This year, a change to Google Ads will affect nearly every Magento merchant. Smart Shopping campaigns are being converted to the new Performance Max campaign type. These new campaigns represent a radical departure from the way shopping ads have traditionally operated, and we can’t just “upgrade” our campaigns and expect them to work optimally. As we give up more control to automated systems, the data integrations with our ecommerce platforms are more important than ever. In this presentation, we cover the key issues that impact merchants. We provide tips and best practices for optimizing Magento to maximize performance on Performance Max campaigns.
Presented by David Deppner
Meet Magento Singapore, August 25, 2022
TRANSCRIPT
Good morning everybody. So, I’ll be speaking today about the Google Ads Performance Max campaigns and quite a bit about, some Magento integrations that can make the campaigns perform better and some other tips and techniques, covering just a wide variety of topics. If this is interesting to you and, any of these slides would be useful, feel free to reach out.
So Psyberware, just real brief, we manage advertising accounts for various e-commerce clients. So we work with a lot of technical integration issues, and we work with everything around conversion tracking, shopping feed data, right. Audience management and e-commerce integrations with with various ad platforms.
There we go. The agenda for today is we’re going to talk a little bit about the Performance Max campaign type. And then we’re going to go through a set of recommendations that are going to cover a wide variety of topics. And I’ll summarize it pretty quickly at the end. We’re running a little bit late, so I’m going to see if I can make up some time here as well.
So whenever an ad platform comes out with a new feature, it’s the best thing in the world ever, right? So Google will tout the Performance Max campaign type as the next best thing, a fully automated, machine learning, AI driven campaign type that’s going to solve all your problems and make you a lot of money. And they’re interesting. These campaign types are replacing the smart shopping campaigns that most merchants use.
So a lot of merchants have already upgraded from smart shopping to Performance Max, and most of the rest will be doing that within the next month or two. These campaigns don’t just run shopping ads on Google search. They run ads everywhere, basically anywhere. Google places ads on their display network, on Google search, discover YouTube everywhere. Right, Gmail. These ads cover the whole gamut of possible ad placements.
Unlike a lot of other Google campaign types, these campaigns have asset groups instead of ad groups. So we don’t actually create ads. What we do is we create a collection of marketing and advertising assets like videos, images, text, and then we turn over control to the machine learning algorithm to figure out where those assets will perform the best, and how to combine those assets together into ads.
These campaigns use audiences in a very different way. Traditionally, Google Ads campaigns would allow you to target particular audiences or allow you to assign bid adjustments up or down for various audiences. These campaigns instead use audiences as hints. They’re signals to the automation. They’re inputs where we say we believe people in this audience will probably perform better. So try them first.
But the audiences don’t limit who the ads can show to. Google will use those at first and then learn what’s working, and then expand far beyond the audiences we set.
Since these campaigns are automated by machine learning algorithms, there are very few settings that we can actually configure.
So the ads can look like this. Just a few examples. We’ve got traditional shopping ads that we’re all used to seeing on Google search, but these campaigns can also serve text ads. In the Google Display Network. We can do all sorts of display advertising, and they’re very similar to the smart display campaigns with, responsive display assets. They can show traditional image based display advertising or the dynamic remarketing product ads.
Over on YouTube, we could see product ads. We could see videos being displayed or, dynamically created display ads as well. So what’s wrong with these campaigns? There are some risks here because they have so many different ways they can try to create advertising. They can combine your assets and so many different patterns. And when they first start, they don’t have any information about what works and where it works.
If people have very small campaigns, then we can have issues with not enough data in a campaign to optimize. Well, these campaigns can take a while to learn what works and what doesn’t. And during that learning period, the performance may not be what you expect. And this is an issue right now as people first start to transition from smart shopping campaigns to Performance Max. The way to solve this, since we don’t have a lot of configuration options, the way to solve this is to make sure that the data we’re feeding into the machine learning algorithms is the highest quality data possible, and that’s going to be a key thing, the theme through the rest of these slides. If we can give the machine learning algorithms good inputs, then we’re going to get much better outputs.
So tip number one is to include all possible assets in your asset groups for Performance Max campaigns. There are a lot of assets that we could create. These are not just as simple as assigning a shopping feed to the campaign and then letting the products run. We can put all sorts of images in different formats. We can put videos on here.
If we don’t add videos to these campaigns, Google will use our static image assets to create slideshows. So if you want to control the videos that are being shown on YouTube, you need to put videos on these campaigns. There are also quite a few different options for the text fields and fields of different lengths. You want to fill those out.
All of them, right? So you have fields that can show in different contexts where there’s more or less text that can show. There is another way that some people think about these, and that is since there’s so many different possible ad combinations that can be created by these campaigns, it might be better to actually limit the number of assets you provide initially when you’re first deploying these campaigns.
So there’s fewer combinations that can be tried. And then to flesh out the campaigns more and add additional assets as you go. But I haven’t seen good data to show that that actually works or not. It’s a competing theory right now, but I think probably the best results are going to come from giving Google the opportunity to figure out the largest possible set of options, to figure out what’s going to work and where it’s going to work.
Tip to minimize the number of campaigns. This is because the campaigns are basically a collection of data about what is being tried and what’s working. If you have a lot of small campaigns, these campaigns take longer to learn what works. If you can combine some smaller campaigns together into larger campaigns, the larger campaigns will perform better and learn faster.
Many merchants like to split up their advertising campaigns by product line or by marketing objectives. There’s a lot of good reasons to split campaigns, like maybe you have campaigns that are market advertising to different geographic locations and different languages, but if you’re just organizing your campaigns for reporting purposes, that’s not good. You’re splitting up the campaigns into smaller buckets of data.
Better to merge those together and figure out your reporting via other mechanisms. You don’t need campaign structure to give you your reporting structure. Tip three make asset groups for major product lines. This relates to the previous point. We don’t want to have a lot of campaigns, but within the campaigns we can create different asset groups for different product lines or different marketing objectives.
So in your Magento site, maybe you have a complex category structure with a lot of different product lines, and those may really be targeted at different audiences, different people. This is where we can set up asset groups within larger campaigns to combine the advertising assets (the images, the text, the videos) and audience signals for the people most likely to buy those products. And instead group them together into larger campaigns.
In order to take effective advantage of the dynamic remarketing display ads. You need to pass the remarketing IDs with the remarketing tags that you need to pass the product IDs. And this is a problem a lot of Magento merchants, face. They think they have this set up right, but it’s frequently not set up right. The problem comes because, well, back up for a minute.
So when a user is browsing the Magento site and they go onto a product page that fires an ID along with the remarketing tag over to Google Ads, that allows Google ads to show these, remarketing ads that show the particular products users have browsed. Right. So that’s what we’re talking about. The problem that comes up on a lot of Magento sites is we really have two unique IDs for every product.
We have the SKU and we have the entity ID, and frequently we get mismatches between these. Whoever set up the shopping feed maybe added the SKU as the primary key in the shopping feed. But whoever set up the front end remarketing audience tagging, set up the entity ID to be fired as the product ID. When that fires over to Google, they don’t match.
So Google does provide a secondary ID field, the display ads ID field, and that allows you to put in an alternate product ID. So this is really, really great for Magento merchants. If your feed uses SKU as the primary ID, it’s oftentimes good to just add your entity ID to the display ads ID field. So both of those IDs are associated with every product.
The other option or the other problem that comes up has to do with configurable products. When you’re browsing a configurable product parent, it’s going to fire the ID of that parent. But the child simple products are actually the products that are in your shopping feed. So the way to solve that is to add the ID of the parent configurable as into the item group ID field in your shopping feed.
So tip number five is to use customer match audiences. Customer match audiences are a mechanism to allow you to define audiences based on user data, like their email addresses. Instead of setting cookies on the browser.
So we do this in a privacy safe way, and we create unique IDs based on hashing the user email address. And we can send a big list of people over to Google. Once we send that big list over to Google, Google checks it against the users that are logged into the advertising systems and can add those that they can figure out that match.
They can add those to your audiences. There’s a lot of opportunities to do work on this and make your audiences better. So, for example, if you have a lot of different asset groups for very specific product lines, you can create customer match audiences of past purchasers of people that bought particular categories of products. And then you can create similar audiences based on those and add those to your Performance Max asset groups that are targeting those product lines. Requires a little bit of custom programing, some integration with Magento to do things like that, but it can really improve the performance of these new campaigns.
Our next tip is to use Google’s built in audiences. Google has quite a few built in audiences, like in-market audiences of people who are shopping for particular products, audiences of people based on their demographic profiles. People that recently started businesses, or people that are looking for particular things, or people that just got married or about to get married.
These audiences can just add additional signals to Performance Max, and help Google ramp up more quickly and find the right people.
Tip number seven use custom audience segments. Performance Max campaigns can’t target based on keyword matching, but you can create audiences based on people who have been searching for particular keywords on Google. And you can also create audiences based on what websites they’ve been browsing, right? This is a really powerful targeting tool. So in this particular example, we have a client who is selling coffee and K-Cup compatible coffee pods.
Well, we can create a list of, based on people who are searching for Keurig products and go to the Keurig website. And this client can then target people like that. We’re signaling to the Performance Max campaign. These are the people most likely to buy.
Tip eight improve your shopping data feed. I could talk about this all day. There’s so many issues. So many developers who are working on feeds just do the minimum possible to make the products work to get them approved in the feed so they’re not disapproved. But if you read down further in the feed spec, for every possible attribute, there’s a section called Best Practices.
And that Best Practices section has a lot of information about what else you can do to improve the quality of data to drive better advertising results. So this is really an opportunity for developers to support marketing teams a whole lot better and do a better job with shopping feeds. In Google Merchant Center. There are we usually focus on getting products approved.
So we’re always looking at the errors. We’re always looking at the ways that we can approve more products by fixing the errors that are disapproved. Oftentimes, we skip over all the warnings that Google Merchant Center is giving us. The warnings in Merchant Center are clues to you that performance could be better if you improve the data in particular areas.
And sometimes that means fixing the data in Magento. Other times that we have integration issues, with getting the right data out of Magento. So for example, in this list it says warning some of your color fields aren’t set correctly. Well, Magento has a color attribute that’s telling us maybe we need to set it, or maybe we need to get the data out of Magento into our shopping feed from that field.
And finally, I’ll just say shopping feeds oftentimes are stagnant. We make them, we set them up, they work, and then we don’t touch them again for years. But the Google Shopping feed changes every year. So here are two fields that Google added to the shopping feeds I think in 2020, two years ago. And most merchants don’t implement these yet.
Product detail and product highlights. Product highlights is basically a bullet point list of the key features of your product. Might be good to put in your shopping feed. Product detail basically allows you to do any arbitrary key value pairs you want to. You can extend the feed with additional data. So all sorts of other attributes that you might have in Magento you can export to this feed attribute, provide more information to Google to optimize who’s seeing the ads.
Tip number nine is to optimize update your conversion tracking. If you’re importing conversion tracking from analytics, there can be a lot of problems with that. There’s a lot of little things can go wrong. And there’s a shift right now between Universal Analytics and GA4 right. So that isn’t a good a good way to do it. Also, most Magento conversion tracking for Google Ads is very out of date.
So Magento 2.4.5 just got released this month and they’ve updated the ad tracking in 2.4.5, but almost nobody’s on 2.4.5 yet. So users, merchants on older versions of Magento, that tracking is now about five years out of date. So you want to make sure that you’re using modern conversion tracking, using the G tags library, or what Google is now calling the Google Tag, or have your tag conversion tracking set up in Google Tag Manager, right?
Those will help quite a bit and provide additional conversion data to the algorithm so it can learn better. Finally, set up enhanced conversion tracking. Enhanced conversion tracking is similar to how customer match audiences work. We provide some hashed user data along with the conversion tag. So we say to Google somebody with an ID based on this email address just made a purchase.
Google can use that information for cross-device tracking or to figure out if somebody’s converted in situations where cookies are blocked, right? Or if somebody, you know, upgrades their web browser, refreshes their cache, all that sort of stuff so it can improve the conversion tracking, not by a lot, but sometimes by 5 or 10%. And that little bit of extra data about what works feeding into the algorithm has a cumulative effect over time, and really trains the AI better. So these campaigns can perform better.
So let’s wrap it up. What matters the most, right? We want lots of assets in the asset groups on these campaigns, large campaigns and as few campaigns as possible. Get larger buckets of data. Focused asset groups allow you to focus your audience targeting, as well as the assets and products that those people are most likely to buy. Figure out your audience integrations. These campaigns use audiences in a different way from other Google campaigns.
Your audience integrations can really matter. Shopping feed data is something that everybody could use more work on, and it requires constant, constant vigilance because things are always going wrong and there’s always changes. There’s always product data issues there. In conversion tracking. Modernize your conversion tracking. Make sure you’re following the latest Google best practices to get the best data possible to train the algorithms.
Because ultimately they’re optimizing based on that signal more than anything else, right? They’re optimizing for conversion value. Remember that the way to make these campaigns work is to feed the best possible data into them. We can’t configure all the things we used to be able to configure. We can’t mark negative keywords. There’s all sorts of stuff we can’t do.
We can’t control where these ads show up. What we can control is the quality of the data that’s being used to train the machine learning algorithms. So focus on that.
And with that, I’m done with the main presentation.