In an Accenture survey1, 94 percent of industry executives recognize artificial intelligence as important in driving innovation and achieving outcomes such as hyper-personalized experiences, new sources of growth, and new levels of efficiency on the Commercial side of the business. But despite recognizing the opportunities, our life sciences leaders are often unsure how to get started.
Standing still is not an option. If Commercial teams continue to go after the same problems in the same ways, they’ll remain stuck with the same issues. AI solves a lot of the longstanding problems that they’ve been grappling with for so long. We’ve identified four steps to get started:
Pick your use-cases
In identifying your own business problems, always trust your intuition. No-one knows your business better than you. Having pinpointed the problem you’d like to solve, the next step is to determine the value drivers and benefits of launching an AI solution to do this. Based on these value drivers, you can identify the optimal focus areas for your organization and tie them back to the potential benefits of AI—such as better insights, cost savings and enhanced interaction.
It’s also vital to build your use-cases and conversations around specific user journeys, grounded in the benefits or outcomes you hope to achieve, rather than just promoting a specific technology.
Activate an AI taskforce and senior sponsor
In any AI deployment, it’s essential to accept that ‘failure is an option.’ AI is, after all, emerging technology and the industry is still working through all the potential use-cases. Eighty percent of ideas go nowhere—but this is more than offset by the 20 percent that succeed.
Given the inevitability that some AI initiatives will fail, it’s important to maintain the momentum (and ensure ongoing buy-in) by educating the executives in your organization about AI and what it can do for the business. You’ll also need to secure a senior AI sponsor with budget allocated to get things moving. A further imperative is to consider the availability of resources and the right operating model. Do you have the talent and skills within your group to move the needle?
Build out the data architecture
Developing a machine-learning application demands large, relevant data-sets to train and test your computer model. This means centralizing and indexing your data assets—whether collected internally or bought in from external partners—to prepare your organization for the move to predictive and proactive decision-making.
The ground rules for AI deployments apply here too: start with running a small data health assessment, rather than embarking on an enterprise-wide project. Approach this as if you were triaging: this includes going through certain basic data checkpoints like data hygiene, focusing on data maturity and data governance, breaking down data silos, and streamlining fragmented data across database to achieve some quick wins.
There’s no reason to approach this alone. These breakthrough capabilities may already exist in the business within the R&D teams or can be leveraged from external partners. This means there may be ready-made opportunities to piggyback on the embedded skills and major investments that have already been made. This is definitely the direction of travel across all industries: by 2020, 85 percent of CIOs will be piloting AI programs through a combination of buy, build and outsource initiatives2.
In deciding which candidates to strike partnerships with, there are several key questions to ask: should it be an off-the-shelf vendor, or are there internal capabilities that fit the bill? What about investigating academic or local partnerships? More broadly, what is your philosophy on intellectual property? It’s an important consideration. Some vendors will store data locally and have full rights to the IP, whereas others may share it.
If all of this sounds like an exhausting process, you may decide to engage with a partner that can support you in selecting and forging the right partnerships. In some pharma companies, that might mean partnering with R&D, building platforms and capabilities that can cross commercial and R&D business units. Think about what data and insights can be leveraged or merged across both. You may also consider establishing functions that can work on an organization-wide basis and act as a center of excellence for skills and resources.
With oceans of life sciences data now being generated every day, this collaborative approach will be critical in the drive toward a patient-centric pharma model. Such a model will be able to provide vital information and insights across key functions, taking advantage of the huge speed and memory capabilities now being developed for AI applications in the pharma industry.
Now is the time to start mapping out the route. Figure out the right use-cases, create a robust value case and determine the quick wins that will ensure traction. Critically important, recognize that this is a journey, with clear stages to progress through.
It’s vital to decide how fast you want to go, whether you want to be a pioneer or a smart follower in this space and, crucially, what your ultimate destination should be. For more information read our full research report.
1Accenture Industry X.0 Survey, Life sciences respondents, 2017