6 ways ad agencies can thrive in an AI-first world
Artificial intelligence (AI) and machine learning have long been part of PPC — so why are AI and machine learning all of a sudden such hot topics? It is, in part, because exponential advances have now brought technology to the point where it can legitimately compete with the performance and precision of human account managers.
I recently covered the new roles humans should play in PPC as automation takes over. In this post, I’ll offer some ideas for what online marketing agencies should consider doing to remain successful in a world of AI-driven PPC management.
Be a master of process
According to the authors of the book “The Second Machine Age,” chess master Garry Kasparov offered an interesting insight into how humans and computers should work together after he became the first chess champion to be defeated by a computer in 1997. In matches after his loss to Deep Blue, he noticed a few things:
- A human player aided by a machine could beat a computer.
- When two human players were both assisted by a computer, the weaker human player with a good process could beat the stronger player with an inferior process.
The first point is covered in my previous post, and it is the foundation for why smart PPC managers will learn to collaborate with AI rather than compete against it.
The second point got me thinking about some other scenarios where the winners aren’t necessarily the most skilled. Does the world’s most successful coffee chain have the best baristas? Do the most successful hotels employ staff who innately know how to make guests happy?
No. In almost any scenario where humans are a big part of the experience, success is achieved by having a clear mission that is supported by a really strong process and tools to achieve the mission.
Hence, I believe that in the world of PPC agencies, a primary focus should be on building an amazing process and equipping the team with tools that make that process easy to follow. So as AI takes over some of the tasks in your agency, make sure your staff knows and follows the process for leveraging the technology to deliver results.
Accept that your old value proposition is toast
Consider how you convinced your existing clients to sign up with your agency. If your pitch included that you produce amazing results because you’re really good at bid management (something machines are getting really good at), you may need to tweak your positioning. You don’t want to make your main value proposition something that can be put on autopilot by anyone — and will hence become very difficult to price at a level that makes you successful.
That’s not to say that you should stop thinking about something like bid management altogether. Instead, you should offer skills that are complementary to the AI system rather than skills that compete against it.
Hal Varian, Google’s chief economist, gives the career advice to “become an indispensable complement to something that’s getting cheap and plentiful.” For example, become a data scientist because we’ll need more people to make sense of the data and to figure out how to turn new insights we get from more sophisticated AI into new strategies.
In the context of an ad agency, this makes a lot of sense. You want to be able to say you have great data scientists who can make sense of what the automated systems are doing and make solid recommendations for the next thing to test.
Determine your new value proposition
Do you know California’s largest agricultural export? I guessed wine, but the correct answer is almonds. How did this come to be? It turns out that almonds are easy to harvest mechanically; you basically have a machine that violently shakes the tree so the nuts fall down to be harvested. So farmers figured they could be more productive by using automation, and all of a sudden tomato fields across the state were turned into almond orchards.
But people want more than just almonds on their plates, so despite how automation moved an entire state’s economy in a certain direction, it also created opportunities for farmers who didn’t automate.
We can apply this analogy to paid search agencies. Thanks to advances in AI, it is a given that they will do a good job of managing bids, and it’s also assumed that this service will be cheap because technology has commoditized it.
Agencies, like farmers, can supplement their highly automatable service offerings with something that commands a higher fee. So figure out what will be your niche in things that are harder to automate. And think about why a client would want to hire you if you’re just as good as the next agency at managing bids. Figure out what additional services you are really good at that are harder to automate (for now) and can be used to win new business.
Be the best at testing because testing leads to innovation
Innovative agencies win awards, which makes it easier for them to land new clients and grow their business. But how can an agency be innovative in a world where a lot of the work is done by a handful of automated systems that produce similar results?
I believe economist Martin Weitzman’s recombinant view of innovation offers a possibility. Recombinant Innovation describes innovation as a process through which new ideas emerge as the combination of existing ideas. Thanks to better prediction systems using machine learning, it is now possible for agencies to test new ideas faster and to iterate faster. Hence, an agency that leverages machine learning for testing and has a really strong process will be able to out-innovate its competitors.
Innovation in an agency is to recombine ideas into valuable new ones. The problem with testing new ideas is that it used to take a lot of time. But thanks to technology, you can test more things more quickly, and the winning agencies will be those that are the fastest at finding new winners. And they can achieve this by prioritizing the most likely winners into the fastest process, with the best testing technology.
You need to monitor the tradeoffs between labor and technology
Business is a big optimization problem. As an agency owner, you balance labor (headcount), and capital investment (technology) to achieve outcomes with a target level of speed, quality and cost. As technology takes hold in more aspects of PPC management, knowing how to optimize the equation becomes critical.
What some advertisers fail to see is that there is no perfect technology (just as there is no perfect human employee), but if a technology gets you close enough to the desired result while freeing up your staff’s time to work on other things, that is a win.
We all hire people for our companies, even when we know that ALL humans make mistakes. But we hire the best we can because it gets us closer to our goals, even if not 100 percent of the way. So why should it be any different when we think about capital investments?
A former colleague of mine who is still at Google shared examples where advertisers told him that they would not use broad match because it resulted in some impressions for their ads on irrelevant queries. But when prodded further, they were unable to quantify the impact this had. In many cases, the additional clicks were negligible, while the time they could have saved by letting Google’s AI handle query exploration was significant.
In my view, this is a poor optimization of that account manager’s time. In exchange for a small sacrifice in targeting precision, they could have freed up billable hours worth hundreds of dollars.
Hire one extraordinary (wo)man
American philosopher Elbert Hubbard said that “one machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man.” And he was on to something. In engineering, a great engineer can do the work of 10 good engineers.
So, as more of an agency’s work gets done by machines and you need fewer humans to do repetitive work, having the smartest possible person to work on the tasks that remain will be more important than ever.
There’s never a boring day when working on PPC, mostly because Google pushes so many changes every year. But this year, AI is going to stir the pot and create some challenges unlike the ones we’ve been used to dealing with. Hopefully, some of the thoughts shared here will get you thinking about strategies for keeping your agency successful in a world of AI-first PPC.
Stay tuned for my next post in this series, where I’ll cover how the technology got us here and what we can automate today.