Most pharma companies have invested something, if not a significant amount, in some kind of artificial intelligence (AI) initiative. However, not all companies are getting the returns they anticipated. A report from MIT Sloan Management Review of over 2500 executives found that 7 out of 10 had minimal or no gains so far from these initiatives.
This got me thinking about whether pharma are seeing similar returns to the companies in the survey and if so, why, and what can be done about it to improve your chances of success. Here are some of the findings based on projects that fail and those that succeed.
1. It's All About the Strategy
AI is a tech tool, but it should not be seen as a solely tech domain. In so many companies the IT department is in charge of creating the AI strategy. IT professionals are brilliant at tech and creating systems, but they are not strategists. Tech implementations without a strategic underpinning can be an enormous waste of money and resources.
One pharma company I know has spent around $500 million (and is still going) creating a massive global data lake of all their past data. Sadly, much of that data will not be relevant or ever used again. The cost was huge, and although the infrastructure will no doubt be useful moving forward, the time spent sorting, cleaning and ingesting all that data was largely a waste of money.
Everything should be underpinned with a strategy first. What is the business need and how can the tech meet this need? How can the results be measured?
Every project we do begins with a strategic planning roadmap. This piece of work examines the company or brand challenges and objectives from a consulting perspective and identifies whether AI could confer some advantages. If it can we then examine which of the many AI project options offers the best cost/benefit ratio that should be implemented.
Then we plan the roadmap for that project. In some cases, we go on to do the project. In other cases, the client internal team take over for implementation. And in other cases it gets outsourced to a cheaper offshore solution provider. However, when pharma companies miss or skip the first steps, the resulting projects can be square pegs for round holes. They may be executed brilliantly from a technical perspective, but they aren’t designed to meet an inherent strategic need of the business. They won’t provide the anticipated returns.
2. Balance Tech and Business Needs Appropriately
We find the projects that succeed put the largest amount of focus in the beginning on the business case. This needs to be planned and analyzed before you even consider the tech. Unfortunately, many companies focus just on the tech. For example, some prioritize getting a specific AI tech into their company. And by the time they implement it, the tech is out of date.
I had a meeting recently with a large pharma company that was busy getting a particular software framework for big data processing into their company. However, everyone in AI has already moved on from that framework to superior software with less limitations. Focus on the business and get the AI and tech to support that rather than focusing on a software or tech implementation a talented sales person persuaded you to purchase.
3. Don't Focus Solely on Cost Reduction
Obviously cost cutting and enhancing productivity is a great goal for an AI project. When implemented well provide some quick wins and measurable returns. However, revenue growth should be a key focus of companies.
The MIT Sloan Management survey I referenced above showed the companies that were effective at gaining value from their AI initiatives were twice as likely to focus on revenue growth than cost reduction.
AI can provide powerful ways to rapidly grow revenue (including in pharma). Utilizing it for precision sales and marketing alongside precision patient diagnosis for specific conditions can land patients who may be picked up in other ways.
4. Do Strategically Planned Projects Even if Higher Risk
Both the MIT and my own pharma survey found companies that take on large, well planned, AI projects receive more value despite the higher risk. Half (50% ) of these type of projects already generate positive value compared with only 23% of small value items.
I believe in the risky bigger well planned big ticket projects as we have done many ourselves successfully. However, the way we approach it is to plan the big picture of the long term vision. Then we break it down into smaller, quick win projects that can ultimately be connected together to achieve the larger goal.
Often the revenue gains from the smaller projects can fund the big vision development. Then eventually all the quick win projects get connected together into a large all-encompassing system. Which brings me to the next point…
5. Focus on the Long Term Transformation
The value of AI is not simply the tech, but how it brings together so many aspects of a company process: the business strategy, the data, the customer needs experience, the sales and marketing channels, all the digital transformations (sales and marketing and other), as well as the AI algorithms and the tech. Create your 3 or 5 year transformation big picture and then work back to the sub-projects that can be tied together to make that happen.
6. Invest in the Right Talent, Team and Processes
One issue is talent. There are experienced data scientists out there who are great at their jobs, but finding them can be challenging. Many companies (including my own but also companies such as Google) ensure their data scientists do not have LinkedIn profiles to prevent other companies from poaching them. As a result, most companies are hiring junior data scientists from who don’t have the experience to tackle the big challenges.
You need these junior data scientists as they will become experienced in time, but you also need guidance from the top data scientists who have been doing AI for over 15 years and have found ways to solve most problems.
Investing in the right data is another critical aspect. Data is expensive. To avoid costly mistakes, put off buying data until you have your business case plan in place and know exactly which data will suit your needs the most.
Ensure your business processes also work. One client started a 6-month project with us with a set delivery date, even though one big data source was not available. The plan was to add it when it became available. It never was, and the client spent hundreds of thousands of dollars on a big data that never made it into the project.
If a data source is expensive and you really want to use it, ensure it is available with a data map at the start of the project.
The report and my own research with executives in pharma found that these areas above were the key aspects that differentiated a successful AI initiative from a failure. The bottom line is to plan the business case, create a long term vision, plan sequential quick win smaller projects that combine to create the big vision, collect the right data, have the right talent to work on it and remember that the AI and tech is only one part of the process to achieve a AI success story with your initiatives.
For more information on any of the topics contained here, or assistance in planning or implementing any of these topics, please contact the author, Dr Andree Bates, at Eularis http://www.eularis.com/contact or sign up for our one day 2020 masterclass (date to be determined) on Innovating with AI in Pharma Sales and Marketing in London, UK.