When it comes to identity data in pharmaceutical advertising, less is more.
At least that’s the bet data platform Dstillery and healthcare data analytics company PurpleLab are making.
When Google decided to start sunsetting the cookie, Dstillery knew it needed to shift from its third-party cookie-based model of behavioral targeting.
Dstillery released a solution in May to help pharma advertisers target audiences without third-party cookies. It uses ID-less signals, such as URL, time of day of web visits and location at the DMA (designated market area) level, then uses that data to find patterns in online behavior that are associated with specific health conditions.
“When we started selling cookieless targeting, we found that the cookieless value prop resonated most with healthcare brands because their access to user signals is more immediately constricted,” Michael Beebe, CEO of Dstillery, told AdExchanger. Other brands that don’t handle sensitive user data, meanwhile, can keep targeting with cookies.
Initially, Custom Patient Targeting relied on first-party data sets from pharma brands. But that data turned out to be limited because pharma brands weren’t always willing to share customer data with Dstillery out of privacy concerns, Beebe said.
In September, Dstillery launched a new version of its Custom Patient Targeting tool integrated with PurpleLab, which uses national insurance claims data to map out online behavior associated with certain health diagnosis criteria, or ICD-10 codes.
Breaking down targeting
Custom Patient Targeting uses a mix of contextual and aggregate signals, so it doesn’t actually touch any personal identifiable information (PII).
PurpleLab’s diagnostic data base is compliant with HIPAA because the company uses statistical analysis to identify and remove PII.
The challenge for healthcare advertisers is to find a marketing solution that still enables some level of audience targeting. Pharma brands haven’t been able to scale audience targeting with precision because tying user data to healthcare information is just too risky from a privacy perspective, said Ted Sweetser, director of sales success at PurpleLab.
Instead, Dstillery creates a model using its ID-less audience signals and PurpleLab’s ICD-10 code data to identify programmatic inventory that indexes against an audience with a specific condition.
An advertiser using Custom Patient Targeting can target an audience with, say, Type 2 diabetes using contextual signals overlaid with claims data instead of first-party data at an aggregate level.
Dstillery, for example, downloads aggregate patient data from PurpleLab that matches corresponding ICD-10 codes for Type 2 diabetes. Then, Dstillery uses that data to “seed” an inventory model likely to reach patients with that condition based on their online activity, including what sites they’re visiting, at what time of day and from which DMA, Beebe said.
Advertisers don’t get any individualized data attributes, Sweetser said, but they do get audience profiles and a sense of where to reach them online.
Because of privacy sensitivity, the aim is targeting behavioral signals at scale “without any user tracking or retargeting,” Beebe said. For example, Dstillery distills (sorry, had to) the user data it ingests from web activity by excluding signals from the actual content of a webpage a patient is browsing. Instead, Dstillery uses aggregate purchase intent signals from third-party partners to model where a brand is more likely to reach its audience and meet its KPIs (e.g., brand awareness or purchase intent).
Dstillery’s focus on direct response by using purchase signals also optimizes the relatively limited programmatic ad inventory that can be associated with specific health conditions, Sweetser said.
The new targeting tool has only been up and running for a couple of months, but Beebe said there has been an “inflection point” in the number of clients interested in Custom Patient Targeting after the PurpleLab integration because the company now has access to health-condition-based data.
Specifically, “the feedback we’re getting is improvement on customer acquisition cost,” Beebe said, adding that one client lowered their CPA by 84% compared with contextual-only targeting.
The leftover cookie jar
Dstillery’s still in the process of weaning itself off cookies.
It may not use cookies for audience targeting anymore, but it does still use cookies for measurement.
As long as cookies are still available, Dstillery can use them for closed-loop attribution to measure the success of its ID-less targeting by identifying what proportion of campaign impressions successfully reached a target audience segment with a particular condition.
Using cookies to measure cookieless targeting sounds like a punch line, but hey, these things take time.
Dstillery knows it’ll eventually need to give up cookies for measurement the same way it had to find alternatives to cookie-based targeting.
One option for Dstillery will be media mix modeling, which would require manually conducting surveys to understand the impact of a campaign. But the company has reservations about whether samples will be acceptable after advertisers start getting comfortable with the idea of pharma as a measurable channel, Beebe said.
The other option is to transition to a measurement API within Chrome Privacy Sandbox when Google finally retires the cookie for good.
But because Dstillery’s pharma solution is still in early days, Beebe said, the immediate goal for now is nailing down performance advertising for pharma advertisers, starting with cookieless targeting.
This article has been updated to distinguish contextual targeting from Dstillery’s ID-less solution.
