What's Better Deterministic or Probabilistic Data?

I had a fun conversation with Copilot this morning around data and the two main types I see/hear in my day to day as a social media product manager. For those that don’t know I work in the pharma/healthcare industry working with DTC and HCP data and those two main data types are deterministic and probabilistic data. Now that we are caught up let’s get into it!

My conversation was understanding the difference between the two data types when it came to marketing and then I narrowed my focus to solely social media marketing and how that could be affected in a positive or negative way by using different data types. It was interesting and made me think of current problems I have been a part of in this industry recently. To note, using one data type over the other is not necessarily bad however it may not achieve your marketing KPIs.

Let’s start with quick definitions:

Deterministic Data – it is data that is not subject to randomness or variability

Probabilistic Data – refers to information that incorporates elements of uncertainty and estimation

I think we know in the world of HCP targeting especially on social media where anyone can be a physician as long, they enter that data within the job title field on the platform. ?It’s important to have deterministic data to make sure you are targeting physicians and not just Joe Schmo pretending to be one. Now I am sure you are thinking but what about scale on social media? I hear you and totally understand your issue here, yes if you are using deterministic data, it might be hard to get scale (not if you use a specific company within my profile but anyway). If you are worried about scale, then using some sort of probabilistic data could be good and it might help with your marketing KPI. It also might not help at all, meaning if you are using interest based targeting you may not hit HCPs at all and therefore waste your media dollars on users that are not converting at any capacity. With deterministic data, you will be able to target those users who are interested in your ad therefore giving you a better engagement rate and better ROI.

Digging deeper in this conversation with copilot, I dusted off the ol’ agency hat and threw that on and looked at how agencies may look at this. What is important to an agency? Making their client happy of course, targeting the right users, staying on budget, and bringing new ideas to the client. Those are the ones I could come up with, I am sure there are more, but it’s been a minute since being in an agency so bear with me here. But if we took those above and looked at the options, we have for social media marketing we come to a fork in the road multiple times on our planning process. Type of data to use? Which platform makes sense? What creative do we use and when? Is it time to try a new social media platform? A lot of questions to answer and all while trying to stay on budget. Yes, the CPMs are higher when using deterministic data than using social media platform data (interest based) but you will be able to more targeted and not only hit the right users but hit them with a more personalized ad! Imagine being able to tease an HCP with a personalized ad based off reading and researching behavior or maybe one step better, off diagnosis/prescription data, wouldn’t that make your life easier? Sounds like using a deterministic data set could check off all the boxes we listed above. What if budget is a problem, how do we combat that? Great question gang! Maybe using a platform where you can add keyword targeting might be helpful, that might get you a more engaged audience at a lower rate. Maybe your client has an email list that can be leveraged and used on social media. Rounding out the omnichannel strategy, we could send an email blast out and then following up with a social media ad and then throw some next best actions in there along with some programmatic campaigns and we are off and running!

Now you all might be one step ahead of me and are thinking this is all great and that’s one hell of a media plan but how can I tell if its working. How do I measure this and what makes sense. Now we all know, or I like to think we do that social media platforms are walled gardens and it almost impossible to get any type of individualized data out of a social media platform. Going back to my original question into Copilot, what are the benefits of deterministic data versus probabilistic in marketing. If we spend so much time deciding which type of data to use to target, wouldn’t we want to spend the same amount of time when it comes to measuring that data? Going back to our lovely industry of pharma and targeting HCPs across social media, wouldn’t it make sense to be deterministic in our measuring capabilities than using probabilistic data? Yes, I understand we want to measure every aspect of a social media campaign like we do with programmatic but if that data isn’t available to us then does it make sense to essentially make it up?

I know as marketers we are going to use any data, we can get our hands on and spin in a way that makes sense to us but also the clients we work for. But this is something we must consider and understand if the data is making sense based on the campaign and does it make sense for our clients. Again, not saying there is right or wrong answer here when measuring social media campaigns, but I do think we would all feel more confident with deterministic data versus probabilistic data.