An overwhelming majority of Hanapin’s State of PPC and Machine Learning survey respondents said that lack of transparency and lack of control were their biggest worries for the role of AI and Machine Learning in advertising. Below are a number of other worries from survey respondents:
Some of the other worries are also correlated to the transparency and control piece, further proving how skeptical we are of machine learning and/or ‘smart’ features.
Personally, I don’t think Google is going to give advertisers more transparency and control when it comes to machine learning.
As we’ve seen in the past, the algorithms tend to improve over time. When the algorithms improve, the performance gets better. When performance is on par with what we were manually producing previously, we adopt more of Google’s automation, despite the lack of transparency and control. That process has worked for Google in the past, so why try anything new?
Where did this strategy come from?
I’ve never worked for Google, so this is purely an educated guess.
I believe Google has 2 main options here:
- Give advertisers more transparency and control with smart features while ALSO improving machine learning to drive better performance from ‘smart’ features.
- Continue giving us broad descriptions of what ‘smart’ features are doing on the back end (requiring no extra work/money) and focus SOLELY on machine learning to improve the algorithms as fast as possible.
Think about how much time it would take to explain to advertisers how the algorithm on the back end of every “smart” feature is working (transparency). My assumption is that it’s MUCH faster, and cheaper, to focus ONLY on machine learning, improve the algorithms, and drive acceptable performance faster than if Google’s focus was split. When the performance reaches the point where it’s on par with what we’re able to achieve manually, there’s no good business reason for advertisers to not adopt that feature. It’s achieving the same or better performance, minus all of the extra work on our end. So we put down our pitchforks and adopt smart bidding, for example. Google wins, but we’re not mad anymore because the ‘smart’ features are working. Transparency and control are now moot points, at least when it comes to smart bidding.
The same goes for giving advertisers more control. We could NEVER process as many data points as a modern-day machine. So why, in the short-term, would Google invest time and money in a dozen levers we can pull to influence bids when in a few months or years the algorithm will get better results than humans can achieve? At that point, we’ll be okay adopting a feature with fewer controls because it’s hitting our performance goals. It just doesn’t make business sense, from a purely financial standpoint, to invest time and money in levers that advertisers can control.
Sure, Google may lose trust or loyalty from hundreds of advertisers by using this strategy. But have we stopped using Google’s advertising platforms as a result? I have yet to meet an advertiser that has stopped using Google Ads in order to make a point about transparency. And even if they exist, their ad spend was probably a drop in the bucket for Google.
All of that being said, I wonder if we’ll reach a tipping point? Will Google introduce a feature that’s lacking so much transparency that the majority of advertisers refuse to use it, at all?
Do you think something like this has already been introduced?
Food for thought!