<< Back to Learning Resources
Approx Time

Ian Edwards, Planning Director at Facebook, talked to us about some of the effects of machine learning on the discipline of planning. Facebook hosted ‘Planning Reimagined’ a sold out event during EffWeek 2018.

Ian Edwards, Planning Director at Facebook

I came to Facebook with 14 years as a communications planner under my belt. I was a passionate supporter of the argument that the media agency adds enormous value, by building segments out of various data sources that allowed for more personalised or relevant communications.

Early on, I had a conversation with our Head of Auction Analytics that challenged me to think differently about targeting and the planner’s traditional role as the owner of the target audience.

The problems with targeting

Facebook is probably the most targetable mass media channel out there. But my senior colleague was adamant that targeting should be kept to an absolute minimum. He explained that we need to give the algorithm the biggest audience with which to work. This allows it to optimise the plan more effectively than if we take the approach of making assumptions about whom we should target. Put very simply; the algorithm can learn very efficiently which people, displaying which kind of signals, are most likely to take a particular action. It does this iteratively, far quicker than a human could. The caveat is, it needs a lot of data to do it. Traditional targeting restricts the amount of data available to the algorithm, so it takes longer to learn and optimise.

This was interesting. The argument here is that while personalising a campaign might make the communications a certain percentage more effective, at an individual level the costs can increase.

“The line ‘what is the minimum amount of personalisation for the maximum amount of impact’ has become a planning mantra at Facebook.”

Evidence for this is easily seen. When creating a campaign on Facebook advertisers have the option to use automatic placements. In that case, the system automatically places the ads across Facebooks entire family of products. However, if the advertiser disables this, they can then exclude certain products, for example, Stories, Instant Articles, or Instagram. We’ve tested this, running ads with and without automatic placements. Advertisers with direct response objectives such as sales conversions or web clicks, who are using automatic placements, see a two and a half times increase in effectiveness.

All this creates profound consequences for planning. It means that to be effective in many auction-based digital channels the emphasis must be on looking for commonality between audiences rather than differences. As Byron Sharp says, you need to target the whole market.. The line “what is the minimum amount of personalisation for the maximum amount of impact” has become a planning mantra at Facebook.

Measuring the effect of brand-building campaigns

This is easy to measure for direct response campaigns, where there are signals that can be used by an algorithm. For brand-building campaigns, it is harder to predict. The audience passively consumes the advertising and reveals relatively few clear signs that it has influenced them.

“The only way to definitively measure the effects of long-term brand-building is to do it in a new way”

For us to put a value on this exposure, and any changes it may have had on brand perception, we need to understand how a brand pays back over the long term. As discussed by Les Binet and Peter Field in their research, this is not something you can see quickly.  The only way to definitively measure the effects of long-term brand-building is to do it in a new way. Brands, industry and platforms like Facebook will have to work together and deliver campaigns with large, and long, holdouts in certain markets. For example, an advertiser might stop distributing any TV brand advertising in Scotland for one or even more years. That kind of measurement is entirely possible, especially on digital channels, but for it to be a real test of all channels, the industry and advertisers will need to make some big bets.

The rise of the machines vs the traditional planner

Garry Kasparov’s book Deep Think, published in 2017, plots the pivotal moment when IBM’s Deep Blue beat the current world chess champion back in 1997. It seems as though traditional planning faces a similar pivotal moment. Much of the optimisation that algorithms can achieve now was done manually in the past.

A man playing chess

When it comes to big data, machines will always win. What Kasparov noted in his book, with fascinating insight, was that computers play chess differently to humans. They achieve their wins through brute computational force, constantly tactically assessing the near horizon. Deep Blue was evaluating two million moves per second. Computers are great at optimising on execution like this, but the leap of creativity will always reside with a person.  Kasparov concluded that there is a point where machine intelligence ends and human creativity begins. Without the lightening-fast number-crunching abilities of an algorithm, human chess players achieve their wins through creative, longer-term strategic thinking, something which computers struggle with.

“No AI would create the Guinness Surfer”

Sir Roger Penrose, the British mathematician who has long spoken about the quantum nature of consciousness, believes that human thought is not merely computational and that we need a new understanding of the physical world that is outside of current science. He argues that there is clearly something special about the human mind that is difficult, or even impossible, to code for.

So, in my view, algorithms will never replace creativity. It remains crucial that any planning process gives people, not machines, the freedom to think creatively. In a world where people can decide what they watch, the ability to come up with exciting and surprising ideas is vital. No AI would create the Guinness Surfer, or determine that the best way to sell Volvo was with Jean-Claude Van Damme doing the splits between two moving trucks. The combination of machine intelligence and optimisation with the creativity and novelty of the human intellect will always be more powerful than a pure AI solution.

Follow @EffWorks for updates and first access to all our new research.

“Just what the industry needs, great collaboration between clients and agencies on the topics that drive business growth.”

Bridget Angear, Joint Chief Strategy Officer at AMV BBDO

“It’s great to see the IPA in the UK bring the whole industry and particularly the trade bodies together to focus on effectiveness. This new Marketing Effectiveness initiative will enable people across the industry to work together to build on best practice.”

David Wheldon, Chief Marketing Officer, RBS

“Effectiveness is a team sport, so it was great to see the industry in the widest sense, come together. In an increasingly diverse and fragmented world, only by using all parts of the brain will we solve effectiveness challenges and design our campaigns to deliver short and long term value. That’s why what happens next is important – if the IPA can help facilitate progress on this with a long-term initiative around Marketing Effectiveness, we’ll definitely crack it.”

Bart Michels, Global CEO Kantar Added Value and Country Leader Kantar UK

“The time spent at #EffWeek was extraordinarily effective. It was great to hear the diverse views from all areas of the industry. All tied together with the common themes of accountability and effectiveness.”

Andrew Canter, Global CEO, BCMA

“It has been a privilege to be part of the inaugural Effectiveness Week. The agenda is one which we at O2 UK feel passionately about. To see and hear perspectives across the industry demonstrates how the breadth of marketing effectiveness is increasingly being valued within businesses. Data, insight, social, customer experience, test and learn, ROI, these are all fundamentals and were covered expansively at the event”.

Sandra Fazackerley, Marketing & Consumer, Telefónica UK Limited

“The full week of effectiveness events brought into clear focus the need for marketers to use data and insight to achieve the key business objectives of growth and profits. Marketers today are in a better position to quantify their knowledge of customers and measure the ability of investments in marketing to increase brand and shareholder value.”

Chris Combemale, Group CEO, DMA