Think about it: How often have you looked up your symptoms on Google, identified possible reasons for your ailment, and researched treatment options for specific conditions?
Most of us do it. If not for ourselves, for someone we know. Data shows that 66% of internet users have looked for information about their medical conditions online and 55% of them have researched treatment options 1
When it comes to purchasing medications, 43% of them leverage online information to inform buying decisions.2 They often rely on reviews and recommendations to better understand the product's effectiveness, complications, and possible alternatives.
As more patients turn to these review platforms to share or understand treatment experiences, platforms like Ask a Patient, OTC online stores, disease-specific communities, and patient discussion forums automatically become one of the most comprehensive sources of data - offering pharmaceutical organizations gold nuggets of information that can tell them what is working, what is not, and how their products compare with those of their competitors.
In addition, pharmaceutical organizations can identify trends and patterns in customer opinions over time and use that information to improve their products and grow their market presence.
For example, the following review tells an eye solution manufacturer that they need to invest in better packaging, ergonomic design, and provide better discounts.
Figure 1: An example of a customer review on an eye solution product
Similarly, reviews on a psoriasis prescription medication may help a drugmaker understand the reasons behind poor sales performance, such as severe side effects, usability challenges, and more.
Figure 2: An example of a customer review on a psoriasis cream
Customer data like this is valuable (and almost effortless to gather). Here are some estimated outcomes that pharmaceutical organizations can expect with an effective review analysis strategy:
Figure 3: Estimated outcomes from review analysis
Primary market research, such as Awareness Trial and Usage (ATU) studies, has been crucial for life sciences organizations for a long time. They track how customers become aware of a product when they start using it, their experience, and if their usage continues - all critical metrics to monitor and improve brand performance over time.
However, while PMR can yield rich data, organizations typically need 6 to 12 months to carry out the extensive process of planning and coordinating personal interviews with hundreds of customers.
Online reviews can offer preliminary insights in the interim, helping organizations keep track of the daily market sentiment while their PMR is underway. These insights can be leveraged to understand customer experience and challenges on the go. Additionally, they can help organizations deep dive into specific pain points and challenges during their one-on-one PMR interviews.
Patient review analysis on its own can be agonizing with a manual process in place. When analysing reviews, you'll want to consider questions like:
What were sentiments like?
Which keywords were most common?
How have review trends changed over time?
What are the most liked and disliked features?
Accurately answering these questions through manual analysis can be extremely timeconsuming. Fortunately, new automated technology, like Natural Language Processing (NLP), can speed up the analysis, easing resource bandwidth and more accurately generating review insights.
NLP models can convert unstructured text into a structured format, enabling you to recognize sentiment in customer conversations by identifying language patterns that reflect their opinions and expectations about a certain treatment.
Figure 4: A workflow of Natural Language Processing
Machine learning models are trained to mine data and classify text by polarity of customer opinion (positive, negative, neutral, and everywhere in between). This is called sentiment analysis, and it helps pharmaceutical organizations understand how customers are feeling about their brand or if there's a change in brand sentiment over time.
For example, a skincare brand is selling an acne liquid solution in 2.5 ml, 5 ml, and 10 ml bottles. An analysis of customer reviews online showed that while customers found the formulation most effective than competing brands in the market, they were more inclined to purchase the 10 ml bottles owing to cost efficiencies. In addition, the reviews also indicated that the 2.5 ml bottles are not likely to see a rise in sales because a majority of the customers felt that the price was too expensive for the quantity provided. However, overall, positive reviews outweighed the negatives - allowing the company to work on better packaging and pricing, without having to worry about its key formulation for now.
Figure 5: An example of customer sentiment analysis
Figure 6: A deep-dive into customer opinions across negative, positive and neutral categories
In addition, trained machine learning algorithms can also be applied to a dataset of product reviews in an effort to analyze customer star ratings on the platform and understand what the majority of the sentiment points to.
Figure 7: An example of customer star rating analysis
At its heart, review analysis is a customer-centric activity - enabling you to make product improvement decisions based on what your customers value. NLP can be very useful when it comes to understanding customer feedback in real-time. By enabling faster access to valuable customer insights, NLP can help life sciences organizations: