AI is disrupting the Search Industry
It is only fair that I start with a confession: Facebook thinks I am German. So they serve me German language advertising. I am many things, Danish national, French resident, Estonian e-resident. I also speak several languages: Danish, French, Spanish, English… but dear Facebook: I am not German, I don’t understand those ads, and you are ripping your advertisers off on my account. Sure, I can understand a little bit of German but somewhere along the line, your machine learning must have choked on me.
Where is the Unlearn button again?
The Search Industry is at the forefront of the upcoming revolution of work. We work directly on the platforms of some of the biggest AI players on the globe: Google, Facebook, Microsoft, Amazon. And we all feel it as our duty to try the “latest functionality” to “join a Beta”, to “try new things for our campaigns”. We are more than willing to try machine-learning and AI-driven features and functionalities and probably have our activity x-rayed and monitored for results.
And indeed, the Search industry is an excellent place to apply artificial intelligence. We work 100% digitally. We handle structured data in the form of numbers, text and the occasional images. We work in automated or semi-automated systems. We do repetitive tasks. And hey, we love tools, we love new features, we are used to constant change.
One of the prerequisites for applying AI is of course Automation, so how are we doing there?
In Innovell’s research for the “Major trends in Paid Search” report, we found that the philosophy with regards to Automation of the leading Paid Search teams in Europe and North America was to completely embrace it: two thirds of the respondents stated either “We are building our own automation stack with external and proprietary tools” or the even more extreme “Anything that can be automated, we automate”. In fact, none of respondents stated that ”Most things are best done by Humans”.
And for those who haven’t noticed, there is already plenty of AI and machine learning in our work environment. The entire paid search industry was founded on the Adwords algorithm which factors in keywords, bids and quality score – quality score? That is machine learning right there. We have been working with quality score since we built our first campaigns. Machine learning was there before we started, what is there to be afraid of?
We also asked the leading Paid Search teams what proportion of their campaigns was managed by AI or machine learning. The average response was around 50% – half of the campaign management is based on AI or machine learning. Are we already half replacing ourselves with Automation and AI? Why would we do that?
Marc Poirier from Acquisio gave a good answer to that question in the research he presented at Hero Conf London comparing performance of campaigns with and without the use of machine learning across a large volume of Paid Search accounts. The “adopters” segments experienced between 39 and 280% better conversion rate depending on the vertical they belonged to Automotive, Education, Medical, Financial, Retail.
Those who used machine learning performed better.
But once you switch to an AI autopilot on your campaigns, you run into a dilemma: will you Learn or Perform?
As of today, AI solutions generate superior results. But as marketers in a moving market, our added value is to understand, interpret, improve in a continuous way! If we use AI, can we still define the winning strategy? Or do we need to find other ways of adding that value? With AI, we will deliver performance today, yes – what about tomorrow? We have built our industry celebrating our successes and learning from our mistakes. It is a deeply rooted principle and one we take pride in. We test and learn and succeed. With AI, there is only 1 option: perform. And once everyone is on AI, that performance could be commoditized away and maybe we didn’t learn anything?!
And there is the problem of AI: the training of an AI goes through many thousands of iterations. We end up with the best result but all the intermediary calculations were wiped out. We don’t know how we got there. We didn’t learn anything from the process…
Fred Vallaeys, in his keynote on Hero Conf London, illustrated the fear of AI with the story of Lee Sedol, the world champion at the game of Go being beaten by DeepMind’s AlphaGo. It was artificial intelligence which had been trained be studying 100 thousand games played by humans and it ended up winning games against Sedol. The scary bit is the next iteration of AlphaGo, called AlphaGo Zero. This Artificial Intelligence played against itself a million times and beat the first AlphaGo 100 to 0. There were no humans in that equation.
The respected economist Schumpeter described the effects of accelerated innovation on an industry as “Creative Destruction”. Something is broken down and something else arises from the ashes. It is a good description of what constantly happens in Digital Marketing.
Sometimes it is a process of disruption, where the creativity comes from new businesses, but sometimes businesses are capable of reinventing themselves, changing business models, changing service offering, changing positioning. I believe that, if there is an industry capable of doing just that, it is the Search industry.
Search Marketers are AI Natives, they were bred with constant change. In a certain manner, the Search Industry is already being disrupted but we only realize this clearly today because of the AI hype of 2018. Other industries are only just starting to face this new challenge but in Search we are better prepared. This doesn’t mean that we have nothing to worry about. If we blindly trust the machines to optimize our campaigns and no longer learn, then we will have lost our added value and commoditized ourselves. I don’t see that happening. Actually, I quite like to watch the way the resilient Search Marketers constantly reinvent themselves, like surfers on the waves of creative destruction.
Get more updates about the speakers and content you’ll find at Hero Conf.