Digital Marketing in an AI World: Futureproofing Your PPC Agency

"You as the marketer need to identify the personas, your major target audience segments, and write compelling creative for them. Machine learning can then deploy your work to the right users at the right times."

"it is worth spending three hours manually pouring over all these query reports to find more negative keywords. Is the three hours of work you did worth more than the five dollars you might have saved?"

"Machine learning can now be deployed to find and categorize any number of microsegments, based either on criteria you supply or by recognizing patterns in data that you would have otherwise missed. You as an agency then need to make sure you’ve provided the system with enough ad-text variations to test which messages will resonate most with each of those segments."

"Your team should include people not only capable of working in Excel but also with analysis-oriented tools like R and Tableau, Google’s BQML (Big Query Machine Learning), and third-party PPC automation tools like Optmyzr. By the same token, the skillsets you will need to hire for tomorrow will equate to tomorrow’s tools. To position yourself for the future, your employees will need to keep reskilling."

"By black box, I mean that you may not be able to tell if the system is taking a certain factor, such as snowfall levels, into account when it places an automatic bid. That’s not at all important in most cases—which is why it may not already be built into the system—but it’s very important if you are selling ski resort passes. That’s why you need to understand these technologies on a business rather than technical level. Is the machine learning model overlooking something that is a big driver for your sales? By all means, continue to use Google Smart Bidding, but realize that only you can determine if you need to tweak the system to consider missing factors or data points that can mean the difference between success and failure, or, just as importantly, between average and optimized performance."

"Reskilling employees requires a real investment of both time and money and should be a triangulation of self-teaching, conferences, and using case studies. Self-teaching means giving time to research topics by themselves or in small groups. Conferences allow people to learn entirely new and proven approaches, and case studies provide a hands-on way to learn, based on highly applicable situations. Once your staff has acquired the knowledge, you need to transfer it to the team and new hires. Most agencies are bad at knowledge management, so this will require some work. Give employees the technical writing skills to capture knowledge, have a central location to store information, and a system for updating documentation."

"The skillsets where machine learning lags behind human ability—and may never surpass it in the future—include understanding your client’s nuanced needs, creativity, and the ability to innovate. It’s people, not machines, who come up with new ideas and experiments. By having a good sense of what has worked for similar campaigns in the past, PPC professionals can bypass the expensive experiments that a machine learning system would need to conduct to arrive at the same conclusion. If you’re able to draw on your experience, you can save a lot of time and money."