Personalizing Non-Personal Promotion With AI and Third-Party Data
Studies and feedback show that HCPs want pharma to tailor its non-personal promotion to meet their individual needs. HCPs want pharma companies to pre-digest clinical developments and for the information to be communicated to them through a preferred channel. But there is a lot to unpack when personalizing non-personal promotion.
The primary obstacle to personalizing non-personal promotion is that the data that is needed by pharma marketers is not the same data that is needed by e-commerce vendors, news outlets, or companies like Netflix. Marketers at such companies have had an easier time implementing personalization because audience characteristics like media consumption habits, hobbies, and interest categories are neatly packaged and standard in most analytics platforms. Those marketers have access to information that is highly relevant to their targets buying preferences.
But Google Analytics can’t tell you if a site visitor is a family practice doctor or a pediatrician.
Recent Successes in Personalizing Non-Personal Promotion
In a recent campaign, GSK augmented the digital marketing of its Shingrix product by using real-time third-party cookies to create personalized web experiences for different HCP audiences. The award-winning campaign had a great re-engagement rate with its HCP audience.
The GSK Shingrix campaign (which focused primarily on email) generated a 66% rate of return to the website when using DMD to act on digital next best action data. Shingrix sales greatly outpaced their already-high forecasts during the campaign year, and are projected to nearly double their original totals only two years after launch.
After talking with a few AI vendors and third-party HCP data vendors, I think there is an emerging way of personalizing non-personal promotion in real time. One can combine AI content personalization hubs with third-party data.
The First Half: Real-Time Content Personalization With AI
AI tools like Acoustic feed cookie and website interaction data into artificial intelligence brains. These brains build models that are most likely to lead to favorable outcomes. The brain can then use these models to personalize the web page that each user sees.
For example, a user who visits a website on a mobile device might see a home page that features video content. Or, if a user engages with several pages about hiking products, then the content hub might populate the home page with new hiking products the next time that user visits the site.
E-commerce and B2C marketers can use these platforms in turn-key fashion because their audiences can differentiate themselves by product category interaction. If they can’t, there is usually standard information contained in cookies that will be a useful starting point.
For example, someone who has searched for “hiking” or who has interacted with several pages about hiking is probably a user who should be shown hiking content. More importantly, “outdoor enthusiast” is a standard interest category for most analytics platforms. The hub would be able to show both users personalized content within the first few page views.
As I mentioned earlier, the problem with personalizing non-personal promotion and content in real time is that there are important HCP traits that aren’t captured by standard cookies.
An e-commerce platform gets a lot of value from knowing a person’s age, interest categories, and other factors that are captured by cookies. These factors can be captured in real time and then met with personalized content that others with the same traits have responded well to.
Standard session data doesn’t include things like an HCP’s specialty, their practice setting, and their prescribing history. We need relevant real-time third-party data. And there are now vendors who offer it.
The Second Half: Investing in Real Data
DMD’s Audience Identity Manager is a unique service that offers the ability to recognize website visitors by their name, NPI number, specialty, and geographic location. This is made possible by the company’s publication network of over 800 sites, which provide free access to clinical developments and news to HCPs in exchange for placing a cookie on their browsers.
NPI-level Information about the HCP is stored on the cookie. DMD’s clients can place a piece of code on their websites that allow them access to this information stored within it. DMD has placed these cookies on the browsers of hundreds of thousands of HCPs across all specialties. There’s a good chance that tagged HCPs will show up to your site in large numbers.
Then your AI hub can get to work personalizing experiences based on this data. As long as you have enough content, the AI can serve tailored pages according to specialty, IMS data, and other factors.
Neither the folks at Acoustic nor those at DMD have given me any reason why the platforms can’t be connected and share data in real time.
In such a scenario, users who have been tagged by DMD can be sent to specialty-specific subdomains that are populated by content that others of their specialty have engaged positively with. For returning users, previously viewed content can be replaced with promotional materials or other news that they haven’t viewed yet.
As long as you produce enough content, they can keep returning to your website for news. If they come back often enough, they will start seeing more promotional materials.
Personalized non-personal promotion on the first interaction
After enough sessions and real-world data have been fed through the content hub, the hub’s AI could start to make models about how to engage each specialty. Customizing an email campaign for endocrinologists could be as easy as sending them to the endocrinologist subdomain. The engagements that happen from that point would be based on the hub’s past experience with similar users.
It would generate substantial ROI
According to 2019 research by The Relevancy Group, advanced personalization can produce ROI of 20:1 in the retail industry
While pharma is not retail, it has seen benefits in personalizing experiences through next best action marketing. Brands have experienced a 2-4% increase in sales when implementing next best action strategies. Automating the digital aspect of next best action marketing could produce an even greater return.
Whereas the Shingrix campaign required daily analysis, this strategy would reproduce that analysis with AI. The amount of time that would be saved would further contribute to ROI.
Audience Identity Manager works by reading cookies that contain NPI and other information in real time. This information is contained within AIM. To engineer a situation in which AIM can take an NPI number and then classify it as an HCP category would require pharma web developers to convert AIM information into cookies that could be imported into Acoustic. This could be extremely difficult and labor-intensive
As the bright data scientists at Aktana Inc. have mentioned, recommending an item to purchase and recommending useful information to a healthcare provider differ greatly in complexity. It’s not really clear how effective AI would be at personalizing non-personal promotion in real time.
It would slow down page load times
Scripts like the AIM reader are usually the biggest culprits when a site is slow. It will take time for the AIM script to convert data into a format that is usable by Acoustic, and then for Acoustic to show custom content to that user. Page load times are essential to healthy user experiences, conversion rates, and search engine rankings, and this strategy could affect that.
Multi-Channel & Personalized Non-Personal Promotion Is the Way Forward
HCPs have let us marketers know that they want us to help them stay up to date with their professional education requirements. A big part of this is that they are seeing more patients than ever and don’t have time to stay current. It’s our job to serve them the content they need in a multi-channel format.
And that’s the last bonus of this strategy. Because DMD’s cookies are placed on HCPs mobile devices and their desktops, the data you get will be a goldmine of insights on the different needs of users. You might see that mobile devices are mostly daytime visitors looking for dosing information. You could tweak the ad scheduling of your Google Ads campaign to capture more users during the day because of this.
Or, you might find that desktop users visit professional education sites in order to research your brand following their visit to your brand’s site. Because DMD also tracks their activity many of these sites, you would be able to identify which publications are most valued by your audience. You could then add the print version of these sites to your media plan.
As long as your website is mobile-responsive, contains a variety of content media formats, and is routinely mined for marketing insights, it should be easy to use this model when serving personalized content in a multi-channel NPP strategy.