USi’s use of Endeca enables employees to search data from over 12 disparate systems to access the information required to help service their customers better and faster, and to decrease the time spent looking through separate information sources to track down relevant data. Although not used within the context of business intelligence, this case study shows how enterprise search can be used for analytical reporting and the management of metrics. In addition, USi’s use of search grants users access to both structured and unstructured data through a centralized user interface. Also, pulling data from independent information sources across the organization enables the organization to identify potential data quality and data integration issues.
Founded in 1998 and based in Annapolis, Maryland, USi is an Application Service Provider (ASP) with over $100 million in annual revenues. USi’s goal is to provide its 150+ enterprise customers with a highly automated, efficient, and systematic approach to delivering managed hosting, application management, remote management, professional services, SaaS enablement, and eBusiness development services. USi has over 850 employees with several regional offices throughout the United States and two global delivery centers in Hyderabad and Bangalore, India. In October of 2006, USi became a wholly-owned subsidiary of AT&T.
With almost 200 employees working to support corporate clients and with customer, account, and other required data being stored in over ten disparate systems, finding information in a timely fashion was almost impossible. The question became how to pull up-to-date data from CRM, HR, Salesforce.com, change management, etc. systems to allow access to pertinent business related information. Additionally, the goal of becoming more ITIL-compliant and increasing efficiency by reducing employee labor hours was the main focus of corporate management. Toby Ford, CTO of USi, identified the main project goal as being to “gather information from many different sources and to reduce employee labor hours by arming employees with more pointed information and by allowing them to search through things to get to information more readily.”
At the same time as USi’s comprehensive evaluation of the organization’s requirements regarding how to access large data sets in a timely fashion, a client was conducting an evaluation of Endeca. Based on successful results at the client site of their evaluation of Endeca for use as an e-commerce catalog search engine, USi decided to prototype the application internally. Additionally, Endeca matched the general requirements of what USi needed to grant employees access to the information required in a timely fashion.
The goal was to use search to decrease data redundancies and to increase process efficiencies. Their hopes were that search would provide employee access to large amounts of information timely, including data stored within Oracle and SQL databases. Also, an in-depth analysis was conducted on employees’ daily tasks, including the amount of click throughs required to find information to see how search could reduce the amount of time to access client information. For instance, how long it took employees to identify what charge codes match which client and what client access charge codes are linked for direct billings purpose. Their prototyping yielded successful results and, after an intense design phase, the organization proceeded to implement Endeca for use by support staff managing client accounts.
The initial implementation took four months to complete, from design through delivery of the presentation layer. This included leveraging ten disparate data sources from PeopleSoft CRM, HR, contract management, the organization’s wiki, and the integration of Salesforce.com. With the addition of Salesforce.com, the organization’s initial goal of decreasing labor hours for report staff was expanded to include providing sales staff with access to sales pipelines without having to look through user accounts.
Benefits and unknown BI
Some unexpected benefits achieved within this project were reporting, data presentation, and data cleansing. Because so many information sources were leveraged to create a unique access point, the differences in data structure and presentation became apparent as search results yielded outcomes that were represented in many different ways due to the structure of information within the source systems.
The creation of a search engine combined with a presentation layer became the basis for additional reporting. For example, the related information pulled from disparate applications enabled employees to see a more complete picture of a customer, their accounts, and the interrelationships that may exist across regions or by customer type. In addition, due to the natural groupings created within search results, unintended relationships were identified, such as the identification of employees at the same location. Because of the ability to report through search, management could identify the number of years of experience and specific expertise employees had through histograms to enable the organization to make better decisions about how to allocate resources to projects.
Search results exposed also the differences in data representation within source systems and identified keystroke errors. Aside from different structures, using search enabled the organization to identify integration issues such as data cleansing requirements. To continue to provide a high level of results, the organization was required to clean up the data sources. With cleansed data across the organization, a more complete customer view was created.
Some challenges encountered were attaining end user buy-in, developing the system design, training, and metrics identification. With every project, challenges enable the organization to improve upon future initiatives.
In many projects, business units or IT identify business problems, but may not be able to attain management buy-in. In these cases, it is rare that a project succeeds. In this case, the project initiative followed a top-down approach, meaning that the project had management buy-in from the start. However, even though management included end users within the process, attaining end user buy-in was difficult. In many cases, disparate business units each felt that the systems they used within their departments could be expanded to provide an overall solution to the organization. Additionally, because of heavy workloads, many employees did not have extra time to add another initiative to their schedules.
Defining system requirements was difficult because this project was a new initiative within the organization. Due to the iterative process of connecting to the required data sources and offering a solution that met the requirements of multiple employees, the development of USi’s enterprise search took several steps. Also, employees had demands of accessing timely information as opposed to weekly or monthly updates. This made the identification phase difficult and time consuming as information needed to be accessed via operational data stores versus a data warehouse. However, this attention to detail in the beginning paid off by eliminating the need for additional requirements once implementation was complete.
Since over 200 support staff were included in the user base with the first release, training had to take place on a wide scale. To be most successful, training was conducted in several small groups. Initially, usage of the new system was high but began to drop of after a time, requiring refresher training courses. USi had to learn to balance initial training initiatives and keeping the momentum of the new system going with ongoing training efforts.
USi’s initial project success metric was to identify the number of clients per support staff to match that with the amount of support time being used. Due to the 2006 acquisition of the organization by AT&T and the growth of overall employees from 500 to 1000 people, the initial metric to determine success was no longer applicable. USi didn’t take into account the potential exponential growth that would directly affect the measurement of project success.
Recommendations and lessons learned
Implementing enterprise search to provide end users with relevant information to enable them to do their jobs more efficiently is not always intuitive. Organizations require in-depth planning activities to lead to a successful implementation. With USi, benefits were seen within seven months of implementation, despite the exponential growth of employees due to the acquisition. This shows that even though unexpected activities may affect the outcome of a project, these factors can be used to highlight project success as well. Because of USi’s use of Endeca, integrating AT&T data into the mix becomes the next logical step as opposed to a new business problem.
The importance of delivering timely information to employees cannot be offset by the fact that those employees need to have access to proper training. This includes refresher courses with the increase of users over time, such as new employees or additional departments. Also, the increase in the number of users generally relates to additional uses within the organization. For instance, Salesforce.com data being leveraged by the sales staff to identify sales pipelines. Therefore, organizations should take into account potential future use during the initial project phases to identify additional data sources and potential integration issues so that when extensions occur there are no additional surprises.
(Copyright 2007 - Dashboard Insight - All rights reserved.)
About the Author
Lyndsay Wise is an industry analyst for business intelligence. For over seven years, she has assisted clients in business systems analysis, software selection and implementation of enterprise applications. Lyndsay is a monthly columnist for DMReview and conducts research of leading technologies, products and vendors in business intelligence, marketing performance management, master data management, and unstructured data. She can be reached at email@example.com. Please visit Lyndsay's blog at myblog.wiseanalytics.com.