Over the last few years, automated ad strategies have moved beyond what you can achieve through manual ad management. Automated Shopping Ads and Performance Max draw from years of historical data and use this information to outperform manual ads.
Let’s look at how you can use automation to optimize Google Ad bidding.
The first thing we need to consider is your ad strategy. When you manually control your campaigns, you bring lots of indirect knowledge. For example, you might want to push a specific strategy because you know that certain products drive revenue. After all, they sell more or have larger margins.
Your strategy is essential because it determines how aggressive your ad volume should be. Many strategies exist outside of direct ad optimization. For example, certain customer acquisition strategies can count on future sales when calculating return on ad spend (ROAS). The value of each conversion can be more than a one-off sale, so it’s worth thinking about what you want to achieve.
How to deal with automation
“The good news about computers is that they do what you tell them to do. The bad news is that they do what you tell them to do.” Ted Nelson
The quote above is a funny way of highlighting that you need to think carefully about what you tell machines to do because they’ll carry out your instructions faithfully. This is true for your settings on Amanda AI or the instructions you give to Performance Max.
Let’s take profit bidding as an example. It’s a sound strategy where you prioritize promoting particular products based on the profit they generate. However, adding these criteria can limit your ad’s volume and reach.
Let’s say you pull up a Google Analytics property and select for profit instead of revenue. You sell this product for $100, and your profit is $10. You let Google Ads know, and you optimize for this profit.
Of course, not all products you sell have strong profit margins. However, they can still be strategically important to your business. These products work a little like “loss leaders” because while their profit margins are low, they help with customer acquisition and eventually contribute to revenue via follow-up sales. However, if you tell Google to show ads based on profit, it won’t prioritize these low-margin products so they won’t act as a magnet for new customers.
Similarly, if your competitors have the same product but they are prioritizing based on the actual revenue, you’ll most likely lose out in the ad bidding war. So, sometimes profit bidding means you increase profit but decrease revenue.
More variables, less volume
The more criteria that you select, the narrower your audience. Let’s use the example of the search term “shoes.”
Let’s say that you get about 100 searches a day from this term. Then, you set “search impression share” to 99% with an unlimited budget. Google will push out this ad as much as possible. But once you add more criteria for Google to consider, the volume will become smaller.
So, if you want to “maximize conversions,” Google will have the following criteria:
- the keyword “shoe”
- a set budget
- and a goal to get as many conversions as possible.
If Google wants to do this efficiently, it needs to consider contextual signals when the user is searching for shoes. It needs to think about things like past history to predict the conversion rate. If it can’t predict the rate, it can’t say that it’s worth pushing the ad to the user with much confidence.
This process means that the 100 searches a day shrinks considerably. Google doesn’t have info on all users due to incognito mode, privacy settings, new devices, and so on. The more criteria you input, the more predictions you ask Google to perform, and the more you narrow your audience.
Different types of data that determine automated bid strategies
Google draws on three different data sources for automated bid strategies. They are:
- Your input
- Local data
- Global data
Let’s explore each one.
You need to input something into the system even with an automated strategy. If you put in one keyword, you limit your ads to appearing for people searching that exact term. If you input lots of keywords, you risk burning your budget by diluting your focus users. You need to strike a balance between the two.
As we said, even with optimization, you need to add keywords, update your feeds, etc. Keywords represent the potential amount of impressions you can get. These impressions symbolize the number of clicks and conversions.
Automatic strategies can work in two ways: exploration vs. exploitation.
- Exploration involves running tests and finding new strategies.
- Exploitation relies on using past data to use strategies you are sure will work.
Google leans towards exploitation because it’s a responsible way to use your ad money. Customers generally don’t want to take the risk of exploration just in case it goes nowhere. While that makes sense and keeps ad costs down, as a marketing team, you still need a way to test and explore new strategies to get clicks on all your products.
The main types of local data are:
- Google Analytics audiences
- Customer match lists
- Your goals
Many businesses underutilize ad goals. There are solid arguments for setting up soft goals and conversions. For starters, they’re a great way to let Google know:
- A) who you consider a valuable user
- B) make it clear who your audience is.
Your local data is a gold mine. It’s a massive part of training algorithms and letting Google understand who your audience is.
Something else you need to think about here is the granularity of your remarketing lists. If you only add an “all users” list, you’re limiting the effectiveness of Google ad automation. If you add “abandoned carts,” you start to get closer because you have a more high-intent audience who are more likely to convert.
Smart bidding best practices for audiences
Once Google launched Performance Max, it became clear to everyone that its processes leaned heavily on audiences. When you set up an asset group, you need to share the audience that you are targeting to use as training material for the algorithms.
Granularity is key if we are thinking about audiences. More audiences are better because Google can combine them, and when you want to prioritize conversions, clicks, or impressions, more granularity equals better decisions.
Global data is the information Google has on users. It’s mostly a black box, but you can do a few things with it.
One thing you can do is improve on data. All data is good data, even bad data! For example, knowing which audiences or demographics aren’t working is very helpful.
While you won’t want to spend too much on finding this information, campaigns that generate clicks but not conversions are still valuable. For example, through the process of elimination, you can use this data to figure out what is working.
In short, the wrong audience can help you find the right audience. In the same way, you can use specific attributes to inform Google about who you should target. You can also tell it who it shouldn’t target.
Understanding Performance Max
Performance Max essentially works by combining global insights with retailer insights. There is a significant correlation between the size of your remarketing lists, your web page, and your potential budget.
The more users, lookalike lists, and relevant audience lists you have, the more you can spend on these campaigns. These campaigns are typically the best performers and rely on feeding Google the correct information to get the best results.
Can you influence an automated strategy?
This is a big question for marketing teams. While there seems to be some mystery around this question, you can find the answer in Google’s documentation. While it’s complex to do, you can influence these strategies in two different ways.
Historically, if you did manual bid adjustment, it was a way to change what you were prepared to pay per bid based on the user. For example, you could have a flat bid for all audiences but a slightly higher bid based on relevant details that make someone more likely to convert.
It’s also a way to both attribute and target smart strategies. If you’re pushing in bid adjustments, you are essentially telling Google which list to prioritize.
The other effect of bid adjustment is that it is an attribution. It won’t affect the cost per click, but it will prioritize between lists.
Essentially, the question becomes which list should be attributed if the user is on several lists. Almost all users are on several lists. With bid modifiers, you can ensure that the right list is being attributed. Doing this is essential because you have much more data on your audience in GA, your local data, and so on. Because you have this information, you can tell which are the common traits that make this user relevant to your business.
Google offers two ways to deal with this: the fine comb and the wide net. Because we use automated bidding, we push for the wide net mainly because the more data you have, the more options you have. Additionally, you lower the chances of missing out on important segments you don’t fully understand.
Conclusion: With automation comes complexity
Automation leads to more complexity. In the past, finding an underutilized category was enough to generate incredible results. But today, the competition is so fierce that we need to rely on automation to get an edge.
Big domains like Google, Bing, and Facebook have a vast amount of user data. When we give them the best possible data, they can perform as well as possible. The process is more complex than changing bids, but it’s the only way to compete in the modern digital marketing environment.
If you want Performance Max to work to the max, you need to combine your audience data with Google’s audience data. You can reap the rewards of better conversions, clicks, and sales from there.