• by Cory Treffiletti , Featured Contributor, 3 hours ago

One of the interesting topics around the use of AI in marketing is synthetic data.    I’m not completely sure I’m comfortable with it, though.

Synthetic data is exactly what it sounds like: artificially generated data intended to accurately represent the ways real consumers interact with media and brands. But the fact is, it’s created using AI, and AI has inherent biases that cannot be accounted for.  Synthetic data is, at its absolute best, a highly educated guess.

AI, however incredible, is still rule-based and relatively linear in its ability to “think,”  unlike humans.   We are non-linear thinkers, and we have the inherent ability to go beyond the data and hypothesize things that may not seem logical. Thus, consumers are not always logical.  If you agree with that statement, then synthetic data is a good foundation from which to make an educated guess about the patterns of consumer behavior, but it is still a tool from which you must make your own final decisions. 

So, the question still remains whether synthetic data is a good tool for filling in the gaps from consumer opt-outs.

That leaves us in a predicament: Depending on the source, as much as 60% of consumers opt out of mobile tracking.  There was an existential threat of cookies going away, but that is no longer the case.  Some data suggests consumers are getting more comfortable with tracking and are simply opting back in, accepting that tracking is part of the industry.  So, the other question remains, do we even need to worry about synthetic data, or can we forecast a world where opt-in rates increase, and the industry does just fine?

This is where I end up.  The concept of synthetic data is, in my eyes, a stopgap.  It is a good trick the industry can use for predictive analytics to be more reliable, but it's still just predictive.  Good marketers need to build in leading, as well as lagging, indicators to help them understand a campaign’s performance -- and no forecast will ever be 100% reliable. 

I think the tools for synthetic data are good for now, and will likely carry us to a state where the data is reliable enough on its own.  I do predict we will get back to a place where opt-outs may drop down again to the 25%-30% range as consumers become more comfortable and, essentially, lazier.  There’s a point where they simply stop opting out because it’s a pain.  After all, consumers still share many things about their lives on social media, and they are fine with speaking to AI and having their phones listening in on conversations, so why would this be any different?

The fact is, society is OK with tracking if it gets what it wants: easier access to whatever makes it happy.  For better or for worse, this is how the world of data works.