I’ve been doing pay-per-click (PPC) since 1998, when virtually every setting was manual. While it was painful to manage everything by hand, it forced me to learn the ins and outs of PPC, and that helped me build a successful career.
Today, with automation playing an ever-more-important role in PPC, new account managers don’t have to learn all the fundamentals because tools handle the details.
But as humans learn how to co-manage accounts with artificial intelligence, I believe that those with the strongest fundamentals will have the best opportunities for career advancement.
Remember the days when photography was an expensive hobby? Before digital, every time you clicked the shutter, you used another frame of film. To see the result, you’d spend more money to get the roll of film developed and printed. And not only was it expensive, it was also slow, with most labs taking an hour or more to turn the film into a print.
In the pre-digital days of photography, it mattered that you understood how to frame a shot and set the right exposure to get a beautiful photo. If the photo you took didn’t look amazing, there was very little you could do easily and cheaply to turn it into a masterpiece.
Compare that to today, where everyone has a decent camera on their smartphone and hobbyists have mirrorless cameras that can shoot 60 frames per second. Results are instantaneous, and all the computing power in the camera almost guarantees correct focus and exposure.
In PPC, where Google is pushing really hard to automate as much account management as possible, we run the risk that new PPC professionals will grow up in an era where they do nothing manually and never learn the fundamentals.
So, why does it matter if the person you hire to manage your AdWords doesn’t know the fundamentals? Even my 4-year old takes great photos, after all. But there is a key difference between taking photos without knowing what an f-stop is and running an AdWords account without knowing how a cost per click (CPC) is calculated.
With photography, we can take shots with ten different exposures in one second, knowing that at least one would be great. It costs nothing to throw away the nine bad photos, but if we’re lazy as account managers and we do nine pointless experiments in AdWords, those clicks cost real money.
I’ve made the point before that testing the right things is what will set great agencies apart from mediocre ones. In PPC, that means that fundamentals still matter so that you can set up a reasonable test that gets as much as possible right, leaving fewer variables to test. This helps find winning tests more quickly, and that can make a huge impact.
When humans rely too much on machine learning, we run into the following issue. Google doesn’t care if it takes ten clicks before their bidding model starts to make good predictions; they still make money on every one of those clicks.
But advertisers should and do care about the amount of investment required to get to a stable situation where machines deliver predictably good results. This matters even more for small companies that may not have the cash flow to wait it out.
The more humans can help machines go in the right direction, the more money is saved and the happier the advertiser will ultimately be. Doing all that requires knowing the fundamentals.
In PPC, the fundamentals include the basics of how entities in AdWords work and interact with each other.
For example, what settings are done at the account, campaign, ad group, keyword and ad level? What are the different types of bids, and what are bid adjustments? Which bid adjustments work at what level? And the list goes on and on.
In my job as chief executive officer (CEO) of Optmyzr, I talk to a lot of advertisers, and it’s surprising which types of questions stump people who do PPC for a living. For example, you know what a return on ad spend (ROAS) bid strategy is and what it’s trying to achieve. But some advertisers who use our rules-based optimizations ask me about the formula we use to derive the CPC for a target ROAS strategy.
The basic equation we start with is simple, and anyone who does PPC should be able to come up with this. You get the correct CPC by dividing the historical revenue from ads by the number of clicks and multiplying that by the inverse of the target ROAS you’re trying to achieve.
The reason I think it’s a problem when advertisers can’t come up with this formula is that they’ll be hard-pressed when the time comes to use new business data to improve their methods for managing bids. So they’re basically stuck with whatever a tool gives them, and they have to rely on the company behind the tool to deliver further improvements.
I was a Google AdWords evangelist for a long time in the 2000s before starting my company. I was frequently on stage, getting peppered with live questions about some really in-the-weeds AdWords stuff.
I didn’t write the AdWords code, so I didn’t know exactly how the system operated, but I spent countless hours working with engineers, product managers, service reps and advertisers, and that helped me understand the fundamentals.
Here’s the thing: The engineer who wrote the code probably doesn’t know every corner case advertisers encounter, so they have an imperfect understanding of the system. And advertisers and reps, because they didn’t build the logic but can only see the results it produces, also don’t have a perfect understanding. But if you know how things are supposed to be built, and you’ve seen them interact in the real world, you can build up a really good intuition of how things will work.
As PPC pros, you have access to what Google says the system should do, along with tons of real-world data, so you’re in a great position to be a master of AdWords.
Here’s an example of a situation where misunderstanding the basics of how AdWords operates can hurt you. Say you’re using a script like the one written by Daniel Gilbert for hourly dayparts, and all of a sudden you start to see ads at times of the week when you thought you had no active dayparts.
If you think that dayparts are managed as budget periods (e.g., I have $0 budget from 10 p.m. until 6 a.m.), you would think using a script to set a -90 percent bid adjustment would effectively leave your ads turned off at those times of the week because there’s still $0 budget.
But that assumption would be false because dayparts are bid adjustments on top of the base bids, and so a -90 percent daypart would serve ads with a max CPC of 10 percent of the original bid.
As more of AdWords becomes automated, we also have to become better at understanding interactions. Fundamentals are really important for this, too. Say you’re using Google’s Target CPA (cost-per-acquisition) bidding method, and then you use a private tool to keep some ads in positions one, two and three.
These systems are both changing bids, so you’re likely to run into unexpected behavior. This is a simple example, but you can only get it right if you know how AdWords works and what levers the tools are pulling.
I get frequent news alerts from sites that deconstruct the latest Android Package (APK) for apps I use, like Google Photos. By scanning the code changes, they are often able to predict what features Google is on the verge of launching.
With Google Photos, that’s nice to know, but with pay-per-click, it’s a chance to be on the cutting edge of the thing that provides my livelihood and to get ahead of a lot of other advertisers.
At the end of the day, I believe it’s just common sense to try to understand the underlying systems that you depend on for a living. And as machines take over many of our day-to-day management tasks, knowing how they operate and how we can work collaboratively to improve results will help us have more successful careers.