Can ChatGPT and Large-Scale Industry Benchmarks Revolutionize Message Generation?
Artificial intelligence and machine learning (AI/ML) have long been trending topics within the pharma industry. The latest buzz-worthy development in the AI/ML space, ChatGPT, once again has the industry dreaming about a new world of potential applications. Given ChatGPT’s ability to create stories, books, and even poetry in any style or theme, a natural area to explore ChatGPT’s potential within the pharma industry is in branded messaging.
ZoomRx has collected ~1M+ data points on the effectiveness of sales messages over the past decade-plus through Promotional Effectiveness Tracking. We ran a few simple experiments exploring the potential power of ChatGPT and large-scale benchmarking to solve a perennial problem for pharma marketers: generating and optimizing promotional messages.
First, we tested ChatGPT’s ability to generate new messaging for a real-world pharma brand. We chose Elahere for our experiment, a recently approved ADC from Immunogen for the treatment of Ovarian Cancer.
- Creating a message: To start, we simply asked ChatGPT to create a promotional message for the brand.
2. Make the message specific: ChatGPT created a plausible message with little input. We wanted the message to be snappier and more precise, so we provided an upper limit on words.
3. Creating Variations: That’s better! But what good is one message? The marketing team will want some options to choose from.
Wow, this is magic! But how do we pick which message is the best? A custom message testing study is costly and time-consuming - but anything less may be guesswork. What if we could rely on a massive database of previously-tested messages instead? Time to turn to ZoomRx’s Industry Benchmark Database, containing over 1M+ message scores within pharma.
4. Building a Message Score Predictor: Leveraging ZoomRx’s extensive benchmark database, we built a beta version of a message score predictor based on a few simple variables. In this beta version, we’ve incorporated a few key variables into the predictor: namely, the word count in the messages, the type of the sentence structure(Simple, compound, complex), the number of data points present in the messages, nature of the message (efficacy, safety, dosing, etc.,) to help in predicting the effectiveness,
5. Message Score Prediction: We fed the 10 messages generated by ChatGPT into our message prediction tool, and highlighted the top 3. In the real world, these messages could then be efficiently validated with actual customers before launch.
Looking Forward: As the field of AI & Large Language Models (LLM) continues to evolve, it is important for pharmaceutical marketers to take advantage of the new opportunities presented by these tools. Though the experiment described here is only a simplistic proof-of-concept, ZoomRx is a firm believer in the potential of technology to enhance pharma decision-making and patient care.
We are releasing a free beta version of the message prediction tool to the Pharma MR community for experimentation. To provide feedback or to learn more about how ZoomRx can help you leverage the latest technology to enhance your brand’s customer engagement reach out to firstname.lastname@example.org
Try this Beta Version Tool to help craft impactful messages for your brand— that would resonate with the HCPs leading to improved awareness & perception.
Note that we are incorporating feedback and hence the platform might be different from what is depicted here.
About ZoomRx's Promotional Effectiveness Tracking
ZoomRx Promotional Effectiveness Tracking (PET) tracks on Rep effectiveness, Message Performance, and Channel Performance, helping 100+ Pharma brands stay ahead of the competition.
ZoomRx PET offers customized tracking for each brand, ensuring that the results are tailored to their unique needs. With Omnichannel tracking for both their brands and competitors, companies can get a comprehensive view of the market. The 30,000-plus ZoomRx physician panel provides valuable insights and access to benchmarks of 10 million data points give brands the information they need to make informed decisions.