Background:
Based in Seattle, Washington, the Fred Hutchinson Cancer Research Center (Fred Hutch) is a world leader in the detection, prevention and treatment of cancer, HIV/AIDS and other infectious diseases. A decade ago, the Center entered a consortium with the University of Washington Medical Center and the Seattle Children’s Hospital to form the Seattle Cancer Care Alliance. With partners responsible for the delivery of patient care, Fred Hutch was better able to focus on its research agenda. Currently, the Center houses more than 300 scientists working in five scientific disciplines, as well as an equivalent number of their support workers, in addition to over 4 thousand employees in other areas.
The challenge:
Fred Hutch is largely sponsor funded. Approximately 80 percent of the Center’s $400 million annual operating budget is funded through grants, contracts and awards from the federal government, the private sector and other organizations, such as the Gates Foundation. According to Stephanie Dennon, BI and planning manager for Financial Management Information Systems (FMIS) at Fred Hutch, awards come with stringent guidelines on spending: “You spend down to the dollar, down to the penny and you try to get the most research you can for your money.” And these guidelines include a requirement to track and report spending, a task that the Principal Investigator (PI) who has typically secured funding for the research program, executes with the help of administrative staff.
Within Fred Hutch, the finance organization’s (FMIS) job is to support the reporting function by making sure that scientists have the information they need to track their budgets and provide regular health checks back to their funding organizations. “We try to provide as much data as we can so they can build up forecasts,” Dennon explained, and “this is where a lot of our challenges lie.” The Center has grown considerably in the last 10 to 15 years, and senior leadership is looking to continue that expansion over the next several years. “All of this requires faster, more flexible analytical solutions so that we can provide the information scientists and their support staff need to continue to perform their research, and do it in a financially sensible and fiscally responsible way,” she added.
The scientists also have a responsibility to FMIS, which oversees and manages reporting on behalf of its research “customers.” Accountable for maintaining compliance with regulatory standards, the financial organization maintains an “Office of Sponsored Research” that provides oversight, and helps manage grants and contract funding to ensure that everyone is in compliance. At a local level, the scientists view Fred Hutch as a “research motel” where they can bring their own funding, but look to the financial organization for help with managing funds that in some cases come from more than 200 different sources. Scientists have their own staff, targets, goals, and standards of operation that are driven by the terms of their own awards, but FMIS is charged with providing the data that will facilitate program management.
This decentralization of management has created its own issues, including problems with data consistency that silos of information, and investment in multiple, different information management systems can create. As Dennon explained: “One of our biggest challenges is getting to that illusive single source of the truth, which is more a matter of perspective than an absolute. In finance, we do all of our reporting out of our specific ERP sources of books of record. But once that information goes out to the local units in the form of reports or data downloads, what they do with it is largely uncontrolled. We were fighting the dreaded ‘shadow book’ – souped up Excel spreadsheets that do everything the scientists could possibly want, to which they added data that isn’t relevant to anything outside individual programs. Inevitably, they [local scientists] would come up with a different number than you would see on the bottom line of our reporting.”
At the same time, users at the local level were experiencing frustration with the inefficient reporting that could be achieved through the legacy system. Fred Hutch had deployed Crystal Reports with PDF output that could not be manipulated or changed – and there were no analytics. “Users printed their reports and rekeyed them into an Excel spreadsheet, so it was very inefficient and very slow,” Dennon explained. “There was a lot of work for the support staff, who would complain to the PI that they couldn’t get the required information because they were too busy rekeying and doing non-value add tasks.” There were also issues with performance: the Chrystal Reports installation sat on top of the Fred Hutch production ERP database, and though the team created a separate database copy once reports began to drag, as Dennon noted, “nothing in a relational database is optimized for reporting. So it was always very slow, and all the calculations happened in the reporting layer instead of the data layer.”
Building the business case for change:
While senior management within finance was very supportive of the need to upgrade legacy systems in order to provide researchers with better access to more consistent data more quickly, funding at Fred Hutch goes through a scientific committee that is generally more focused on directing resources to scientific research rather than to administrative activities. And because it was so old, the legacy system had exhausted its support term and Fred Hutch was no longer paying maintenance fees. This set a high ROI bar for any new project.
To address the committee’s concerns, Dennon’s team built a strong business case for savings, based on quantifying the soft benefits a new system could deliver. They began by interviewing administrative workers who were responsible for the reporting function to uncover the specific soft benefits, surveying them on how much time they were spending doing non-value added tasks, such as keying data into spreadsheets, or trying to run reports that take 10 minutes or that do not complete. Ultimately, Dennon was able to calculate the hours per month devoted to these activities, and compare this with early estimates on what productivity gains would be delivered through new technology; the result was quantified as a number of FTEs (full time employees) that would be saved. “That made a really compelling case,” Dennon noted, “as did some of the user stories, especially when staff were able to tell the scientists, ‘this is my pain’.”
Integrated Information Builders deployment:
To resolve issues with data accessibility and reporting, Fred Hutch decided to replace their aging Crystal Reports system with the more versatile Information Builders’ WebFOCUS analytics platform. WebFOCUS would deliver new efficiency in meeting financial reporting requirements, while providing portal technology that easily combined reports and graphs in a flexible and user-friendly interface. The Information Builders implementation has been live for just over a year and a half, and so far, the Center has tackled part of the financial reporting solution – it is slowly phasing out the legacy system and phasing in WebFOCUS through progressive integration with additional data sets.
As part of the WebFOCUS deployment, the organization revamped its underlying data infrastructure: Fred Hutch now has a series of data marts that come out of its PeopleSoft Financial ERP system. Inside the ERP is a denormalized table that captures all detailed transactions from subsystems and puts this in one place. From here, the Center rolls all data up to a summary level, and these two tables, along with a couple of additional tables for dimensions and reference, are put into a reporting database. Based on SQL server, this database was created using Information Builders’ iWay Data Migrator, which moves data from the ERP into the reporting database. The initial implementation included work from the point of the data migrator and the creation of the ETL logic to the reporting database. As Dennon explained, “we are developing our data at the same time we are developing our reporting. It’s really an evolution for us – with each phase, we have a new way of looking at the data that we have to incorporate into our data mart.” Ultimately, the team is working towards a full on data warehouse that will accommodate more financial data, and potentially data from other areas such as HR and other related departments, but for now has layered WebFOCUS over its data marts.
Change management to drive adoption:
Dennon characterized reporting at the grassroots level before WebFOCUS as a “kind of like guerilla research where you are on a shoestring budget and you’ve got to get it done the best you can.” And though this ad hoc approach had become “a pervasive mentality,” armed with WebFOCUS the financial team was able to engage in discussions with program areas on a solution that would provide some local flexibility, while improving reporting efficiency and data consistency. “We are trying to balance rooting out the ‘shadow books’ with providing flexibility so that the scientist can take something like the financial report of record and add, append and massage data as they see fit – to put in some flexible tooling in a flexible data set that will also enable track back to what the golden record is, the number that we all agree is the bottom line.”
To encourage adoption of the solution, the team engaged in extensive communication and socialization of the WebFOCUS initiative. “No matter how much people hated the old system, the idea that they had to move to something new was painful,” Dennon observed. FMIS’s tactic in this struggle for solution acceptance was to win over the administrative workers tasked with the actual reporting work. Since the PIs are focused on curing cancer – rather than on the use of a standardized tool to manage funds – to support the initial implementation Dennon’s team targeted a small user group made up of the support staff who worked right under the top scientist layer.
A key piece of this effort consisted of iterative portal development aimed at creating a user interface that would be intuitive and easy to use. Starting with business analysts from FMIS, who had an average 10-15 years of experience in the organization and deep understanding of the business needs, and working with RDC Business Solutions (RDC Biz), an implementer with expertise in WebFOCUS UI and the design, FMIS held a number of design sessions. The goal was to talk through what elements to keep in a new system, what items in existing parameterized reports were no longer useful, what the portal objects should be and how different row/column reports could be re-imagined. Conclusions were brought back to the user group, and in combined design discussions, users would vet the design or suggest changes.
According to Dennon, these efforts to engage users from the beginning of the process of building the Hutch Analysis and Reporting Portal (HARP) paid off. By demonstrating to the test group how they could relieve some of their work burden, the FMIS project team turned these users into supporters of the initiative, proponents that would carry the message across the organization. They became ambassadors who could support the roll out of WebFOCUS to the remaining 300 support staff by distributing information and answering questions. “They are great partners for us,” Dennon noted, an important addition to end user training initiatives.
Phase one outcomes:
Scientists want simple answers: can they extend their research, buy a new piece of equipment, or hire another scientist? But prior to the deployment of WebFOCUS, support staff were unable to quickly or reliably provide that kind of information. With HARP, staff can deliver answers in minutes rather than weeks through ready access to graphs that show expenses by resource type, category, or other attributes. Since WebFOCUS supports rich integration with Excel, they no longer have to key in data, and system performance has been enhanced by moving to a data mart structure that is optimized for reporting.
Beyond more rapid response to simple queries, the Information Builders platform has also created new visualization capabilities. Many of the objects deployed in the HARP portal represent real level, pre-built graphs or charts that users can see immediately by logging in and pushing a button or two – they no longer have to run a report, do data massaging or run their own calculations to see, for example, where they are with spending relative to budget. Users can also drill down to see greater data detail off the objects, and in the interests of improving productivity even further, FMIS is now working on moving staff towards monitoring using an exceptions-based type of reporting structure so that they will only drill down into data that looks out of the ordinary.
Next steps:
According to Dennon, users are now anxious for the next phases of the portal: “stemming the tide is now our challenge, but it’s a good challenge to have.” Moving forward, FMIS is looking to further optimize use of the HARP portal. For example, they have identified a couple of calculations where the user could potentially get some data errors, and so are building in a “red flag” based on a percentage threshold that would isolate data points that may be problematic and need further investigation. FMIS is also hoping to develop the portal visualizations to allow users to capture trends by seeing the data rather than having to crunch numbers.
The team is also working towards complete exit out of legacy systems. To support this, and to provide additional capability, FMIS has upgraded WebFOCUS, positioning to capture some of the functionality in v. 8.2. Specifically, the team is looking to leverage responsive portal technology to give users the most space possible on the screen. Currently, there are a couple of locked elements, and it would be helpful to slide these around and change the organization of the portal so it is optimized for viewing. Fred Hutch is also focused on offering users the ability to use any kind of device – rather than tie information access to a desktop PC. While the Center is just now approaching mobile deployment, tablets are standard issue, Dennon explained, and the science population consists of a lot of MAC users: “being able to deploy on any browser, any platform, and device is a big win for us.”
For future:
Down the road, FMIS is looking to reach additional user constituencies. A “game changer for us in terms of providing flexibility is the Infosys Plus,” Dennon explained, as it would enable them to roll out WebFOCUS capability to power users who sit in the local program areas who could then develop access to the data sets and build reports independently. This would allow department heads and administrators in the scientific divisions to quickly answer questions that are very specific to their respective areas, structured in a way that is specific to local needs. Currently, she noted, Fred Hutch builds solutions based on “an 80/20 rule” – and this roll out would satisfy the needs of the 20 percent. Ultimately, the goal is to deploy an interface or dashboard that would allow the PIs themselves to directly serve up information.
Best practice guidance:
- To prepare for initial implementation, the FMIS had laid out a detailed roadmap. However, deployment was not simple; there were a number of complex elements that had to be taken into account, including the human. “Change management can never be underestimated,” Dennon explained. “You have to hold your user’s hand all the way through the process, provide them with support, and provide them with an outlet to vent their frustrations and to answer their questions – you have to lead them into the process.”
- Stabilization of the system is critical. There is always a backlog of changes/new features that need to be addressed, but implementers need to make good on their promises to maintain the trust of the user group.
- The need to build out the internal support structure should also be included in planning. Dennon devoted several months to finding additional resources for her team, but felt this was necessary to the development of a realistic roadmap. FMIS worked with a developer from its implementation partner, IT hired a systems administrator, FMIS also engaged a report developer, and six months later, found it had to hire another project analyst to support management of the solution rollout. Over time, the project team grew – and that involves training and effort that should be budgeted.
