On the back of an extremely fast-moving quarter in the Google Ads world (aren’t they all?) here are the standout updates from Q2 – and why they matter.
The Big Picture (i.e. A.I.)
Last quarter brought some stark clues about where AI may take advertising next, not least from Meta, who laid out their vision for AI to take over the entire ad lifecycle.
As Mark Zuckerberg put it in his interview with Stratechery in May:
“You’re a business, you come to us, you tell us what your objective is, you connect to your bank account, you don’t need any creative, you don’t need any targeting demographic… [you just pay for ] as many business outcomes as you can achieve”
The timeline for this advertising Skynet? 2026!
While all of the smart money is surely on the continued increase in automation across the industry, Google meanwhile does not seem to be charting quite the same course.
Right now in fact they’re enjoying praise from the marketing community for the unexpected increase in both visibility and control over their systems in Q2, particularly for Performance Max which – in quick succession – has now received campaign-negative keywords, channel performance reports, ‘real’ search term data, and age and device exclusions.
The rise of AI is no longer news, but what it means for advertising – and advertisers – is still very much to be determined. Meta’s all-in bet is not the only path available, and Google may well be watching it unfold with interest before deciding whether to follow or differentiate.
Key Google Ads Updates
1. 𝗣𝗠𝗮𝘅 𝗖𝗵𝗮𝗻𝗻𝗲𝗹 𝗣𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗿𝗲𝗽𝗼𝗿𝘁𝗶𝗻𝗴
Rolling out in Q2, this long-awaited feature shows exactly how your Performance Max campaign budget is split across channels, presented in both table and (visually pleasing) Sankey form, with cost, clicks, conversions, and value all broken out.
While it would be nice to have the accompanying control over which channels to enable (a control brought in for Demand Gen in Q1) there’s plenty of channel-specific work that can be guided by the information in the report.
The latest unit of my course - 'refining and expanding PMax’ - goes through all the new levers and how to get the most out of them.
In truth, we have had similar detail for two years already, courtesy of groundbreaking script work by Mike Rhodes – but it’s great to have it natively in platform – both for the convenience, and for what it implies about Google’s current alignment with its advertisers (Meta take note).
2. Accessible Incrementality Testing
True incrementality testing – once reserved for the highest-spending advertisers – is now being opened up more widely.
Thanks to a new Bayesian methodology, advertisers can now run tests with as little as $5K in budget, across all campaign types.
Many assumptions about the value of branding campaigns on one end, and brand search campaigns on the other, should soon be fleshed out with valuable data.
3 GA4 Improvements
Q2 also brought the announcement of a suite of updates to GA4 – the highlight being a revival of the ‘Assisted Conversions’ report (one of the most useful in Universal analytics).
As Google Ads Liaison Ginny Marvin promises: “The Assisted Conversions reporting that you loved in UA is coming to GA4 – but better. Get clearer insights into which channels and campaigns influenced conversions in easy-to-read out-of-the-box reporting.”
GA4 will also soon have a new AI chat agent to help find the reports we need… (a notoriously harder task than it used to be in the days of UA).
Very positive updates! GA4 has a long road to redemption – but every step helps.
4. Agentic tools
AI assistance within Google Ads continues to evolve. Building on the “conversational experience” launched two years ago, Google is now introducing more advanced capabilities – described as an “agentic expert” that can offer personalised recommendations and even implement them on your behalf.
This promises to be a more fully-fledged AI assistant – Google cites its ability to build “multiple themed ad groups with matching assets” – and may bridge the gap between the more limited (and too often ‘quantity-over-quality’) native recommendations we have had up to now, and the freer analysis and creation for which many of us use external AI tools.
For now, these are touted as suggestions that can then be implemented rather than autonomous account-managing bots – but no doubt we’ll be able to take the safety off soon.
Google’s recommendations – especially in-platform – have never yet shaken off the need for a health warning, so expanding their capabilities is not without risk… But these agents could be extremely useful… if not too tightly bound to the party line.
5. AI Max / Mode / Overviews
In a quick rebrand of “Search Max”, AI Max rolled out in the latter part of the quarter.
This optional toggle for Search campaigns offers increased reach – with DSA-style ‘keywordless targeting’ and auto-generation of assets (with both elements independently togglable).
While it is – mostly – a repackaging of existing AI features, it has sparked a fair bit of excitement in the community – particularly around the prospect of gaining access to AI Overviews (and its big brother, AI Mode), which are also accessible via PMax, Broad Match and Demand Gen.
In my view, this will become important inventory – but it isn’t, yet.
There are still plenty of search impressions to go around, and no evidence that AI Overview placements are especially valuable compared to traditional search. (Nor should we expect them to be, given that AI Overviews are typically triggered – and read – in response to more informational queries.)
That hypothesis, though, should now be testable, with a new ‘AI Max’ entry to the ‘search term match type’ dimension in reporting, making it easier to evaluate the incremental activity from its expanded reach.
The search term match type dimension is great (and under-appreciated) for evaluating match type performance more deeply than we can by comparing keywords alone – breaking down results by the matching rules under which ads were triggered (exact, phrase or broad).
When analysed – intuitively enough – it often shows ROAS decreasing as targeting expands beyond the core search term.
With AI Max enabled, a fourth ‘match type’ (‘AI Max’) appears. We may well see ROAS decline further as we move beyond broad match into this keywordless territory – but with this reporting feature, we can isolate the incremental reach, and put that assumption to the test.
When either the quality or the scale of ad placements among AI-generated answers reaches a point where it meaningfully/positively impacts results, that’s when it becomes worth pursuing.
For now the calculation is similar to using broad match or DSA: If volume is the priority, and you need more inventory to generate it, it’s an interesting option – but it’s not generally the most cost-effective one.
In short: put your ads where the best inventory is. Not the shiniest.
Want to stay ahead of these changes – and actually use them to grow accounts? We dive deep into how to get the most out of every feature in the Google Ads Level Up membership.
