We had the wonderful Ginny Marvin of Google Ads join us for both of our chats the week after this year’s Google Marketing Live (GML) to answer your questions! This is a monster post, so you might want to grab a snack or more water or coffee before you begin…
Note, that Ginny has also updated some of her responses from the chats here to provide more information, clarification and links where helpful.
Q1: I read about Trusted Execution Environments in github but it’s a bunch of adtech gobbledygook. Can you explain it in plain English please?
Great question. More to come on this, but the short of it is that a TEE is a place where data is secured and encrypted. In this case, it means only advertisers/designated users will have access to their data and control and transparency over how it is used.
TEE = Trusted Execution Environments
This will be available for advertisers over the course of this year. (No action needed.)
Q2: Can we get more clarity on the optimization score stat? Were the gains because they acted on recommendations or dismissed them?
The stat shared was that advertisers who increased Optiscore 10 points also saw an average 15% conversions lift. In other words, on average, overall improvements correlate to a lift in conversions. It isn’t looking at improvements from any one tactic.
Optiscore is a diagnostic tool that provides an estimate of how well your account is set up to perform, and recommendations are focused on your goal based on your bid strategy, such as conversions/conversion value, impression share, or clicks. You’ll see a score uplift for each recommendation that reflects the estimated impact of the recommendation on the campaign.
Changes to your Optiscore can come from a broad range of tactics, including actions taken outside of specific recommendations. You can dismiss recommendations you don’t find relevant to your campaign. You know your business best. My suggestion is to focus on what you think is most relevant to the specific account you’re optimizing.
Q3: I wonder how/if the AI Overview will affect the ad auction?
It’s still early days, and we’re experimenting with multiple options to learn what works best for consumers/users and advertisers.
Q4: Can you provide some more detail on image generation tools? Specifically if your uploaded image assets include product photos with brand logos in them, will the AI be able to render variations that include the products with the branding?
There are some content restrictions when generating images. Logos are one example. The image generation tools will avoid generating content with products with brand names and logos. The tools will also avoid generating images with individual people, for example.
We also have strict ads policies that govern the types of ads and advertisers allowed on the platform. Note that our policies and reviews for Brand Identity assets (such as business name and logo assets) are not changing with this feature.
This page has more details on content restrictions and eligibility for using image generation tools
Q5: How firewalled is your profit margin data? Will it be used for recommendations such as “your competitors are 20% more profitable than you are…”?
The COGS data will not be shared with other merchants / advertisers and is protected using the same security measures applied to all merchants’ GMC data.
We already enable merchants to share COGS data in Google Merchant center for reporting purposes. Using conversions with cart data, you can get additional reporting and insights, including profit reporting, by providing a value for the cost_of_goods_sold attribute under “Price & availability” in your GMC feed. More on that here.
Q6: Which of the announced features and functionalities will we be able to get access to first? Can you give us a rough idea of when the big ones will roll out (and which ones are already available and where)?
Here’s a rundown on some of the announcements:
- Ads in AI Overviews is starting soon with a small test in the U.S., English-only.
- We’re testing the dynamic experience with AI-generated recommendations in the U.S. — English-only to start. (The example shared during the keynote was a search for “short-term storage” and being able to click into an ad to upload photos of your rooms to get a recommendation on how much storage space you’ll need as well as packing supply quantities.)
- For PMax: brand guidelines is now in beta — English-only to start.
- Image references and image editing for PMax are rolling out globally in English.
- And additional reporting for PMax is coming soon, globally.
- For Shopping ads, 3D Spin is starting with shoes and will be available soon to all applicable advertisers. Virtual try on is starting with tops and will also be available to all applicable advertisers soon. Video highlights will be running in pilot to start.
- And Google Ads Data Manager and Measurement Diagnostics are now available globally.
- As mentioned earlier, Trusted Execution Environments (TEE) is expected to be available for advertisers over the course of this year.
Q7: Are there any plans to make offline conversion imports easier? Even the platforms with direct integrations (Hubspot, Salesforce) have issues in implementation and are not plug and play.
Google Ads Data Manager is now available for all advertisers. We are continuing to focus on making these processes easier for advertisers.
Q8: What Metric are RSA ads being optimized for? I’ve read as much public documentation as I could find and there doesn’t seem to be an answer.
Responsive search ads combinations are optimized based on relevancy and expected performance based on the query. (This is also why you may see some assets serve infrequently but have a high performance rating.) Performance is based on your stated campaign goals. This whitepaper on RSAs may also be helpful.
Q9: Are there plans to add targeting and exclusions to PMAX campaigns beyond location? We currently have several Lead-Gen clients and the initial conversion does not equal quality leads and we’re struggling with getting enough offline conversions imported.
Note that PMax also supports placement exclusions (YouTube videos are also now included in placement reporting, announced at GML) and brand exclusions as well as account-level keyword exclusions.
You may find this guide to driving high quality leads helpful. A few additional thoughts to increase lead data & improve lead quality:
- Enhanced conversions for leads can help capture additional conversions across Search & YouTube.
- If phone calls are part of the customer journey, tracking & importing call conversions can also help.
- With a value-based bidding strategy, you can assign values to the important conversion actions in your customer journey (e.g. lead, MQL, SQL, closed deal).
I also wrote this article on using value-based bidding. And I enjoyed this article on VBB for lead gen too.
Q10: Is documentation coming as to how to optimise our campaigns for a few of the AI functionalities – especially for AI overview and new Search experience [the self-storage ad example], where recommendations are given on the search page itself before a consumer clicks through to an ad?
There is no advertiser action needed for ads in AI overviews. Ads that serve within AI overviews will be Search and Shopping ads from existing campaigns (Search, Shopping and PMax). Relevancy will be based on both the query and the information in the AI Overview.
More to come on Search ads with AI-generated recommendations where we showed the storage unit example. Stay tuned.
Q11: Will broad match will have a leg up on other match types to serve in the AI overviews?
It is about having a relevant match to the query and the information in the AI Overview, not necessarily about match type. When used with Smart Bidding, broad match provides the greatest flexibility to match to relevant queries using additional signals, including the other keywords in the ad group, your landing page and user context. And by providing a diverse set of assets in your RSAs, you’ll give the system the flexibility to optimize your ads for the query coverage that broad match offers.
Q12: Will broad match campaigns or PMAX search themes will have priority?
Our keyword and search theme prioritization still applies:
- When you have the identical keyword to the search term (regardless of match type), it will be prioritized over Performance Max.
- If you have the identical PMax search theme only, it will be prioritized over Search campaigns.
- If you have the identical keyword and search theme to the query, they’ll have equal prioritization and compete based on Ad Rank.
Q13: Will we be able to build audiences based on interactions with assets rather than just the ad/initial search?
Not at this time.
Q14: Can we get an answer on the role of Ad Strength again, since it was brought up at GML and talked about as if it has performance importance?
Ad Strength for responsive search ads is not a factor in the auction. It really is a diagnostic tool. Why it is shown and emphasized is because it’s a reflection of four factors that we’ve seen correlate to an increase in performance on average (e.g. more conversions). As you build out or edit your ads, Ad Strength will adjust to reflect the relevancy and diversity of the assets you’ve added. We recommend having at least one ad with “good” or “excellent” Ad Strength per ad group.
When you use pinning, you will typically see your Ad Strength go down because pinning restricts the number of ad combinations that the system could show. That’s also why we recommend pinning two to three assets to a position. It’s important to understand why pinning is reflected in the rating, but it shouldn’t keep you from using or testing it.
When you have a diversity of assets, RSAs have more flexibility to be eligible to show for more, relevant queries that you may not have otherwise. That’s why you may see some assets not show often but have high performance ratings because they are relevant to a smaller audience subset.
Q15: If a pinned and unpinned ad run in tandem and the pinned ad shows a better conversion rate, will the Ad Strength reflect the reality of that improved conversion rate in the pinned or does it default to always favoring unpinned? Regardless of performance or brand?
Ad Strength is not a KPI and it doesn’t look at performance metrics. It’s a diagnostic tool indicating the diversity and relevancy of your assets and gets reported as you add and/or edit your assets. Ad Strength can also change over time according to factors such as keyword relevance.
The four factors Ad Strength looks at are: Number of headlines, Keyword relevance of headlines and descriptions, Uniqueness of headlines, Uniqueness of descriptions.
Will a high ad strength always beat out a lower ad strength ad? No. An ad with excellent Ad Strength is not guaranteed to be a high converter. Use Ad Strength as a guide like any other diagnostic tool. The important thing to using it is to understand what it reflects and why. I go deeper into this in this article on Ad Strength in Search Engine Journal for more.
Q16: Does “boring creative” have any advantage over testing more unique creative or phrasing – I’m asking about the quality of creative – for example “we’re excited to talk to you” vs “call us”?
Responsive search ads are designed to serve the ad combination expected to perform for a particular query. They optimize for relevance to the query and predicted outcomes based on your goal (e.g. the likelihood that the user who engages with the ad will convert), not based on whether the assets might be considered “boring” or “exciting.” (I’ve definitely experienced the letdown of seeing my “out-of-the-box” ad test underperform a straightforward version on Search — but don’t let it keep you from continuing to experiment to find what will engage the right people.) Keep testing and iterating on assets that will drive conversions or conversion value, depending on your goal. This whitepaper guide to RSAs dives deeper.
Q17: As specific conversion metrics were announced for Performance Max Assets, are there any plans to do the same for responsive search ads? If they are no plans to do the same, please may you explain why not?
I don’t have anything to share at this time but will pass this along to the team.
Keep in mind when looking at the performance labels of your RSA assets, that they reflect each asset’s performance relative to other assets of the same type within the RSA.
The rating shows what each asset could bring to the overall performance of your RSA, based on your advertiser goal. Assets that outperform other assets (i.e., are most effective at driving clicks and conversions) are rated “Best”.
Q18:If you have low rated assets, does the system ever try them again?
We recommend swapping out “low” rated assets to test new ones. It doesn’t mean that those assets may not work in another ad group with different keywords or landing page, but it means it was underperforming relative to other assets of the same type in that ad.
Q19: But does the system necessarily try to show the headline that is most related to the query itself?
It aims to show the combination that is predicted to perform best based on the query and your goals. More details on how RSAs work in this guide.
Q20: Would also love more info on whether stickers will be available for non shopping industries?
Stickers in Shorts ads are not limited to retail advertisers and will launch in direct response campaigns to start.
Q21: In terms of using AI generated assets, what guardrails are in place to protect brand voice, look and feel? example, one of my competitors is McDonalds, so how will the ai understand how our burger ad is different than a standard mcd burger ad?
AI generated asset features are purposely grounded in your existing assets in your account, including landing pages. The more that your existing asset reflect your USP and branding
Q22: Can you please ask about the “find exactly what I’m looking for” searches? Think it was pretty in the beginning of the presentation where they presented the stat about (I think) searches with 5 or more words grew 2.5x faster (again, can’t quite remember).
What would be the best course of action to optimize for these searches?
“Exactly-what-I-want” search was the first trend that Phillip Schindler discussed at GML: Over the past 6 months, the volume of searches with five or more words grew 1.5x as fast as shorter queries compared to the same time period last year.
It’s not going to be practical to try to cover all of these types of queries with individual keywords. Longer queries also tend to be fairly unique and lower volume, so there may not be a ton of visibility into them. Broad match is ideally suited to help you cast a wider net and match text ads to relevant long-tail searches without having to try to anticipate and build out extensive keyword lists. And looking back on the Ad Strength discussion, this is also where having a diversity of assets can provide the system with more flexibility to match to more specific and detailed searches.
Additionally, Performance Max can complement your Search campaigns. And Performance Max and Standard Shopping campaigns will be eligible to show Shopping ads in AI Overviews and in new Shopping ad placements at the top of visual search results.