How pharma-backed websites are easing doctors’ search woes through AI

Getting healthcare professionals to visit, spend time, and engage with a brand’s HCP website is essential to any pharma marketer’s digital strategy. Yet, the experience doctors encounter on these sites has fallen far short of those goals.

That’s despite all the fanfare around employing artificial intelligence (AI) to help personalise content.

“For a very long time, there was the tease of what looked like it was going to be personalised content—that little widget on the bottom right-hand corner of your screen that says, ‘I’m here to help,’” recalled Robert Schildt, VP of go-to-market and omnichannel strategy for NewAmsterdam Pharma.

“You’d go there, and it would offer you the same content you’re already looking at,” he added. “It was the illusion of tailoring.” 

Many pharma-owned websites still force clinicians to sift through a lot of what they don’t want to find what they do. As a result, according to Schildt’s estimate, most brand-HCP.com sites garner just a small fraction of the traffic seen by their patient- and consumer-oriented counterparts.

But that situation is poised to change, he and other marketers say, thanks to third-party tools that can discern clinicians’ preferences before they arrive on the site and then serve sought-after content more directly to them. 

The technology derives this ability from “understanding the data of a doctor — what their prescribing behaviour is, what their search history is—and then personalising the content that actually meets their specific needs,” explained Arpa Garay, who formerly held executive-level commercial roles at Moderna and Merck. 

“In the past, a lot of pharma marketing has been mass marketing,” said Garay. “Now, everybody’s trying to figure out how you personalise content, both to HCPs and consumers, but the level of success is pretty varied.”

Indeed, digital engagement is a bit of a Catch-22. Relevant content is often necessary to get someone to complete an action online. But without that engagement to understand preferences, it’s hard to know how to make the content more relevant.

Two solutions

A few technology companies are trying to solve this quandary by tapping into the doctor’s web activity. Consider Ostro Health’s tool, which appears on a brand’s website as a boxed listing of content categories. If a doctor arrives on the site after clicking on an ad, for instance, the tool arranges the content categories based on the parameters embedded in the ad.

“If I click on an ad around clinical trials, don’t put ‘Getting Started’ at the top [of the website],” urged Chase Feiger, M.D., Ostro’s co-founder, referring to a common issue seen by physicians. “Don’t send me to a one-size-fits-all brand dot-com. Show me personalised one. You already know what I want to click on, so just put it at the top.” 

Known as “Tailor,” the tool distinguishes between prescribers and non-prescribers (their content needs usually differ). Ostro also uses AI to serve up “next best content dynamically,” Feiger said, in a similar fashion to the recommendation engine on Netflix, as well as other doctors’ most-requested items. 

Its AI-powered search and conversational AI capabilities also understand misspellings and medical abbreviations— “If I type in, ‘What is the EFX?’ it knows EFX is efficacy,” he noted — and can handle requests in other languages. 

Swoop’s “Agents” tool, which appears as a chat framework, also focuses on the clinician’s web activity. If, before landing on the pharma site, the clinician has clicked on a programmatic ad focused around a new indication or clinical data, Swoop can “pre-launch” its tool with an already established topic, said Eric Redline, Swoop's EVP of product technology. 

A spokesperson said, “It can offer suggestions on what content might be relevant based on the current user input. " The tool is currently only available in English and is focused on the U.S. market.

Redline said marketers are also doing a better job of leveraging data to pinpoint where clinicians are in their prescribing journey, then adjusting the flow and tenor of the conversation to guide them along the path to the right information. 

“That’s the other big change,” he said, “using more sophisticated logic to identify the right timing and messaging for that HCP to say, ‘Okay, they have this patient in front of them. They’ve ordered this diagnostic test, put in this ICD-10 code; they want this information about these certain treatment opportunities.’ Within a data ecosystem, you can surface that right content to then pull them through to the right place on these conversational agents.”

In both Swoop’s and Ostro’s cases, the content that surfaced is only PRC-/MLR-compliant. Companies insist they also offer more than the mere illusion of relevance. 

A pre- versus post-implementation study of Tailor involving an oncology brand shows it drove a material improvement in bounce rates, a meaningful uptick in high-value actions like “request a rep,” and an increase in average engagement time, Ostro said, adding that there was no change in media spend for the associated brand during the study period.

According to the firm's data, Swoop’s HCP Agent was associated with a 6% abandonment rate, a 96% rate of “inquiries resolved”, and a satisfaction rate of 4.7 out of 5. 

Both tools can be implemented relatively quickly. Agents’ average implementation time is around 12 weeks, according to Swoop, while Ostro’s averages less than eight weeks.

A peak under the AI hood

Agents was originally developed by conversationHealth, which was bought by Real Chemistry in 2022 and then brought under the Swoop umbrella. 

While built on a chat framework, the underlying technology wasn’t automated. A medical staff had to manually build the “intent layer,” an industry-oriented framing of user actions, like dosing and administration, and a taxonomy that included all variations and synonyms for those actions, Redline said. 

Swoop has since pivoted to what he described as an LLM (think ChatGPT, Claude or Gemini) that’s been fine-tuned using the above taxonomy for more accuracy to understand, for instance, the difference between a member of the general public and a doctor asking about side effects, or when a patient may need to report an adverse event (both patient and HCP versions are available). 

GenAI “brings a lot more contextual relevancy to the table to address more complex questions,” Redline noted, adding that Swoop is seeing “more willingness to engage with a chat framework today than we were even a couple years ago.”

Ostro, which was founded in 2019, is careful to distance Tailor from merely a static chatbot feature. Feiger said its patent-pending technologies actually leverage AI to decode what the doctor is trying to find, based only on what’s available and approved by compliance, and to detect user persona, such as an oncologist versus an office administrator, while adjusting its follow-up content on the fly — i.e., the next best resource.

“It doesn’t say, ‘Sorry, I don’t understand; please try again,’” Feiger said, “because it’s understanding within the context.”

The tool has seen adoption across multiple life science brands, including on the patient side, and across many disease categories, from oncology to rare disease, he added.

And therein lies the potential to simplify a complex process. The typical HCP or consumer website houses clinical trial, safety, dosing and savings information; copay cards and other patient or HCP resources; along with the ability to request samples or visits from representatives. There may be 10 tabs running along the top navigation bar, each of which requires multiple clicks to access reams worth of content.

Even if they incorporate cutting-edge navigation, today’s pharma brand or medical information websites often still rely on traditional search. That makes it challenging for a user to find the right content.

“The amount of information on [those sites] is staggering,” said Schildt. “The idea is to try and figure out ways where you can, within the flow of navigating a website, offer people shortcuts.”

Forward movement

Stacy Stone, executive director of omnichannel at Pacira BioSciences, said she’s also excited for what the newer tools might mean for enabling HCPs to connect with the right information in a less challenging way. 

As the industry adopts an omnichannel marketing mindset, “it’s about offering a really high level of customer service,” she explained. “And if that’s your objective, most websites today, which should be the hub of your omnichannel ecosystem, are not often very easy to navigate. They’re refreshed maybe once every other year, depending on the maturity of the brand.” 

HCPs and patients, like other consumers, “may get frustrated to the point where they just drop off” if they can’t find relevant content quickly, Stone added. “Being able to find content on a branded or unbranded site immediately that’s relevant to what you’re asking is critical to moving the process forward.”

As such, the newer tools are designed to mimic the intuitive user experience (UX) that has taken root in nearly every aspect of digital life, from banking and retail to entertainment and travel.

“I have hope that we can at least get closer to that with some of these new technologies,” Stone said. 

While this level of customer service has proven elusive for pharma brands in the past, given the technological advances combined with a strong undercurrent of consumer expectations almost everywhere, conditions could be ripe for change.

“Whether you’re a doctor or a patient, you’ve come to experience and expect a certain level of ease at which you are searching for, absorbing and consuming content,” Schildt observed. “It would be very short-sighted to think that in this one industry, you don’t have to deliver that, because if that’s the case, they will just stop looking at branded websites.”

On the other hand, the quicker pharma can accelerate to what other platforms are doing in terms of removing friction from the content search, the sooner HCPs can become more comfortable with the treatments available to them and their patients, and the sooner branded and unbranded pharma sites may see an increase in traffic levels.

Thus, while the thought of AI-powered content search may still conjure images of that “widget in the corner,” pharma marketers believe that this time, the tech companies may be on to something by promising an improved digital UX for clinicians.

Schildt said he “wouldn’t be surprised” if, five to 10 years from now, every pharma brand with a significant HCP presence is offering more relevant content to users based on who they are. 

“Otherwise,” he said, “you’re going back to that same approach, which is like, ‘Here’s everything, hope you can find what you’re looking for.’”