Feature

posted 15 May 2006 in Volume 3 Issue 1

Project planning for the 21st Century

At pharmaceutical giant AstraZeneca, a data mining application has given project managers access to the information they need to run drug development projects more efficiently, to get better use of available resources and to pay closer attention to cost control.

By David Scanlon

As one of the world’s leading pharmaceutical companies, AstraZeneca has to focus on turning good ideas into innovative, effective medicines that can make a real difference in important areas of healthcare.

The company employs some 64,000 people worldwide. Of these, around 12,000 work in research and development (R&D) centres in any one of six international sites. Reflecting the importance of R&D to the health of a drug company, AstraZeneca spends more than $3 billion on R&D every year – equivalent to more than $10 million every working day.

Drug development is a highly competitive activity and every day saved in getting a new branded drug to market can be measured in millions of dollars. Patents for pharmaceuticals last between 20 and 25 years, depending on the country or region, but it takes between eight and 12 years to bring a new drug to market. The shorter the product development time and clinical trial, the longer a company has to establish brand leadership and generate revenue before a competitor can bring generic products to market.

Reducing time-to-market depends on a wide range of factors, including efficient project management, careful resource utilisation and rigorous attention to cost control. For employees, meanwhile, immediate access to one consistent view of R&D information is absolutely critical. Research and development is the engine of value generation for AstraZeneca and it is the project teams that drive the value generation through the efficient and effective delivery of projects.

The Matrix

With that in mind, AstraZeneca developed MATRIX, an Oracle-based data warehouse for project managers. A number of tools are used to extract or ‘mine’ the information that MATRIX holds in order to provide complete visibility into R&D information across the world: a business intelligence (BI) system called MARS and reporting tools from Business Objects. The focus of this case study is the implementation of MARS at AstraZeneca. In essence, MARS is designed to support the following:

Project planning;
Scheduling of projects;
Estimates of costs and resources for projects;
Control of projects;
Provision of project information.

As such, MARS is optimised for the extraction of such data from the MATRIX data warehouse.

Project history

The MARS project goes back about two-and-a-half years to an AstraZeneca internal project management conference in the fourth quarter of 2003, where the shortcomings of existing project management systems were discussed. This conference was attended by the top 200 people in drug development – a reflection of its importance to both the company and the people that rely on that information.

At this point the project team had developed a concept – in just six weeks – to demonstrate the potential future of reporting in AstraZeneca. One of the problems with data warehouses and business intelligence tools is their complexity. We therefore knew that we needed to provide a relatively easy-to-use tool to project managers – one that would enable them to wring the information they needed from the system without demanding a degree in advanced software engineering from them.

The response to this early concept from users was encouraging – ‘When can we have this?’ they asked. It was a major validation of our ideas, but also presented us with a significant challenge in terms of us having to meet and manage their expectations. Indeed, MARS needed to meet many different demands, so we went back to the users to find out exactly what they wanted. They said they needed the following:

A reporting tool connected to MATRIX;
A powerful analysis tool providing complex portfolio and benchmarking information;
A tool that could give them the right information at the right time;
A tool that would be simple to use;
The capability to mine fully all the rich data in MATRIX;
The capability to release the full potential of historic data;
A transparent vehicle for data quality issues, so that further quality improvements could be easily and transparently moved forward;
A system providing integrated project information.

A key part of the deliverable was going to be a tool owned by the business, which could be evolved by the business according to its own needs.

The basic goal, meanwhile, was simply to give people their data back. When we talked to people, we felt that there was a strong perception that existing information systems had become ‘black holes’, which sucked information in that would subsequently prove difficult to get back out again.

That is not an uncommon problem: a fundamental flaw with many business intelligence (BI) systems is their inability to present complex information to users in a way that is simple and intuitive for them to see and manipulate. Unfortunately, our existing Business Objects tools – designed to meet the needs of BI experts rather than business users – fell into that category. They were not perceived as ‘user friendly’ and expertise in using the technology tends to reside in particular ‘pockets’ of our organisation, mainly finance and accounting groups.

Given these usability issues, we were therefore strongly attracted to the new breed of ‘visual analytics’ software that was, serendipitously, beginning to emerge at the time. This software promises to make it easier for end-users to understand information presented to them from multiple sources – and can create simple visual representations of data, such as charts, graphs and maps. These are easier for users to understand and follow than pages of data, and the tools are more flexible, too. With a few clicks, users can manipulate the visual representations, to provide many views of data.

As a result, we selected a visual analytics package, called Discovery, from software supplier Advizor Solutions, which we could bolt on as a front-end for our existing – but over-complex – Business Objects reporting software.

That means that we have not discarded Business Objects, but use the new system in combination with it. By bolting on Advizor to the Business Objects environment, we have the ability to present project managers with graphs they can easily understand without them having to acquire any particular BI expertise. In this way, they can access and examine information on project schedules, costs, and resources and measure them at a glance against key performance indicators (KPIs) across R&D.

We had originally decided to develop the front-end for project managers in-house, but the need to do so was fortuitously mitigated by the delivery, ahead of schedule, of a new version of Advizor software which could effectively replace the in-house system that we had been working on.

This was a truly collaborative development in which innovations from AstraZeneca directly led to further product development for Advizor Solutions, leveraged through our partnership with our systems integration partner, E-Matrix Global (EMG). EMG is one of only a few Advizor platinum partners and has one of the largest installed bases of Advizor licences in the world.

Our MARS project provides a perfect example of the benefits of being an early adopter of a leading-edge technology. Because it is a new technology, both EMG and Advizor have worked hard to meet the demanding requirements of MATRIX users within AstraZeneca.

At the same time, this has helped the software supplier to develop its product from a largely scripted application requiring ongoing support, to an analytical development environment (or ADE) for dashboards and has also helped with its integration within the Microsoft Office environment.

Direct impact

MARS has had a big impact on the way project information is accessed, viewed and shared across AstraZeneca. Senior management, development teams and therapeutic area teams now have instant access to up-to-date R&D information to help them make quick, confident decisions. Easy access to R&D information also gives senior management at AstraZeneca a clearer picture of new-product project costs and helps them to improve project scheduling, budgeting and resource allocation.

Of course, systems such as MARS are only as good as their data sources and the information put in by users. A limitation of the information in the MATRIX data warehouse was perceived to be the quality of data related to scheduling information (in particular, project ‘milestone’ and ‘tollgate’ information). One benefit of MARS has been the improvement in the quality of milestone data quality within MATRIX by enabling global project managers to effectively manage the quality of their own data more closely.

From external presentations and discussions with peer companies, we believe that the MARS development has helped put AstraZeneca’s project management capabilities among the best in the industry. It has led to greater transparency of the drug development process at the company. MARS was instrumental in supporting key business analysis of risks across the portfolio, as part of an internal review.

One of the most surprising, but pleasing, aspects of the work was the collaboration among a diverse and diffuse project management community. By effectively displaying the ranges of delivery, global project managers are able to see who is more effective in delivering projects through the development process.

As an instrumental part of a risk assessment of our early portfolio, the work of the MARS team has enabled more effective and transparent discussion of risk. While this has always been a strong aspect of project management at AstraZeneca, MARS has provided greater transparency of the data that supports such decisions.

The business analysis team has been able to provide the correct data to the right people in hours, rather than weeks, by more efficient use of the data in our systems. This has been witnessed at functional, project and board level.

Having this higher level of quality data has helped staff to make further, deeper analysis thereby supporting the transparency of data all the way through to the board of AstraZeneca.

MARS offers plenty of other possibilities for AstraZeneca, too. Behind the MARS tool is one of the most powerful data analysis tools available on the market today, which can easily be applied to other areas of business. Its capabilities for analysts to understand complex data from systems and publish that easily in various formats will be of help to other departments in our company.

So far, ‘proof of principle’ exercises have been carried out to demonstrate that the system can be extended into a wide range of areas, including:

Project management baseline reports;
Portfolio analysis;
Regulatory analysis;
Clinical analysis;
Productivity analysis and reporting;
Industry analysis;
Discovery informatics;
External benchmarking.

This will encourage the sharing of good practice and help the company in its process of continual improvement.

David Scanlon was the lead project manager on the implementation of MARS at AstraZeneca. He has since moved into another role at the company, but the project’s original partner, E-Matrix Global (EMG) is still involved in the MARS project and, in particular, an enterprise programme to roll out Advizor, the technology behind MARS, across AstraZeneca. For more details, contact George Frangou, CEO of E-Matrix Global, at gfrangou@ematrixglobal.com

Project facts and figures

The challenge

To improve visibility and access of global research and development (R&D) project management information.

The solution

Business Objects provides summary information via reports to more than 5,000 R&D staff globally;
Advizor provides highly visual and self-service advanced reporting & analysis;
Delivered to more than 1,000 R&D users/managers globally;
Advizor delivers key performance indicators to senior management.

The benefits

Global transparency of R&D information;
Improved resource utilisation;
Tighter cost control;
Improved time-to-market for pharmaceutical products.

Development techniques for creating analytic applications

Analytic applications

An analytic application is not just a set of randomly created reports in a person’s in-box or "My reports" folder. An analytic application contains business logic that takes users, step-by-step, through a series of interactive reports so that they can access, analyse and take action to optimise the activities in a specific business domain. Although some organisations currently deploy dozens of analytic applications to hundreds of users, most do not and are still inching their way toward maturity in this sector.

Customisation challenges

Building intuitive analytic applications is not easy. Organisations spend too much time customising and extending commercial products to meet user requirements – products sold to them as ‘packages’. On average, organisations customise about one-third of the final analytic application, mostly using SQL and other hand-written coding languages. The total process takes seven-and-a-half months, on average, which is simply too long to meet fast-changing user needs.

In practice, organisations are most likely to build analytic applications around a BI tool, not surprisingly, but then add substantial amounts of their own custom code (mostly in SQL) to customise and extend the application. Most organisations also customise the BI tool itself, focusing on the graphical user interface (GUI), calculations and navigational elements.

Developers spend significant time customising mappings for extraction, transformation and loading (ETL) of data from source systems to the data warehouse, not to mention designing data models in packaged applications. Developers, who are mainly IT staff and application programmers, make frequent changes to analytic applications, and power-users are also often enlisted to help change the front-end environment.

Buy and extend using ADEs

BI vendors have recognised the need to deliver "buy and extend" capabilities. Most are starting to deliver analytic development environments (ADEs), which expose BI components in a graphical drag-and-drop development interface. ADEs are the analytic complement of integrated development environments (IDEs), which are used to build transactional business applications. ADEs promise to accelerate development time and reduce costs by enabling power-users in every workgroup to tailor an analytic template for their immediate colleagues.

There are a wide range of ADEs on the market today. Pure ADEs, like those from Arcplan, Business Objects, Microsoft, ProClarity and SAP give developers almost unlimited control over the look and feel of an application and the way users can navigate through it. Newer report authoring tools, such as those from MicroStrategy and Advizor Solutions, give report developers greater flexibility to create a range of reports or dashboards to meet the unique requirements of a broad range of end-users.

ADEs for dashboards

Many ADEs are now used to build customised dashboards and scorecards, which are quickly becoming the most popular way for users to navigate analytic information. While most dashboards today are strategic in nature and enterprise in scale, the number and type of users supported indicate that we are still in the early stages of dashboard deployments.

Business intelligence and data warehouse investments have been underway in earnest for more than a decade, leading to an overall maturing of infrastructures that has opened the way for the widespread adoption of advanced reporting and analysis solutions across a range of industries.

During this period, the boom in data warehousing and e-commerce has helped businesses to amass unprecedented amounts of data – causing them to be data-rich, but information-poor. This phenomenon has rendered obsolete many traditional forms of data analysis and reporting, which have focused on the ‘What?’ whereas visual analytics software from vendors such as Advizor answers the ‘Why? and enables users to quickly analyse and present ‘highly dimensional’ business data.

Source: The Data Warehousing Institute