Search Beyond: How Will Machine Learning Change the Way We Do Marketing?

March 2017

Search Beyond is a forum that brings together the latest discussion from some of the UK’s most innovative independent digital agencies. A handful of senior practitioners meet on a quarterly basis to debate a topic facing the digital marketing industry, with the insights and output appearing here on Think with Google.

In our second Search Beyond session, we wanted to hear how independent agencies are preparing to face down increasing complexities in the digital landscape by adopting machine learning. The participants launched into the discussion by revealing how they’re already putting these tools to use in their own work on behalf of clients today.

“An element of automation is machine learning, so that is a lot of what we do”, said Maria Yiangou, Group Account Director at All Response Media. “In really basic form, using dynamic creative and optimising for the highest click-through is machine learning.” While many shared examples of their use of machines to take over grunt work, Angela Knibb, Head of Search at Greenlight, pointed out that the benefits go beyond that. “It does automate general process, but what’s great is that we can see patterns or trends where we need to focus our time. It makes us a lot more strategic in what we’re doing, as opposed to just pushing buttons.”

For Sam Fenton-Elstone, VCCP Media’s Chief Digital Officer, the task now is to expand the application of machine learning to not just unlock simple challenges like time, but to unlock real value. “A lot of elements, for example bidding or optimisation of creative, that’s a micro focus on one specific thing or one particular type of media using a simple rule-based system that ignores all the context around it. The exciting thing for us now is: how do we unlock value in the macro context?”

The group felt that bringing machine learning to the bigger picture isn’t so much a technological challenge as an operational one. For example, an agency might be interested in performing attribution modelling and exploring how different marketing activities affect in-store footfall. The client’s organisational structure can create a roadblock, though. Digital and offline teams may be separate, while TV, display, social and outdoor often occupy different silos. “Getting the best out of machine learning does require businesses to be joined up and thinking a lot more holistically. In my experience, a lot of businesses are surprisingly not joined up and not holistic”, Angela observed.

While a current debate around machine learning is the fear of human input and jobs becoming obsolete, the panel perceived far more opportunities than threats, arguing that machines can be better, faster and less prone to error than humans. According to Mark Williams, ‎Head of Search at iCrossing UK, they also reveal trends and anomalies people might not be able to perceive. “All humans tend over time to learn a certain route to solve a problem, and inherently you’ll always be biased towards routes that are easiest to follow. The great thing about machine learning is that it gets us into areas that humans just ignore. A disadvantage as a human is we’re just looking for the easiest route to get through the day!”

At the same time, the group was quick to defend the role of people in enabling the machines to do their best work. “You still need a human at the end of it, because an algorithm is only as good as the rules or process that you put in place”, Angela said. “So you still need a person saying, ‘Yes that’s a good idea’, or ‘Oh don’t do that!’” Sam expanded on this notion further, observing, “An algorithm only knows what it knows, and it doesn’t know what it doesn’t know. The beautiful thing about humanity is that we’re able to compute context much quicker, because we’re aware of the wider world.”

Far from being pessimistic, some participants – like Richard Hartley, Jellyfish PPC Director – are eager to see machines get smarter and smarter. “How long do we have to wait until we don’t have to ask any questions? When will the machine simply tell me what’s significant?” he asked.

A few in the group underlined the importance of having an understanding of how the back end operates, since many clients will always have a need to know what goes on inside the proverbial black box. Vim Badiani, Head of Search at Merkle | Periscopix, said, “I think that the lack of transparency is difficult for clients, so to tell a client that it’s machine learning that’s doing 'X' on your account doesn’t sit very well.” The consensus was that practitioners should at least understand enough about how their agency is using machine learning to be able to explain it to clients.

In terms of accelerating adoption of machine learning within the agencies, almost all of the guests spoke of the vastness of opportunity versus knowing just where to start. “The first step is to understand what’s possible, and I don't think that we’ve really scratched the surface”, Richard affirmed. “There are people who are data scientists and I am not; I’m a marketer so I need to say to these guys, ‘What can we do, how can we build it, how can that help make better campaigns?’ Then we need to roll it out to the agency and help people understand that it will benefit them, that it isn’t complicated, and that it can be simple, save time and drive real results. It’s not just this scary thing in the future, and you’re not going to get left behind!”

The Search Beyond contributors include:

Angela Knibb, Head of Search, Greenlight
Maria Yiangou, Group Account Director, All Response Media
Mark Williams ‎Head of Search, iCrossing UK
Richard Hartley, PPC Director, Jellyfish
Sam Fenton-Elstone Chief Digital Officer, VCCP Media
Vim Badiani, Head of Search, Merkle | Periscopix

Search in the Age of Assistance