Google has integrated AI into its Paid Search ad platform, leading the industry in this regard. It has done so with its Smart Bidding feature and automated ad creative, as well as various other innovations that we might categorize under the audience engagement umbrella.
These changes are intended to enhance campaign performance by leveraging the massive datasets the company possesses and the numerous near-instant decisions it must make daily.
For instance, Smart Bidding utilizes machine learning models to analyze vast volumes of historical and real-time data. This allows advertisers to set bids that drive conversion performance (or achieve other KPIs) at scale. Automated ad creative improves efficiency by generating and optimizing (read: split-testing) copy variations. Dynamic search ads are even smarter; they generate keyword groupings and copy that are a much better match for users' search queries. Then there is Enhanced audience targeting, which uses behavioral data to segment audiences ever more finely and boost relevance and performance.
Although these AI features bring about a more efficient and effective advertising system, they aren't without some drawbacks. The biggest concern is around control; with many 'smart' decisions being made on our behalf, the risk of ceding too much control is at the forefront. What happens if the AI system doesn't make decisions that we're happy with? And is it even possible to realign the AI system with our old control and decision-making processes?
And then there are our data dependencies (or dependences, if you prefer). Leaving aside for now the critical issue of data privacy and what can happen if our data ends up in the wrong hands, data-quality risk is a huge potential downside to our advertising efficiency.
A mix of feedback has been received from industry experts and case studies about these implementations. Some have seen apparent successes with some metrics, including engagement rates and conversion volumes, especially for companies that have achieved a high level of data maturity. However, even those we've spoken to who view AI as a positive step for the industry believe that underperforming campaigns could pose a significant risk for the type of companies we're primarily discussing. Why? Because failure could lead to a fresh wave of skepticism toward AI. As we've noted, some experts believe that campaigns utilizing AI are already underperforming in some instances.
Campaign performance outcomes when using AI and automation inevitably involve a trade-off between efficiency and control. AI can scale massively and perform well under certain conditions, and automated solutions can often replace inefficient human campaign managers. But when human beings do campaign management, they place at least three essential ingredients into the mix, with which AI and its output must get along:
I. Brand Messaging
II. Strategic Alignment with Business Outcomes
III. Integration Across Marketing Channels
The evolution we're seeing now with AI and automation in campaign management inevitably leads to this rethinking of where we humans fit back into the equation.
Today's marketers must transition from hands-on bid adjustments to interpreting AI-driven insights and providing strategic oversight of campaigns. This has led to the emergence of a new and somewhat nebulous set of marketing skills. Data analytics. A new era of critical thinking. And, something that is almost taboo to mention in certain circles, a deeper understanding of machine learning algorithms. We have long been a society of storytellers and, even with the emergence of the AI-based age of advertising, these aforementioned skill sets aren’t going to be replaced anytime soon. If anything, our new era of machine learning will only add another layer of complexity to the journey from data to insight, and ultimately to the increasingly liberated world of human-to-human storytelling.