Machine learning for pharma marketers: The time is now to budget, experiment and adjust, expert says

Doctor on computer
Machine learning can help pharma marketers move from reactive to proactive, says one agency expert. (Getty/Vladdeep)

When does CMI/Compas executive Paul Kallukaran think pharma companies should start looking at using machine learning in their marketing strategies? "Yesterday," he said at a recent digital conference. The good news, however, is that it’s not too late.

Machine learning, a kind of artificial intelligence, is when computers are programmed to “learn” and make decisions without human input. In pharma marketing, machine learning can be used to figure out what media or creative executions are working in near real-time and make adjustments for effectiveness.

Kallukaran, CMI/Compas' executive VP, performance analytics and data science, speaking at Digital Pharma East, pointed to consumer examples of machine learning that are common in many people’s lives, such as Netflix movie recommendations and Amazon purchase suggestions.

Free Daily Newsletter

Like this story? Subscribe to FiercePharma!

Biopharma is a fast-growing world where big ideas come along daily. Our subscribers rely on FiercePharma as their must-read source for the latest news, analysis and data on drugs and the companies that make them. Sign up today to get pharma news and updates delivered to your inbox and read on the go.

RELATED: FCB Health launches next-generation big data agency to better tell pharma brand stories

In the same way those platforms recognize, learn and predict human preferences based on past behavior and other factors, pharma companies can plan and tweak marketing and advertising campaigns. With machine learning identifying patterns and making predictions in real time, each engagement can be different. CMI has used machine learning to determine media budget spends, find the most effective sequence of marketing engagements, study the impact of positive or negative news on a brand and predict drug adoption.

And Kallukaran offered some tips and advice for pharmas looking to adopt similar machine learning strategies.

  • Make sure you know exactly what the problem is. As with any experiment, the more specific the query, the more precise the research and the more effective the result.
  • Determine that it’s a problem you can do something about. If you can’t change a budget or investment in media channels, for instance, don’t bother using machine learning to determine where dollars would be most effectively spent.
  • Experiment, experiment, experiment. It’s a bit like the old saw in real estate about location, but in this case it’s experimentation that is key to unlocking the value of machine learning for your company.
  • Allow enough time to find a solid conclusion, which is at least 6-8 weeks, Kallukaran advised.
  • Reward the process, not the outcome. That is, be prepared for failure but don’t give up. Every fail is an opportunity to learn what not to do the next time. Make sure to adjust expectations so that stakeholders know failure is a possibility.

RELATED: Hey, pharma: Data's more than a 'defense mechanism,' ad executive says

One of the important things machine learning can do is make marketers proactive instead of reactive, Kallukaran said. With so much data available now, pharma should be using it to its best advantage.

“More and more data is being collected all the time today. If you are not using it, then why are you even collecting it?” he said.

Suggested Articles

Traditional generics bigwigs Teva, Mylan, Sandoz, Amneal and Endo have lost out to a group of six competitors in terms of U.S. weekly prescriptions.

Roche’s Tecentriq scored some big wins lately, including a pair of first-in-class approvals. But in other cancers, it's decided to pull the plug.

Biosims haven't worked, experts say. Instead, it's time to for an independent body to set fair biologic prices after the drugs lose exclusivity.