Pharmaceutical companies are making big investments in analytics ecosystems, but not without some disappointment in terms of ROI. Because investments are typically limited to people or technology in isolation, companies can’t deliver the kind of value that makes such programs worthwhile. At the same time, cost pressures mean that analytics and data management groups have to deliver more with less. Executives who sponsor such programs also need to be very clear on how to define the success of such initiatives. Creating hundreds of new reports does not equal success. To give your analysts the advanced tools that they need to truly succeed, you need the right combination of people, data processes and technology to get the most out of your advanced analytics investment.
What should my team look like? What kind of processes do I need to support an advanced analytics capability? What kind of technology? These are great questions to ask at the outset, and here are some answers:
- People: Go beyond cross-functional to promote cross-understanding. While you need to form a cross-functional team, don’t create specialty silos. The division of labor through specialty roles was mastered by manufacturing plants to drive efficiency and productivity, but it doesn’t drive creativity and innovation. Instead, look for people who bring a blend of business, engineering and AI/machine learning expertise. Yes, it’s hard to find such people, but you need team members with perspective across these three areas so that the solutions you build aren’t biased toward one of the three. A domain expert, for example, may trivialize the challenges with training machine learning models and have false expectations of the accuracy. Conversely, a machine learning expert may not fully appreciate the business challenges behind training a machine learning model, leading to significant effort when engineering features.
You’ll also need a leadership sponsor who can encourage a “one team” mindset and will understand the importance of providing your team with the right training across the three areas. In addition, the leader should ensure that the objectives of different groups are aligned. It can’t be left to individuals to define their own objectives based on the group they represent. Cost, efficiency, productivity and innovation often contradict each other. Having more people with cross-functional understanding reduces this friction.
- Processes: Think evolutionary, not revolutionary. The days of the big, splashy rollout of the multiyear project are over. You can no longer afford to pursue a project that takes two years to collect data and another two years to design the perfect reports. Business is changing too rapidly for such an approach to be successful. Instead, you need to build something small, build on its success, and then take it to the next level. Define the business value for this smaller solution. Have a lean, agile process. Define success metrics and KPIs that can be shared within a culture of transparency. Release a solution and demonstrate the value and rewards seen with this new system. And then move on to the next business problem and keep prioritizing as your business needs evolve.
For example, one company wanted to reduce the operational cost of data stewardship through AI, so instead of solving an enterprise-wide stewardship problem, the company picked a very specific use case and designed a model to reduce the need for human labor. This allowed the company to identify gaps in people, processes and tools, show value to executives, and build upon this success so they could move on to solving problems in other areas of the organization.
- Technology: Emphasize business architecture over technology architecture. Not only is technology changing fast, it’s also becoming a commodity. One tool that may be vital to our business today could be obsolete in a few years or a matter of months. Leading cloud vendors such as AWS, Azure and Google have made it easy for anyone to use these tools with ease. For this reason, IT must no longer pursue monolithic architectures that need to be built from the ground up. Think in terms of a subscription and web services, with plug-and-play, fit-for-purpose solutions that can be easily replaced as business evolves.
Most companies have moved away from private data centers. The public cloud gives you the ability to experiment, up-size, down-size and change quickly rather than being caught up in the old-school world of managing servers and upgrades.
As you embark on this journey, remember to start small, learn and scale. Have a program mindset but plan to deliver value over a few months rather than years. And don’t do it alone. There’s so much complexity in technology and our changing business needs: The group that drives this should not be just a business group or just an IT group. Also, remember to invest in partnerships with software vendors or managed service providers to fill in weaker capabilities on your team.
Last but certainly not least: Don’t forget change management. You can do everything right but still fail without a good handle on how you plan to manage adoption. Most changes will impact your analysts, so focus most of your attention there. Analysts will need to learn not only new tools and technologies but also the changes to the way that your data is governed and managed.
But remember, as long as you’re moving methodically and proving value as you go, your new program shouldn’t be a hard sell.
For more on this topic, be sure to visit Niroop’s session, “From BI to AI: Future-Proofing Your Analytics Investments,” at the ZS Impact Summit, held Nov. 6-7 in Chicago.
BLOG POST: The State of AI Adoption in Pharma