Artificial intelligence (AI) and machine learning (ML) are the trendy buzzwords in business today.
Just look at how searches for these concepts have risen in Google Trends over the past five years:
The increase in searches for AI and ML over the past five years.
Unlike the previous technology buzzword (“big data”), I suspect these technologies will have a far wider reaching impact on how we work.
But AI has been around since the 1950s. So why is it such an important topic today?
It’s because we’ve reached a tipping point where the impact of Moore’s Law (the doubling of processing power every 18 months) is accelerating the frequency with which new capabilities are launched that can eventually displace the need for human PPC managers.
The Acceleration of AI
A great example of the accelerating pace of innovation can be seen in Google DeepMind’s work on the AlphaGo algorithm. They spent years building AI to beat the world’s best human player of the game of Go.
It is estimated this work costs $250 million and, once completed, the machine took several months to study thousands of human games to become good enough to beat Lee Sedol.
What happened next is the crazy part.
Within the two years that followed, the team created AlphaGo Zero at a fraction of the cost. The new system took just days to play games against itself to become good enough to beat its previous iteration (AlphaGo) by a score of 100-0. It was faster, cheaper, and required no knowledge of human experts players of the game.
In less time than it takes most of us to get an undergraduate degree, the machines basically went from not playing this game, to becoming virtually unbeatable.
While AlphaGo doesn’t manage ads, the same technologies underpin how Google is changing how AdWords works.
Just this month we already saw two announcements that might make us question how much longer humans will be able to add value without adapting to how PPC is done in an AI world.
Responsive Search Ads Automate Ad Testing
First, we heard about responsive search ads.
Rather than writing entire ads by providing two lines of title text and one line of description, now advertisers submit several variations for each ad text component and AdWords automatically combines these to find what performs best.
Taking the individual components and multiplying them into different ads obviously is not the hard part of this, it could easily be done in spreadsheets already. But making sure that the resulting ads make sense and have a decent chance of being winners is hard.
AdWords gets paid for clicks so they can’t afford to waste limited ad space with nonsense ad combinations that nobody clicks on. Their AI must have gotten good enough that they feel most ads will be good enough to at least keep click volumes steady and will eventually improve to lead to better results than humans might not have been able to produce.
Now, this is not to say that humans can’t create better ads. But PPC pros tend to be busy and often run out of time to do everything they know they should be doing.
That’s why PPC tool companies exist in the first place. It’s also why Google believes that the overall performance of AdWords will be better if they help the masses automate ad testing, something they may not have done as much before.
Now busy advertisers can throw a lot of variations at the system and let it figure out what works best.
New Shopping Campaigns Are Automated
The other AI-driven launch we heard about recently is the introduction of a new shopping ads campaign that is more similar to universal app campaigns than to search campaigns.
As Google says:
“This new Shopping campaign type will automatically optimize your bids, identify the right audiences, and determine where your ads show to maximize your conversion value. We’ll do the heavy lifting so you’ll be able to focus on more strategic initiatives.”
This is getting close to the vision former Google CEO Eric Schmidt talked about over a decade ago. He envisioned a world where companies could ask AdWords to grow their business by providing some very simple inputs, like a business goal, and a list of what they sell, and let the system deliver the results.
So does this mean that figuring out the perfect structure, the right negative keywords, different bid adjustments, etc., no longer matters?
For now, I don’t believe so. Instead, think of this as the easy way to run shopping ads for someone who doesn’t know AdWords very well and who doesn’t have time to become an expert.
For PPC pros (like you reading this article), it does mean that competition is going to get a little stiffer.
By making it easier for non-experts to run decent campaigns, more businesses can come online to try AdWords and see good enough results to keep buying more of it.
Whereas in the past we’ve benefited from being experts who could figure out the complexities of shopping ads, now more and more advertisers can be good enough to be in the ads auction and we’ll need to work a little harder to keep an edge.
Why This Isn’t the End of Human PPC Management (Yet)
AdWords has always had a bit of a tough time getting casual advertisers to stick with AdWords.
It’s just hard to figure out the right combination of settings to get good results, and when the casual advertiser is up against competitors who have dedicated full-time PPC experts, even someone who did set it up right may feel they’re not getting as good results as they hoped.
These new AI-based solutions open up AdWords to new players. In a short amount of time, the AI may get so good that humans would be better off shifting their focus to things the machines are still not as good at.
In a way, we’re seeing the intrusion of AdWords Express into AdWords. AdWords Express is supposed to make things easy, but it’s a separate system and experts shun it. But experts could use a little help because there are now so many new ways to advertise that even they can fall behind on their knowledge.
For example, an ecommerce advertiser may have been doing search ads for 15 years but showcasing shopping ads is so new that they simply haven’t had as much time to figure out how to optimize these.
With all the demands on a PPC expert’s time, it’s just nice to be able to set it on autopilot and maybe deal with manual controls later when the time is available or when tools have caught up.
Humans Help the System Evolve
When it comes to writing ads, humans introduce variety into the ecosystem with ads that machines might simply not consider.
There’s the example of the jittering image ad on Facebook that was accidentally created by a human who didn’t save the animation correctly. But then they found the jitter did much better than a static ad and could be created at a fraction the cost of a full animated ad. That accidental discovery introduced something new for AI to consider and learn from.
In the same vein, while Google will run all ad variations, we still need creative humans to feed the components into the system.
Automation Has Co-Existed with Manual Management Since Broad Match Keywords
While full automation helps non-marketers take advantage of Google’s massive ad platform, it’s usually not favored by professional marketers who like the extra control.
Google’s broad match keyword was one of the earliest ways marketers could let Google just figure it out. But what percentage of experts’ accounts’ keywords nowadays are broad match?
If you have the time and tools to manage keywords properly, more tightly controlled match types can deliver better performance. Broad match still serves the function of generating new keyword ideas but they are just a small portion of the account of a PPC expert.
When Google launches fully automated account management, they want to make sure that in aggregate, things get better. They want user satisfaction to rise (and they measure that partly through improvements in CTR) and they want to protect their revenue (as measured in overall ad revenue, not necessarily better average CPCs).
But each of us is not average so by playing the game of AdWords a little better than others we can benefit significantly.
So for now, try the new automations and be ready for AI to evolve very rapidly and results to become very good.
Until then, use these automations to get into new areas, discover new opportunities, and be a helper to get you to your goal of getting more sales, more leads, and grow your company.
More Paid Search Resources:
- 5 Ways to Automate More of Your PPC This Year
- Machine Learning in AdWords: How & When to Use Smart Bidding
- 5 Things You Could Be Automating with AdWords Scripts
Screenshots by Frederick Vallaeys. Taken May 2018.