Plans & Pricing
Does My Company Subscribe?
Matt WhiteEMEA Managing DirectorQuantcast
When it comes to placing and measuring digital ad impressions, machine learning can play a major role. eMarketer’s Bill Fisher spoke with Matt White, EMEA managing director at Quantcast, about some of the ways machine learning can help, particularly with regard to the recent furor around brand safety.
eMarketer: What experience does Quantcast have in the area of machine learning?
Matt White: Quantcast was a technology company from the get-go in 2006. There were vast amounts of data that humans couldn’t decipher but machines could, and that was [co-founder and CEO] Konrad Feldman’s Ph.D.—it was machine learning at [University College London] UCL. The starting philosophy of what Quantcast does is that there’s so much data—most of it noise—that you need pattern recognition and machine learning to pick out signals and give those signals relevancy and value.
The reason it’s trending is because there is now enough data and technology to make use of pattern recognition. Machine learning has only started to get some kind of cadence in ad tech because there’s a realization that it’s far more optimal than a human interface.
eMarketer: Can machine learning play a role in helping to avoid the brand safety issues we have been hearing so much about recently?
White: Absolutely. You cannot stop the proliferation of dodgy sites being created, but if you set the right rail guards for whitelists and blacklists, it can alleviate the problem.
What you’re finding is that there are impressions being served—a lot of impressions being served by some companies and very few by others—but only one impression needs to be served to create offense, and to give a CMO food for thought as to where his brand’s being displayed. Machine learning can help identify the content on sites, but it’s a continuing challenge, because how long does it take to set up a website? Not long.
eMarketer: Among all this talk about machine learning, is it still important to have humans involved at some point?
White: The finesse of advertising will always require the emotional [involvement] from a human. I don’t think it will ever be as binary as one or the other. What machine learning allows you to do is the heavy lifting of making out the signals from the noise that no human can do anymore, because there’s just too much data.
At Quantcast, we process 40 petabytes of data per day, which is huge. You can’t have somebody looking at all that—it’s just impossible. As an infrastructure, machine learning is there and you have a human to finesse parts of it. That’s probably where you get the sweet spot.
eMarketer: Do you think in the long term that artificial intelligence (AI) might be able to add that finesse?
White: I don’t think it’s so black and white, whereby AI is going to take over everything and we’re all going to be made redundant. I think AI and machine learning will do a lot of the heavy lifting in jobs that nobody wants to do.
You think about maybe 120 years ago when the first Ford [Motor Co.] factory came out. It was all about the production line and people would sit there putting a screw in a wheel, and that’s all they did. That’s gone, but other more interesting jobs came on the back of that. I think that’s what will happen with AI. It will make a lot of jobs that nobody wants to do redundant, but it will create a lot more of the “finesse” jobs.
Shifts in how retailers and consumer packaged goods (CPG) brands think about ecommerce, combined with an accelerating acceptance among consumers for buying food digitally, have boosted online sales of groceries. Retailers and brands are taking note of these changing consumer behaviors and offering more digital options for grocery shopping and delivery, which will continue to drive the trend upward in 2017 and beyond.
Not a PRO subscriber? Find out how to become one.
You've never experienced research like this.
Nearly all Fortune 500 companies rely on us.
Inquire about corporate subscriptions today.