Pharmaceuticals & Biotech

How gen AI is rewriting the script in pharma marketing

May 30, 2024 | Article | 9-minute read

How gen AI is rewriting the script in pharma marketing

“Most of all, I think generative AI can completely change the way we create content and campaigns … but it is a question of adoption, of process, of operating model and not of technology. It will rock our world … only if we let it.”

—A participant in a ZS roundtable of chief marketing officers


B2C leaders outside of the pharma industry began their digital transformations approximately 10 years ago and already have realized the value of this evolution. For example, Nike has seen significant growth from a hyper-focused customer-first strategy, anchored in digital capabilities and personalized engagement. But many pharma CMOs and marketing excellence leaders are still reeling from prior digital transformations. Many pharma marketers have experienced what we call “omnichannel drift,” where the gap between the promise of digital transformation and reality continues to widen. This drift prompted many marketers to hesitate to explore solutions with generative AI. Many CMOs lamented, “Omnichannel showed up everywhere, except in our profits … will gen AI be the same?”

Despite the doubts, the promise is still significant, with potential to improve quality, productivity, speed and cost ultimately saving biopharma organizations large sums to reinvest in serving more patients and bettering outcomes. And increased customer expectations for personalization and engagement compel the industry to move forward.

As AI sprints, pharma marketing lags behind

ZS surveyed nearly 60 pharma marketers and held roundtables with 15 CMOs and marketing leaders to understand potential, progress and barriers.

When surveyed, almost no line marketers and only 10% to 15% of executives said they had a high degree of comfort with gen AI tools like ChatGPT, Microsoft Copilot and others. In some cases, organizations at the forefront have given their executive teams enterprise tools to experiment with, but scaling has been limited.

Some assume that gen AI will not affect pharma due to regulatory and compliance concerns. But even if the immediate disruption is less than what the consumer packaged goods industry is experiencing, one CMO said, “40% of what we do will be affected by gen AI within the year and more beyond that. Even though pharma will be a little behind, we have already started the journey.” There was broad agreement in roundtables that there are ways to use these tools, and that partnership with legal, regulatory and compliance colleagues is needed for success.

Hope beyond the hype: Marketers see use cases

Identifying the right use cases for gen AI is the first part of the challenge. Our respondents defined 13 unique use cases that are most applicable in marketing in pharma, as detailed in Figure 1.

We asked marketers to rate each of these use cases on their priority versus feasibility (see Figure 2) and have prioritized some key use cases.

Use case 1: Quick, efficient insights generation

In our research, brand marketers and CMOs continually lamented the mountain of research, quarterly business reviews, dashboards and presentations that leave marketers scratching their heads to find the critical insights that should shift strategy. “I joined the brand as the lead on day one and asked for the core insights that informed brand strategy,” one marketer said. “I was sent over 300 slides and links to four different dashboards. When we do a situational assessment, it usually takes over a month. We can do better.”

Some organizations are investing in insight agents, where Q&A interfaces are being developed to allow insight leads and marketers to ask questions of the data. From dormant market research PowerPoint files to various tableau dashboards to even processed secondary data, marketing and analytics organizations are hoping insights can be mined quickly to create more efficiency. A few organizations are using gen AI techniques to gather new insights from call center transcripts (patient and provider) to inform more real-time shifts to strategy, which is used consistently in other industries.

Imagine if: Marketers contemplated the idea of asking an AI copilot if access perception for a given brand has shifted and getting an immediate text response that could identify a specific improvement in overall perception due to a recent payer win, but also identifying a PCP subsegment that still has poor perceptions. They envisioned a tool that could query surveys, HCP call center transcripts and qualitative research transcripts at the click of a button.

Use case 2: Easing the path of first-mile content

One area where marketers were less comfortable (but CMOs were significantly excited about) was “first-mile content” or the process of generating “concepts” for testing before a full campaign is built. Typically, a company gives a creative brief to its agency; this brief is typically a multi-week process with reviews and revisions. Next in the process is a four- to eight-week process of concepts and ideation, primarily owned by the agency. In the first use case, gen AI can be used to write the creative brief, cutting down on the amount of back and forth and helping both inexperienced and experienced marketers.

Beyond the brief, in some cases, these concepts are turned into sketches or even images based on initial ideas and then tested in research. ZS found that a few marketers have been experimenting with image generation platforms (e.g., Midjourney, DALL·E, Firefly) to generate innovative ideas for campaigns with careful prompting. However, this is not a skill set that pharma marketers have today, and it requires careful upskilling on ideation, usability of large language models and prompt engineering on various platforms.

Imagine if: CMOs wondered if first-mile content could be part of the creative brief or even if the skill set could be somehow insourced to speed up the content development process, saving up to $5M per organization. Freeing up this $5M means a reinvestment to help improve patient outcomes. 

Use case 3: Tweaking last-mile content

The highest-rated use case by marketer feasibility was the process after a campaign has been developed, photo shoots have occurred and creative assets finalized—tweaks to creative. “We want inclusive and global campaigns, but we cannot capture every variable,” one global marketer said. “In some cases, the South Korea team wants an ethnically appropriate model, but that was not part of our core content. It is not just global, but many populations in the U.S. are not always represented in our campaigns. I am curious if this can be solved.”

A few organizations are piloting loading their core campaign asset into image generation platforms and prompting slight tweaks to the model or even the setting, potentially saving at least $20M to $40M. In some cases, the setting may not feel culturally appropriate and other shifts need to be made. Others are experimenting with shifting messaging or creative to be segment-specific for greater personalization. CMOs also hypothesized training would be needed on best practices to uphold U.S. copyright laws and educate marketers on what not to do (e.g., swap the model for someone who “looks like” a given celebrity).

Imagine if: Global marketers agreed this capability could fit nicely in the increasingly important in-house creative services being built in large pharma organizations. “Agencies think this is their value to capture, but we need to make it ours,” one CMO said. This opportunity has big bottom-line implications; finance, procurement and senior executives need to pay attention.

Use case 4: Easing personalization and medical-legal review

Most mid-level marketers were extremely skeptical that gen AI could help them with their medical-legal review (MLR) processes, which they describe as “cumbersome,” “convoluted” and “the hardest part of my day.” CMOs, however, agreed this should be the “top priority for marketing organizations” as it will have significant implications for personalization.

Many pharma organizations struggle with true personalization because there is a narrow funnel for regulatory approval that requires between 20 to 50 days for each promotional asset, including two to three reviews, based on a ZS analysis. Part of the issue is the rigid process; part is junior marketers needing to go through it several times to learn what not to do; and part is a potential bolus of content as personalization and new iterations increase MLR workload by 200% to 500% compared to three years ago.

Only a handful of organizations have started using gen AI to highlight messaging that has already been approved versus net new, to auto-populate references, to auto-check for minor label updates and even provide “similarity” or “risk” scores to help identify content that may require a more thorough review than simpler, more derivative content. Our research found that a few companies are exploring AI to provide “first draft feedback,” which is especially helpful for junior marketers, effectively learning from prior regulatory and medical comments and providing automated redlining of messaging and images. These applications have potential to improve speed to market by 50% and increase content delivery volume by 25% to 40%.

Imagine if: One CMO envisioned that “AI could be ‘the lawyer’ for low-risk, derivative content ... you change a few headlines, colors, fonts, visuals, nothing too risky … it’s possible and would make the velocity of content incredible and on par with more consumer-driven industries, but in healthcare a human may always need to be in the loop.”

Three steps for pharma marketers to begin the AI journey

Regardless of the use case, marketers agreed—these technologies are changing the world and have the power to disrupt the pharma marketing process, especially for content. The other consensus was that all organizations were overlooking the need for change management with marketers, insights colleagues, agencies and even senior leaders. There are three actions a marketing executive should take to begin the journey:

  1. Prioritize and pilot use cases: To move from tech-driven to business-driven use cases, marketing leaders should drive use case prioritization for their function and have a heavy hand in experimenting with specific marketing teams that will become “champions.” We have defined clear value measures to enable scaling, which many leaders overlook when piloting.
  2. Disrupt and define process: New AI tools will not just automate parts of the process. Transformation will require reimagining, not just with marketers, but with external agencies, in-house agency services, regulatory and even insights colleagues. We have been partnering closely with marketing executives to define the operating model and incentivize the right behaviors.
  3. Upskill and reskill talent: First-mile content, for example, is typically owned by an agency after a creative brief is written, but marketers may now need to learn how to prompt for innovation and understand copyright best practices. We are learning this is an apprenticed skill, and adoption must be driven by levers beyond training modules.

The marketing workforce of the future will have a higher velocity of content and insights that have the potential for not just efficiency gains, but real effectiveness improvements at a brand level. Is your marketing team ready? One thing is clear: AI will not replace marketers. Marketers using AI will replace marketers who do not.

Add insights to your inbox

We’ll send you content you’ll want to read – and put to use.

About the author(s)