Inside Google Marketing: How we’re unlocking the power of machine learning

Inside Google Marketing: How we’re unlocking the power of machine learning

Published
September 2019

Machine learning is a term we hear all the time at Google. “Everyone loves it — machine learning is everyone’s BFF these days,” says Avinash Kaushik, who leads a team of analysts in Google’s marketing department. But putting all the hype aside, can the latest marketing trend actually help companies achieve important business goals? That's a question Kaushik’s team is trying to answer. "I spend most of my time trying to figure out how we can unlock the power of machine learning," he says. Here's how his team is tackling that challenge and what they've learnt along the way.

0:01

At the moment, every company on the planet is doing some things that are

0:09

super fricking cool that they're not doing as much as they should.

0:14

That's where data plays a role. My name is Avinash, I work with a team that

0:19

focuses on super-advanced analytics at Google. Where I have spent most of my time is

0:22

trying to figure out how to unlock the power of machine learning when it comes

0:25

to analytics. Everybody loves machine learning. It's

0:31

everybody's BFF these days. But for us, when we deal with the size and scope of

0:35

data we have, being able to figure out how to have machines do things that

0:42

humans can't detect has been sort of a big challenge as well as a very big high for

0:47

us. With all the change that we have to deal with, I have a very simple framework

0:53

we use which is 70/20/10. 70% of the time we're going to focus on things that we know

0:57

that are very core to our business. 20% is where we're trying to push the

1:01

boundaries. You get into the known unknowns. And the last 10% is a true

1:06

crazy experimental stuff. Trying to figure out how to do uncomfortable

1:10

things that we're going to fail at more than we will succeed at. But with every

1:14

success you build a competitive advantage. A lot of marketers worry about

1:19

data as being the death of creativity. But I see data as allowing you to

1:25

unleash even more creativity than in the past. For me, knowing that we have access to

1:32

data, almost in real time usually, means I can take on a lot more risk. In terms of

1:35

trends coming down in marketing, automation, and intelligence are the two

1:40

things that are the most disruptive force on the horizon. Automation will

1:44

improve efficiency, which will improve agility, which will improve the speed at

1:48

which we can move as marketers. Then the other one is intelligence. Just our

1:52

ability to understand things at a scale that we've not been able to in the

1:55

past. Automation and intelligence combined

1:59

together will allow us to have longer term, more meaningful relationship

2:03

with our end users and customers. I'm very, very excited about it, because

2:09

marketing's job will not only be to interrupt, but to actually deliver delight.

Inside Google Marketing: 4 ways we’re planning for an automated future