A Q&A with Lifeblue’s Senior Data Scientist Adam Olson
We sat down with our Senior Data Scientist Adam Olson to get his perspective on the importance of a data strategy, the ethics around data collection, and the demise of the third-party cookie.
So, Adam, you have your PhD, and your MA and BA in political science. Now you’re living the digital agency life. How did that happen?
You know, people always rip on the social sciences, saying, “What are you gonna do with your job?” But what you do get out of it, from a social science standpoint, is that you get a set of tools that help you to think better.
It’s actually interesting that I wound up here because I knew I wanted to be a political science major, but I thought I would wind up doing something in journalism. I fell into agency life because after reading an agency job description, I thought, “This is basically social science in the real world.” You get to work in a lot of different areas, with a lot of different challenges. You get to learn a lot and be exposed to a lot of different methods.
Let’s get into it: Where do you see some of the biggest needs for data and where have you seen the most successful use cases?
I think in the day-to-day data science landscape, the real value is to be able to create insights.
Something we did that was really interesting involved a series of methods called clustering, where you take a data set with as many variables as you want and estimate which observations are closest to each other based on all the other variables. There’s a lot of applications for that in marketing, but the biggest probably is in creating segments.
There’s an art to it. We used something like this to create targeting for digital ads where we created a cluster that we thought would be good for propensity to show interest in the ad. We had one segment that we thought would be interested in the product, a segment that would be kind of interested, and a segment that we thought would be more interested — which, subsequently, had a three- to four-times-higher click-through rate.
That alone shows the value of owned data because we used the client’s owned registration data with addresses to merge in a bunch of census data to get average income in that census area, average age, all sorts of information. We were able to augment owned data with freely available government data to produce this result.
Compared to other ways data is collected (i.e. third-party cookies), this sounds ethical. As a data scientist, how do you balance the ethical concerns of data privacy while still designing systems that provide insights clients need to be successful?
The angle is both to be a responsible steward of the data and also to be honest about the brand’s self-interest. If a brand has a robust and rich data collection infrastructure and then they just pass you junk ads or they spam your emails, users will eventually opt out. There’s sort of a double-edged sword in that you have a responsibility to use the data in a quality way.
For example, I don’t know how much you’ve read about the European Union privacy law, but one of the big storylines is what they’re calling legitimate interest. If you want to collect data from someone, you have to get explicit consent from the consumer — and you also have to offer something of actual value. In the “olden days,” like a couple years ago, the way of getting around that was a “more tailored ad experience” that makes ads more relevant.
But today you’ve got to give them something legitimate — just providing a new experience with the possibility of you selling them something isn’t a better experience. I think a lot of people in our industry would prefer a standardization, and I would guess that there’s going to be some version of this legitimate interest clause going forward.
Considering that third-party data is still being used, what are your thoughts around it?
This point you’ve been alluding to, that you do have to own first- and zero-party data is super important. Third-party data is already, on some level, low quality. It’s only going to get worse going forward, regardless of the steps big vendors take to try to shore it up. So, as this resource that is universally used across digital marketing platforms for ad targeting is being hollowed out, it means that the amount of resources a firm has at their disposal is going down.
At the same time, what can they do? The only thing they really can do to combat this is to start collecting their own data — showing up with the habitual CRM data: collecting email addresses. From there, you can start augmenting that data with census data or you can model it out and send out surveys to learn about the contact. But I would also say, regardless of how good third-party segments ever were, having your own data is always better.
How feasible is the implementation of a zero-party data strategy for nonprofits and businesses? Is this a heavier lift than just relying on third-party cookies and running display ads?
To the company that does not start developing its own data strategy, they’re going to be left behind — they’re increasingly going to be left behind as time unfolds. Depending on what you’re using third-party purchased data for, it might already be pretty bad. For example, four or five months ago, Meta lost a whole bunch of billions in market share. And a lot of people were ascribing it to the Facebook rebrand, but what’s really happening is the effects of Apple making people opt into the geographical tracking.
Any final words?
A lot of the performance marketing people I talk to are ringing the bell, saying buying targeting segments isn’t long for the world. If you have your own data sources, those make the best targeting and can help create the most personalized experiences. There are massive negative costs of not implementing a system and the costs to start collecting email addresses or phone numbers or stuff like that can be relatively low just to get started.
Long story short: Get a data strategy!