multiple communication channels (face-to-face, voice (VoIP, voicemail), websites, SMS, E-Mail, group calendaring, blog, podcasting) but also spaces (conference room, team rooms, discussion forum, focus groups, webconferencing, wiki). These are becoming perfect media for embedded analytics. The collective thinking, augmented by the new media, is transformed to collective intelligence to guide people make the right decisions. There are different types of embedded analytics. The way of embedding is evolving as the technology develops. (1) view level embedding: this is a very basic and primitive way of embedding. It uses an analytics view inside portal entry pages which lead user to an analytics tool. This is, in many cases, called embedding but it is really an entry to the tools, you can’t really solve the problem within your business context. Advantage of view level embedding is simple to implement. (2) Data level embedding: This embedding is right in the context of application, or business process. The transactional table representing the business context

5

contains the fields coming from analytics data source. One example is the spend management system where you can track spending on a particular commodity category by checking the progress to the planned value which is coming from analytics data source. You can use this information to adjust the target right inside the application without actually leaving from it. (3) Activity level embedding: There are many cases you need analytics information in the middle of the business activities. For example, in procurement sourcing process, the selection step for qualified suppliers to invite is ideally coming from the supplier evaluation analytics based on the past delivery, freight, location, etc. Also in this case, the process to evaluate the qualified suppliers is a collective process: typically the purchasers will upload their own records of the suppliers they dealt with in the past and consolidated in the BI system and evaluate them together manually or

use certain kind of criteria and determine the final ranking. (4) Context aware embedding: This level of embedding is like smart tags (address can be tokenized to link to a map, product name can be link to product specification page, etc). All numbers, alert messages, locations, time, etc in any document can all be smart tagged to analytics information. This also includes the on-going document which is in the process editing. This gives the user the insights what the significance of the number the user entered is. This level embedding can also introduce location specific analytics powered by GIS. For example, customer analytics retrieved once you are onsite.

The Authors

Daniela Busse, Ph.D. Daniela Busse is a Product Owner and User Experience Expert in the Business Process Renovation group – an internal, global innovation team in the Office of the CEO at SAP Labs that lives and promotes at design-led innovation in the enterprise space.

Richard Hong Head of SAP Analytics User Experience Design and Architect responsible for analytics application design and analytics user interface framework, currently working on SAP’s new generation analytics framework with special focus on collaborative and embedded analytics.