In August 2016, Econsultancy published a report in association with IBM called The Secrets of Elite Analytics Practices.
Part of this wide ranging report seeks to discover just how automation and AI have changed analytics in marketing.
Let’s look at some of the talking points…
From automation to automaton?
Time was identified as a business’s most precious resource. Being able to streamline the marketing function through automation and, in particular, the analytics portion, was something executives deemed hugely valuable.
But is automation driving out innovation and originality? With so much potentially determined by machines and algorithms, do brands risk losing the essence that made them unique and the innovation that could keep them alive?
Understanding where automation delivers real results
Nearly 60% of respondents stated that their analytics solutions produced data-based insights without analyst involvement.
A further 80% of those stated it saved them significant time as a result. Either the analysts themselves could be redeployed to focus on trickier tasks or the insights generated pointed to opportunities elsewhere.
Hotel group IHG’s head of CRM, Jim Sprigg, explains his position on automation thusly: “Automation and machine learning will be critical for the sort of thinking that requires many calculations done in a somewhat predictable way.”
“It is definitely in our roadmap for broad use in predictive modeling which can drive, say, the assignment of offers and content in digital media based on individual customers’ attributes, behaviors and transaction histories.”
Dealing with the routine but complex
The idea of automation means different things to different people. For some organizations, particularly those that operate well on defined processes and rules-driven decision making, automation saves a great deal of time.
This is either because it can create a trickle down effect of automation allowing other systems to take appropriate action without human intervention, or alert business users when intervention is needed.
In complex, real time environments such as programmatic advertising, the automated processing of information into insights, and insights into action is viewed as essential to realizing opportunities before they pass brands by.
Danish AI-based media buying agency, Blackwood Seven, is expanding across Europe based on the success of its model that claims clients get a 25-50% improvement in the effect of their media (according to Campaign magazine) through using the company’s AI technology.
The software analyzes 82 different data inputs (such as sales, YouGov data and weather info) to determine a media plan’s likely outcome and optimizes in real time accordingly.
Can we automate creativity?
The advertising community is already looking at the question of whether or not automation and machine learning can actually create ad executions, not just supply humans with the insight with which to build their own creations.
However, the idea that AI would be integral to developing creative is still a pipedream. This begs the question, beyond a degree of grunt work or speedy number crunching to get the the right ads to the right audiences in real time, does automation have anything to contribute to the creative, innovative side of marketing.
In a discipline that has always been described as the marriage of art and science, can science begin to replicate (and replace) art?
The limits of automation
There is a sense, however, that analytics will never fully be automated. The feeling persists that strong marketing is an intelligent marriage of art and science, even in today’s data obsessed environment.
“Humans still have an advantage over computers,” Sprigg insists. “We used to call these the big ‘ah-ha’ insights. The sort that come from intuition and highly synthesized recognition.”
Sprigg gives the example of a time he showed the output from an automated learning process that suggested some offers landed differently with customers who came to the company via customer service than for those who used the web.
The group to whom Sprigg presented this data made the connection between the streams of information the computers already had access to – that there was a gap in the merchandising. “Humans were synthesizing information along with practical human experience in ways that we would have never known to code into the computer’s consideration set,” he adds.
Sprigg identifies that the biggest problem with automated analytics may yet be human in origin – it is a case of scenario planning.
Programmed with the information around any given scenario, a computer could undoubtedly come up with the relevant insight. It’s just that humans cannot prepare the machines to anticipate every possible nuance or scenario.
“Marketing functions can’t build automation for out-of-the-box thinking, but they can recruit for it,” Sprigg concludes.
Humans, while lacking in the number crunching abilities of their automated colleagues, benefit from years of emphatic “programming” that contributes hugely to strategic success.
The dangers of machine-based innovation
While marketers may be in danger of forgetting the worth of their human resources in favor of the speed and efficiency of automation, there is also a danger in relying 100% on data outputs to inform future direction.
Some executives interviewed for this report warned that an over-reliance on data to substantiate decision-making was hampering innovation.
The Hard Rock Cafe’s Claudia Infante complains that “the ideas that get shelved are the victims of a hybrid data-driven culture that we’re creating around ourselves.”
“We’re no longer as nimble and willing to go looking for the new shiny thing because we have to look at the data. There is no data to back those ideas up and you can’t get data unless you activate the idea.”
Paralysis by analysis
Automation builds a data-driven culture because it allows for faster reporting, analysis and optimization of existing channels, building on what is known. This use of data as a comfort blanket, however, can slowly suffocate innovative organizations.
On that note, Infante adds that “a company may have been innovative and forward thinking but of course it grows on the back of that success. Then, when you’re a big player, you have to take care of the day-to-day: make sure the lights are on, guarantee growth. As a result, we end up leaving behind a bit of that ideation process.”
It’s clear from Infante’s illustration that companies need to make sure they continue to use automation as a solution to an existing problem rather than a panacea for everything.
It may seem faster, and the results from it more tangible, but automation is not going to deliver on every aspect. It’s all about finding its place.
“Companies have many people analyzing reports trying to identify tactical performace gaps and opportunities so they make predictable adjustments. As a result, companies hire a lot of people who like trying to think like a computer. If that can all be automated, the goal should be to recruit different types of thinkers,” Sprigg explains.
Automation must be omni-channel
The immediate challenge for developers of analytics automation is in creating a solution that moves beyond the point and into an omni-channel environment. IHG’s Sprigg explains:
“Out-of-the-box solutions tend to be limited in their scope of operations. They are designed to optimize one channel or only one page at a time. This is multichannel optimization. We want to optimize in a way that allows us to maintain a consistent omni-channel experience across all channels and even extends to customer service and in-hotel interactions.”
Understand the question before anticipating the answer
Over and over again however, executives have reinforced the old computing adage of “garbage in, garbage out.” Automation in analytics is only ever going to be as good as the premise it is set up to work toward. Marketers must understand its power and its limitations to fully benefit from its potential.
For some, it is a simple question of plug and play to speed up number crunching and use the efficiencies to divert resources elsewhere.
Other organizations may find that embedding automation in their analytics process requires a wholesale change of departmental and HR organization. In some cases the whole culture of the company could change.
This post was co-written by Morag Cuddeford-Jones.