The end of Dynamic Search Ads is here! Insights from 200+ AI Max campaigns

Episode Transcript

Mike 00:00:00
Welcome to another episode of Growing E-commerce. I am one of your hosts, Mike Ryan, and with me today is Chris Scharmüller. There was a rare departure and I won’t say you let us down, but it felt like you were gone for a while.

Chris 00:00:19
That’s why I’m back. Cheers, by the way. To the people listening, this is not a paid product promotion, but it is an Austrian product—one of the best products we have ever produced: Red Bull. Kevin, the producer, is probably wondering if he should edit this out. Anyway, Mike, we have a massive topic on our hands today: AI Max search. You did it again and broke the internet. What did you publish in our small bubble of the internet?

Mike 00:01:07
We’ve talked about AI Max on this podcast before, and I believe I teased that I’d be working on a larger report about it. So that’s what I published: an 18-page report called “The Ultimate Guide to AI Max.” It covers one year of AI Max for Google Search.

Chris 00:01:29
Don’t undersell yourself; it really is the ultimate guide. It reminds me of what everyone was calling their blog posts ten years ago for SEO purposes, but this is the real deal.

Mike 00:01:47
Believe it or not, it has been over a year since AI Max has been around. It was in a very small pilot phase back then called Search Max. Most advertisers started getting in touch with it around June or October of last year when it reached general availability.

Chris 00:02:13
For the listeners, maybe explain in ELI5 mode: what is AI Max search and what makes it special? Let’s set the stage before going into some super interesting findings.

Mike 00:02:41
AI Max is basically not a new campaign type; it is an expansion layer. It’s a combination of features you can activate on your classic keyword search campaigns with the intention of broader expansion. It uses a lot of broad match to increase your query coverage. Google says the goal is to generate incremental conversions—not necessarily in terms of marketing incrementality, but just additional volume on top of what your campaign was doing before. They also promise it will happen at a similar efficiency, which is a standard Google claim: percentage uplift at a similar CPA or ROAS.

Chris 00:03:26
Google claims to have the key to solving probably one of the most underlying business issues: the curve of diminishing returns. That is probably the reason why Google is such a successful company.

Mike 00:03:48
We can challenge that claim in a bit. First, regarding the characteristics that stand out, I made a table on page two of the report. It is a tabular overview of all AI Max features compared to PMax, Dynamic Search Ads (DSA), and broad match. The goal is to help you understand everything that’s inside of AI Max and where the overlaps or differences are from other technologies already running in your account.

Chris 00:04:32
One of the fair criticisms is to look at AI Max search objectively; not everything is gold. There are advantages, but you have to consider the overlap with existing technologies. Can you shed some light on that?

Mike 00:05:11
AI Max is in some ways just a repackaging of existing features, and even Google admits that. The official answer is that it’s a repackaging of broad match, DSA, and some things we know from PMax. They make a case that it’s more than the sum of its parts, but of the people using AI Max, almost all of them are already using broad match. The biggest value proposition should be for advertisers using exact and phrase match because there’s more room for Google to deliver expansion. We see a high overlap with DSA and PMax users; about 48% of advertisers using AI Max are also using DSA and PMax.

Chris 00:07:06
What does that mean in general? If Google handles this perfectly, I wouldn’t care, but there’s a theory that it does something to your CPC dynamics. Is this something you’re concerned about or that came out in your analysis? If I have three technologies running in parallel with the same core, am I driving up my CPC?

Mike 00:07:42
It’s tricky to know. I’m slightly less concerned about it here than with PMax and Shopping overlaps because PMax and Shopping build from the exact same feed and target the same landing page. In AI Max, conflicts are solved by Ad Rank, which looks at predicted CTR, ad quality, and landing page quality. Because these campaign types do more dynamically—customizing text and selecting landing pages on the fly—they can target the same query but have different outcomes. This mitigates the risk, but the concern remains because Google denies that self-competition ever happens, and I’m not 100% sure about that.

Chris 00:09:19
What is your take regarding this overlap beyond bidding?

Mike 00:09:39
These are automation-forward technologies, so you have multiple paths into the same actions, which could result in bidding problems. You don’t always know which campaign is going to serve and why. Serving for the same search term from different campaigns or technologies on different days is bad account hygiene. It makes reporting and analysis harder, and predictability is lost. Furthermore, Smart Bidding thrives on consolidated conversion volumes. By splitting conversion volume between different places unnecessarily, you might be limiting the velocity and volume that Smart Bidding needs to make the best decisions.

Chris 00:10:56
That makes sense. I had a think tank session at an event in London two weeks ago where I spoke to C-level executives. We talked about the future of search and the “power pack” of Google Demand Gen, PMax, and AI Max search. I showed them what each campaign type is good at. AI Max search is designed to drive incremental revenue, and for the C-level, that was a no-brainer. But then we discussed the downside: incrementality costs you. Google claims AI Max search drives incremental revenue at the same return levels. Is that true? Because if it is, it’s a no-brainer to activate.

Mike 00:12:32
Google specifically claims it generates 14% more conversions or conversion value. It’s a bit of a red flag to me that they suggest those two mechanics are interchangeable. For broad match, they claim it delivers more conversions and more conversion value due to average order value effects. They say your ROAS or CPA will be “similar,” which is a vague cushion. If you could really get 14% more conversions at the same efficiency, everyone would activate it right away.

Chris 00:13:15
How do we actually see it in the data? What are the real numbers?

Mike 00:14:03
I was suspicious of the claim because it doesn’t follow what we know about saturation curves and the law of diminishing returns; every extra dollar you spend eventually returns less than the dollar before. Keyword search is a mature technology, and I don’t believe most advertisers have huge unmet pockets of demand that can just be tapped efficiently. I looked at data from over 600 search accounts. Adoption is still only around 16%, so we’re looking at about 100 accounts. I found that it generates 13% more conversions, which is very comparable to Google’s claim. However, efficiency was not “given.” Cost per acquisition (CPA) was typically 16% higher. For an online shop in dire need of top-line growth, that might be acceptable, but it is a trade-off.

Chris 00:17:10
We talk to clients about AI Max search as an option to drive growth. If we follow Google’s playbook and expect 16% more revenue at the same price levels, those expectations are often not met, creating unnecessary friction. If they framed it closer to reality, this campaign type would be fine.

Mike 00:17:58
Happiness is expectations minus reality. Most of these advertisers are using “Maximize conversion value” with a ROAS target. Looking at ROAS, there was a 0% difference from the campaign average, which seems to support Google’s claim. But that hides the broad range of outcomes. The distribution is flat; it’s not a bell curve where everyone is at 0%. You have nearly equal odds of having a good or bad outcome, and it likely varies month to month.

Chris 00:19:16
The technology is good and drives incrementality, but Google should be closer to reality with their framing. It’s a nuanced message. As a client, I would ask: where is this incrementality coming from?

Mike 00:21:16
AI Max can find new queries and high-intent pockets of demand, specifically your brand traffic or competitive brand traffic. It has granular brand controls at the ad group level, which is better than what you have in Performance Max. One concern is that traffic might just be shifting between campaigns with little control. Is AI Max capturing stuff that was previously handled by PMax or DSA?

Chris 00:22:33
Those concerns are real. What is your take on brand-related and competitor-related search queries? If it drives incrementality, I’m fine, but is it really that easy?

Mike 00:23:09
It depends on your setup. If you aren’t targeting brand terms with broad match—perhaps using exact or phrase match for control—AI Max will jump in with its broad match component. You can control that, but you need to watch it. Regarding competitor terms, AI Max might treat a competitor term as synonymous with your brand or find competitor traffic that isn’t being met. Some people have “gentlemen’s agreements” not to bid on competitors to avoid a bidding war. You need to verify if AI Max is doing this and manage it accordingly.

Chris 00:25:38
Exactly. This is one of the most positive things about AI Max: potential shortcomings can be mitigated.

Mike 00:26:06
It’s not a perfect technology, but it’s a very good starting point compared to how PMax started. It’s a moving target, and Google will improve it. I have mixed feelings about cross-campaign interactions, but it is worth experimenting with.

Chris 00:26:29
The core is good, and you have controls to mitigate risk. It’s worth exploring, but not as a “one-click” set-and-forget experiment. Advertisers with more basic search builds might see even stronger effects. PMax search did a lot for clients who didn’t have the most advanced setups.

Chris 00:27:55
Mike, given the time is flying, one last thing from your document: the overlap with Dynamic Search Ads. You said they are going the way of the dinosaurs.

Mike 00:28:17
Google has confirmed that Dynamic Search Ads will be sunset. We don’t have an exact timeline, but Google has described it as a long-term goal. If you are a heavy DSA user, AI Max is the perfect place to start your migration. You’ve been able to voluntarily migrate DSA to PMax for a while, and Google might eventually prioritize AI Max more heavily.

Chris 00:29:34
There’s no official timeline, but Google has ways to make it a less important campaign type to push the “power pack” of PMax and AI Max search.

Mike 00:30:18
In the last Google earnings call, AI Max wasn’t very present, but it was highly present in the one before. When the executives talk about it to Wall Street, it’s a sign. In Dublin, executives told us this is the future.

Mike 00:31:04
Please go read the full report. It exists as a long-form article and a PDF, which I consider the canonical version. I also published a script to help you get better AI Max reports. All three pieces are free.

Chris 00:31:48
Ladies and gents, I read the long copy after a day in London and I didn’t fall asleep at all. It’s a great article with a lot of findings. Shout out to the team.

Mike 00:32:07
The team worked super hard on this. I reject the philosophy of writing at a fifth-grade level; I think people like to read real sentences.

Chris 00:32:29
This level of content makes me proud to be part of this company.

Mike 00:32:45
Great job. This has been another episode of Growing E-commerce brought to you by Smarter Ecommerce, also known as smec. Learn more at smarter-ecommerce.com. Please leave us a review or check out “The Ultimate Guide to AI Max.”

Chris 00:33:03
Thank you. See you next time.