Before you turn on AI Max, fix these things first
AI Max is not the first thing we would fix in most Google Ads accounts.
It is one of the last things we would test after the account is clean enough to handle more automation.
That may sound cautious, but we have seen what happens when automation is switched on before the foundation is ready. In one bathroom and kitchen renovation account, AI Max started pulling ridiculous queries like “Bathroom” and “Bathroom near me.” The business was not trying to pay for vague category curiosity. It needed serious renovation inquiries from people looking for premium remodeling work.
That is the difference between automation helping and automation wandering.
Before turning on AI Max, a business should fix:
- Conversion tracking
- Lead-quality feedback
- Landing page relevance
- Offer clarity
- Search term hygiene
- Negative keyword systems
- Brand control
- URL control
- Campaign structure
- Budget and bid guardrails
If those are weak, AI Max can give Google more room to make decisions from bad inputs.
That is the core warning of this article.
AI Max can be useful.
AI Max can also be expensive garbage if the account is not ready.
This guide explains what AI Max is, why Google introduced it, how it works, where it can help, where it can go wrong, what early results are showing, and what we would audit before trusting it with more budget.
We are writing this for business owners, marketing managers, and agencies who do not want to be left behind by Google Ads automation, but also do not want to hand the steering wheel to a system that has not earned trust yet.
This article is based on four inputs.
First, our own account experience at River Stone, including ecommerce, local lead generation, renovation, interior design, dentists, SaaS, and WordPress product accounts.
Second, the foundation rules we use for Google Ads thinking: offer, landing page, unit economics, goals, conversion tracking, bids and budgets, campaign structure, creatives, and targeting. AI Max touches almost all of those pillars.
Third, Google’s current AI Max documentation and announcements.
Fourth, current third-party PPC testing and commentary, including practitioner reports where AI Max produced both extra volume and irrelevant search terms.
Google’s own content guidance says helpful content should provide original information, reporting, research, or analysis, and should demonstrate first-hand expertise and depth of knowledge.
That is the standard we are trying to meet here: not “AI Max is good” or “AI Max is bad,” but “here is what we would check before we let it spend.”
The short answer: should businesses turn on AI Max?
Sometimes.
But not blindly.
AI Max is best treated like a controlled experiment, not a default upgrade.
We would consider testing it when:
- Conversion tracking is accurate
- Primary conversions represent real business value
- Search terms are already clean
- Negative keyword lists are mature
- Landing pages are strong
- The website has relevant pages AI Max can use
- Brand exclusions or inclusions are planned
- Final URL expansion is controlled
- The account has enough conversion volume for Smart Bidding
- The business can judge lead quality after the click
- Someone will review AI Max reporting every 1-2 weeks
We would be cautious when:
- The business has a small daily budget
- CPCs are high
- Lead quality is already poor
- Conversion tracking counts weak actions
- The website has many unrelated or thin pages
- The offer is vague
- The account has brand/non-brand confusion
- Search term review is not happening
- The business serves only narrow locations or premium jobs
- The sales team cannot report which leads were qualified
AI Max is not the problem by itself.
The problem is turning on AI Max in an account that cannot tell Google what a good customer looks like.
1. Why Google introduced AI Max
AI Max did not come out of nowhere.
It is part of a long automation arc inside Google Ads.
Google Ads has been moving away from exact manual control and toward machine-learning systems for years:
- Smart Bidding
- Responsive Search Ads
- Broad match paired with Smart Bidding
- Dynamic Search Ads
- Optimized targeting
- Performance Max
- Automatically created assets
- Final URL expansion
- Search themes
- AI Max for Search
- AI Max for Shopping
The direction is obvious: Google wants advertisers to provide business inputs, creative assets, landing pages, goals, conversion data, and guardrails, while Google’s systems find more demand across more complex search behavior.
There is a real reason for this.
Search behavior is changing.
People are not only typing short keywords like “roofing contractor” or “interior designer.” They are asking longer, more specific, more conversational, more contextual questions. They compare, describe problems, include locations, reference brands, and ask for solutions in ways old keyword lists may not fully cover.
Google’s AI Max announcement says it brings “targeting and creative enhancements” into Search campaigns and helps advertisers expand beyond their keyword lists with broad match and keywordless technology.
Google also says AI Max is designed for “the next era of Search,” where search is more conversational, visual, and intent-driven.
That is the strategic reason.
The business reason is also clear.
Google wants more advertisers using more automation across Search, Shopping, and cross-channel campaigns. Performance Max already moved many advertisers into more automated campaign management. AI Max brings that kind of expansion mindset deeper into Search.
Google says Dynamic Search Ads are moving toward AI Max. In April 2026, Google announced that DSA, automatically created assets, and campaign-level broad match settings would be upgraded into AI Max starting September 2026. Later, Google extended the DSA auto-migration timeline to February 2027, while ACA and campaign-level broad match upgrades remain part of the transition.
In plain English:
AI Max is not a small feature.
It is Google’s next major step in automating Search.
2. What AI Max actually is
AI Max is not a new campaign type.
Google’s developer documentation describes AI Max for Search campaigns as “an optimization layer” for existing Search campaigns.
That distinction matters.
You are not creating a totally separate AI Max campaign the way you create a Performance Max campaign. You are enabling AI-powered features inside Search campaigns.
The main AI Max features are:
Search term matching
Search term matching expands beyond your exact keyword list by using broad match, asset-based signals, landing page signals, and keywordless technology.
Google’s AI Max help documentation says search term matching uses “broad match, asset-based and landing page-based technology” to optimize campaign reach.
This is the feature that can find incremental searches.
It is also the feature that can bring in strange, vague, or low-fit searches if the account is not controlled.
Text customization
Text customization is the newer version of automatically created assets.
It uses your existing ads, landing page copy, assets, and generative AI to create more tailored headlines and descriptions.
This can help ad relevance.
It can also create messaging risk if the business has strict claims, compliance needs, premium positioning, legal language, medical language, or brand standards.
Google has introduced text guidelines to help advertisers steer AI-generated messaging. Google says these guidelines allow advertisers to define exclusions and restrictions such as “don’t imply our products are cheap.”
That is useful.
But it also proves the point: if AI can generate messaging, you need controls.
Final URL expansion
Final URL expansion lets Google send users to the page on your domain that it predicts will perform best for the search.
Google says Final URL expansion can send traffic to the most relevant URLs on your website, and that it requires text customization so ad copy can match the landing page.
This is powerful when the website has strong, relevant pages.
It is dangerous when the website has weak pages, old blog posts, unrelated service pages, thin pages, outdated pages, or pages that should never receive paid traffic.
Brand settings
AI Max includes brand inclusions and exclusions so advertisers can guide which brands ads should or should not appear with.
This is important for brand safety, competitor strategy, and brand/non-brand reporting.
URL inclusions and exclusions
AI Max allows advertisers to include or exclude URLs so the system knows which pages it can use.
This is one of the most important controls.
If you do not want traffic going to a blog post, old service page, low-intent informational page, hiring page, pricing page, or unrelated page, you need to control the URLs.
Reporting
Google says AI Max reporting includes search terms, keyword report rows, landing page reporting, asset reporting, and a view that combines search terms, headlines, and URLs.
This is better than the old black-box feeling, but it still requires someone to actually review the data.
Reporting is not control unless someone uses it.
3. How AI Max works in practice
The cleanest way to understand AI Max is this:
AI Max looks at more than your keywords.
It looks at the meaning of searches, your existing keywords, your ads, your landing pages, your assets, your domain, your conversion data, and your campaign goals. Then it tries to match users to ads and pages that are likely to perform.
That means the inputs matter enormously.
If your landing pages are strong, AI Max has better pages to choose from.
If your landing pages are messy, AI Max has messy options.
If your conversion tracking is clean, Smart Bidding has cleaner learning.
If your conversion tracking counts junk leads, AI Max can help create more junk.
If your offer is clear, AI-generated text has stronger source material.
If your offer is vague, generated text may become vague too.
If your website has well-built service pages, location pages, category pages, and buying-intent pages, AI Max may find useful paths.
If your website has thin content, old posts, weak pages, or mixed intent, Final URL expansion can get risky.
This is why PPC Mastery fundamentals are so relevant here.
AI Max touches the same foundations:
- Offer
- Landing page
- Unit economics
- Goals and KPIs
- Conversion tracking
- Bids and budgets
- Campaign structure
- Creatives
- Targeting
Performance Max already taught advertisers this lesson. Automation works better when the fundamentals are strong. We would like to describes Performance Max as the ultimate test of your fundamental understanding of Google Ads because it combines automation across bidding, creative, targeting, and placement. AI Max brings a similar test into Search.
Our view is simple:
AI Max is not a replacement for strategy.
It is a stress test of your strategy.
4. AI Max can be good
AI Max can be useful in the right account.
It can help when keyword coverage is too narrow.
It can find long-tail queries humans missed.
It can adapt to conversational search behavior.
It can match ads to more specific user intent.
It can route users to more relevant pages.
It can reduce manual maintenance in large accounts.
It can help advertisers move beyond old keyword-only thinking.
It can reveal search paths that should become new campaigns, landing pages, or content ideas.
Google says AI Max campaigns already using search term matching see an average of 7% more conversions or conversion value at a similar CPA/ROAS when enabling text customization and final URL expansion.
Google’s DSA-to-AI Max announcement uses the same 7% figure for the full feature suite compared with search term matching alone.
Google also says brands using text guidelines in AI Max have seen gains. In one example, BYD increased leads by 24% at a 26% lower cost while using text guidelines to protect brand standards.
Those results are worth paying attention to.
But we would not treat them as permission to switch AI Max on everywhere.
Google’s averages are not your account.
Your account has its own tracking quality, budget constraints, search terms, landing pages, brand issues, sales process, and economics.
AI Max may work beautifully for one advertiser and create garbage for another.
5. AI Max can be bad
AI Max can be bad when it expands into the wrong intent.
This is the risk most business owners need to understand.
The account may not look broken at first. Spend may increase. Impressions may increase. Clicks may increase. Conversions may even increase.
But the business has to ask:
Are these the conversions we actually want?
In our renovation example, the client wanted premium bathroom and kitchen remodeling inquiries. When AI Max started showing queries like “Bathroom” and “Bathroom near me,” that was a clear warning sign.
“Bathroom” is not enough intent.
It could mean:
- Bathroom ideas
- Bathroom accessories
- Bathroom stores
- Bathroom cleaning
- Bathroom repair
- Bathroom design inspiration
- Bathroom remodel cost
- Bathroom showroom
- Bathroom near me
- Bathroom contractor
Some of those might be useful. Many are not.
For a premium renovation company, vague searches can burn budget quickly because renovation CPCs are not cheap and the sales cycle is high-consideration.
This is the same lesson we learned from broad match and Performance Max:
Google can find more volume.
The business has to prove the volume is worth paying for.
Third-party testing is showing the same mixed picture. Fraud Blocker reported turning on AI Max for two non-brand Search campaigns and seeing daily spend jump nearly 5x within days. They said the system found new volume, but also brought in irrelevant searches such as “ad blocker,” “web scraper,” and “como crear un bot.”
That is exactly the concern.
AI Max can expand.
Expansion is not the same as profitable expansion.
PPC Live’s practitioner guide puts the concern well: Google’s 7% lift is an average, not a promise that “switch it on and watch your account roaring.”
That is how we would explain it to a client.
AI Max is not bad.
Blind AI Max is bad.
6. Initial results: useful signals and garbage signals
The early results around AI Max seem to fall into two buckets.
The useful side
AI Max can find incremental demand.
It can show what users are asking that your keyword list did not include.
It can show which landing pages Google thinks match search intent.
It can reveal headline and landing page combinations that work.
It can help large sites with many relevant pages expand beyond manual keyword maintenance.
It can make sense for advertisers who already have strong conversion tracking and enough data.
It can work well when the business has broad but commercially relevant demand.
The garbage side
AI Max can also bring in weak, vague, or low-fit searches.
It can route traffic to pages you would not have chosen.
It can generate messaging that needs review.
It can blur brand and non-brand learning if settings are not controlled.
It can increase spend faster than the business can judge quality.
It can make lead gen accounts worse if form fills are treated as equal conversions.
It can be especially risky for small-budget accounts where a few wrong clicks matter.
Google’s AI Max reporting documentation recommends reviewing search terms every 1-2 weeks and using the search terms and landing pages view to understand search terms, headlines, and landing pages.
We agree with the need for review.
But in early tests, we would often review more aggressively than every 1-2 weeks, especially for high-CPC lead gen. If the daily budget is small or CPC is high, waiting two weeks to discover junk can be expensive.
7. Research: where AI Max appears to be working
AI Max seems most promising when the account has:
- Strong conversion tracking
- Clear business goals
- Enough conversion volume
- Strong landing pages
- A clean website structure
- Commercially relevant long-tail demand
- Strong URL controls
- Brand and competitor controls
- Someone actively reviewing reports
Google’s own case-style examples show performance gains for advertisers using AI Max features. The most specific number from Google is the average 7% more conversions or conversion value at similar CPA/ROAS when enabling text customization and Final URL expansion in campaigns already using search term matching.
Google’s text guidelines announcement also cites BYD improving leads by 24% at 26% lower cost while using AI Max controls for brand standards.
Bizi Digital’s 2026 AI Max review takes a balanced stance: AI Max is real, the migration is coming, and the right accounts should test it, but “turning it on blind will cost you.”
Fraud Blocker’s test shows the other side: AI Max created more volume but pulled clearly irrelevant terms in their account.
So the honest answer is:
Yes, AI Max appears to work somewhere.
No, that does not mean it will work everywhere.
The best accounts for AI Max are probably not the accounts looking for a magic switch.
They are the accounts that already have enough discipline to control the system.
What deeper PPC research says about AI Max
The deeper research is more interesting than Google’s headline numbers.
Google’s own story is that AI Max helps advertisers expand beyond manual keyword limits while keeping more reporting and control than older automation products. But independent practitioner research shows a messier picture: AI Max can improve relevance and performance in mature accounts, but it can also dilute intent, inflate spend, and make brand/non-brand reporting harder to trust.
Search Engine Land reported on 23 AI Max tests across 16 mature advertisers. The authors found that text customization improved weighted Quality Score from 6.8 to 7.3, with ad relevance showing the strongest lift.
That is a meaningful signal.
It suggests AI Max can improve ad-to-query relevance when the account already has strong inputs. Better relevance can help the auction, improve user experience, and potentially lower friction between search intent, ad copy, and landing page.
But that same research context matters. These were mature advertisers. The test was not “turn this on in a broken local lead gen account and watch magic happen.” It was structured testing in accounts that already had enough history and discipline to make the feature testable.
Another Search Engine Land piece covering SMX Next summarized the practitioner view clearly: AI Max and Performance Max can find new opportunities faster, but human strategy still decides where automation fits. The article notes that AI Max is not a new campaign type, but an opt-in setting within existing Search campaigns that expands keywords using landing pages and site assets, then personalizes ad copy and landing page selection.
That is exactly why we keep coming back to fundamentals.
If AI Max uses landing pages and site assets, then landing pages and site assets become targeting inputs.
If AI Max personalizes ad copy, then brand messaging and offer clarity become automation inputs.
If AI Max expands beyond keywords, then search term review and negative keyword discipline become quality-control inputs.
Search Engine Land’s brand-campaign analysis is even more direct. It warns that AI Max is often pitched before advertisers fix “measurement, conversion, and account structure issues.” It also notes that independent testing is mixed: Smarter Ecommerce reportedly found AI Max delivered 35% lower ROAS than traditional match types across 600 accounts; Xavier Mantica found AI Max had the most expensive conversions in a four-month test; and Ezra Sackett tested 30,000 search terms and found 99% of impressions delivered zero conversions.
That does not mean AI Max is useless.
It means the average business should stop treating Google’s recommendation as the same thing as a strategy.
One Search Engine Land audit article makes the operational issue very clear: AI Max has increased the volume of irrelevant queries flowing through some campaigns, including queries with no connection to the advertiser’s business or keywords.
That is not theoretical for us.
That is what our renovation example looked like.
The business wanted premium kitchen and bathroom renovation opportunities. AI Max brought in searches like “Bathroom” and “Bathroom near me.” Those searches may look adjacent to the category, but they do not prove remodeling intent, project size, budget, urgency, or fit.
Another Search Engine Land article on when to trust Google Ads AI gives a useful rule: test AI Max in a new campaign first, and do not apply it immediately to your most profitable, proven campaigns because it fundamentally changes how keywords work.
That is practical advice.
We would rather test AI Max in a controlled expansion environment than switch it on inside a campaign that already works and hope the machine does not disturb a profitable system.
Search Engine Land’s earlier 30-day testing article landed in the middle: AI Max is not PPC heaven and not PPC disaster. It expands Search targeting beyond keywords and can deliver results, but context matters.
That is also our position.
AI Max is not the enemy.
Unprepared AI Max is the enemy.
What experts are really saying beneath the headlines
When you put the research together, the expert consensus is not “never use AI Max.”
It is closer to this:
Use AI Max when the account has enough signal quality, structure, and operational discipline to manage expansion.
Do not use AI Max as a shortcut around bad fundamentals.
Here are the patterns we see across the research:
- AI Max can improve relevance when inputs are strong.
- AI Max can expand into irrelevant queries when intent controls are weak.
- AI Max should be tested, not blindly applied.
- AI Max is risky in brand campaigns if the goal is defensive control.
- AI Max reporting is better than old black-box automation, but only if someone reviews it.
- AI Max can distort performance if brand, non-brand, and expansion traffic are not separated.
- AI Max depends heavily on landing pages, creative assets, and conversion signals.
- AI Max may work better in mature accounts than in messy or low-budget accounts.
This is why a serious AI Max article cannot stop at “Google says 7% more conversions.”
The real question is:
What kind of account produced the gain?
What conversion actions were used?
Was brand included?
Were landing pages controlled?
Were weak search terms excluded?
Was the advertiser optimizing for lead volume or lead quality?
Was the business measuring qualified opportunities or only form fills?
Was the account mature enough for automation?
Without those answers, the headline result is not enough.
What Reddit and LinkedIn are saying
Official documentation tells us what AI Max is designed to do.
Practitioner communities tell us where it gets messy.
We do not treat Reddit and LinkedIn comments as controlled studies. They are anecdotal. But they are still useful because PPC managers often report the problems they see before polished case studies are published.
The community signal is consistent:
AI Max can work in accounts with strong data, but it can become a budget leak in accounts with thin signals, niche intent, weak conversion tracking, or small budgets.
Brightsprout reviewed Reddit discussion around AI Max and found a mixed picture: some advertisers saw international search behavior uncovered by AI Max, but others reported top-of-funnel “idea” and “inspiration” traffic that did not convert. One Reddit-sourced example quoted in the article said AI Max produced so much irrelevant traffic that it was turned off after one day.
MediaPost also summarized Reddit user reports and cited a severe mismatch example: an appliance repair campaign reportedly served for the query “obgyn near me.”
That example may sound absurd, but it matches the kind of problem we care about:
The machine can see relationship signals that are not commercially useful.
In our renovation account, “Bathroom” and “Bathroom near me” were not as extreme as “obgyn near me,” but the underlying issue was the same. The system expanded toward terms that looked category-adjacent but did not prove serious renovation intent.
LinkedIn is showing the same split.
On a Google Ads LinkedIn post announcing AI Max upgrades, Google positioned AI Max as a way to adapt to new Search behavior and find untapped queries. In the comments, practitioners immediately raised the control issue. One commenter wrote that AI Max works when “guided, not trusted blindly.” Another summarized the strategic question well: the issue is not simply “Should you switch to AI Max?” but whether your inputs are strong enough to make it work.
Ginny Marvin, Google’s Ads Product Liaison, also posted that AI Max became the default setting when creating a Search campaign, while noting that Google had extended the DSA auto-migration deadline to February 2027 and still recommended manual upgrade tools for oversight and control.
That matters.
If AI Max is becoming a default or migration path, the question is no longer whether advertisers can ignore it forever.
The question is whether they prepare before Google pushes more account behavior into that system.
Some LinkedIn practitioners are positive but conditional. Josh Blackman wrote that AI Max can make sense when a campaign has 100+ conversions per month, uses Target CPA or Target ROAS, has already proven broad match, is not limited by budget, and has weekly search term review.
That is a high bar for many local businesses.
It is especially high for under-$50/day accounts, high-CPC local services, and niche lead gen campaigns where qualified conversion volume is already limited.
John Aspinall made a similar point on LinkedIn: accounts with thin conversion data can turn AI Max into a broad-match-like expansion that burns through test budgets, and the question is whether the account has enough down-funnel conversion data for Google to know what good looks like.
That is exactly our view for lead generation.
If the account only tracks form fills, it is not feeding AI Max down-funnel truth.
If it tracks qualified leads, booked appointments, and closed deals, the test becomes much more intelligent.
Dave Chaffey shared Xavier Mantica’s four-month AI Max comparison on LinkedIn, where AI Max came in at $100.37 per conversion versus $43.97 for phrase match and $52.69 for exact match. The post framed AI Max as a cautionary tale and warned not to be seduced by the AI label just because Google makes it easy to select.
Thomas Eccel’s LinkedIn discussion added nuance: he reported competitor and irrelevant terms in both tested accounts, but noted AI Max terms were more useful in the account with more data. His practical conclusion was that AI Max is more useful for scaling existing campaigns that already have a solid base.
John Malefakis shared another controlled testing view on LinkedIn: across Travel, ecommerce, and lead gen brands, AI Max produced anywhere from 6% to 33% of clicks compared with standard match types, but some queries were broader and more loosely related. He also reported 20% to 40% lower CTR and 15% to 50% lower ROAS in some cases, while still saying AI Max can have a place in mature accounts as part of a controlled structure.
This is the theme:
AI Max is not a beginner shortcut.
It is an advanced expansion tool pretending to be a simple toggle.
What the community signal changes
The Reddit and LinkedIn layer makes our recommendation more conservative.
If only Google data existed, we might say:
“Test AI Max because Google is showing conversion lift and the migration path is clear.”
After looking at Reddit, LinkedIn, and practitioner testing, we would say:
“Test AI Max only after you define what failure looks like.”
Before a test, decide:
- What is the acceptable CPA or ROAS?
- What is the acceptable qualified lead rate?
- What search terms would immediately concern us?
- What URLs should never receive paid traffic?
- What brand terms must be protected?
- How much budget can we spend before evaluating?
- How often will search terms be reviewed?
- What would make us turn it off?
For River Stone, the “Bathroom” and “Bathroom near me” example is not a small annoyance. It is the kind of search-term drift that tells us the test needs tighter controls.
AI Max can absolutely discover useful long-tail searches.
But a serious advertiser should not confuse discovery with permission.
Discovery needs review.
Review needs action.
Action needs business judgment.
How this changes our AI Max recommendation
The deeper research makes our recommendation stricter, not looser.
We would not say:
“AI Max is trending, so turn it on.”
We would say:
“AI Max is becoming part of the future of Search, so prepare your account to test it safely.”
That preparation matters because Google is moving advertisers in this direction anyway. Search Engine Land reported that Google is retiring legacy Search automation tools like Dynamic Search Ads in favor of AI Max, and that campaigns using DSA, ACA, or campaign-level broad match settings are part of the migration path.
Google later extended the DSA auto-migration timeline, but the direction is still clear.
The old world was keyword-first.
The new world is signal-first.
Keywords still matter. In fact, Search Engine Land quoted a Google spokesperson saying keywords remain an essential component because they provide “fuel” for AI and intent signals.
That is the right way to think about it.
Keywords are not dead.
They are training signals, control signals, reporting signals, and intent anchors.
AI Max does not remove the need for keyword strategy. It changes the role of keyword strategy.
What businesses should fix before turning on AI Max
Now let’s come back to the most important part.
If a business asked us, “Should we turn on AI Max?” we would not answer until we reviewed the account.
Here is what we would fix first.
Fix 1: Conversion tracking
Do not turn on AI Max if your conversion tracking is weak.
Google’s Smart Bidding and automation systems rely on conversion data. Google’s bidding documentation says, “Implementing a bid strategy with a solid foundation of conversion data will help drive results faster.”
That line should be printed above every Google Ads account.
AI Max can expand reach, tailor ads, and send people to different pages. But if the account is optimizing toward weak conversions, the system can learn the wrong lesson.
Before AI Max, check:
- Are form submissions tracked correctly?
- Are phone calls real calls or just clicks?
- Are duplicate conversions removed?
- Are primary and secondary conversions separated?
- Are qualified leads imported?
- Are booked appointments tracked?
- Are closed deals uploaded where possible?
- Is ecommerce revenue accurate?
- Are enhanced conversions configured where appropriate?
- Are low-quality micro actions excluded from bidding?
For lead generation, this is non-negotiable.
If Google only sees “lead submitted” and not “lead qualified,” AI Max may help create more form fills that sales does not want.
Fix 2: Lead-quality feedback
AI Max needs a business-quality signal, not just a website-action signal.
This matters for renovation, roofing, dentists, interior design, SaaS, and local lead generation.
The system needs to know which leads were real.
Not all inquiries are equal.
A premium renovation lead is not the same as someone searching “bathroom near me” with no clear project intent.
A roofing replacement lead is not the same as someone asking for a cheap patch.
A dental implant consultation is not the same as a free-care search.
A SaaS demo request from an enterprise prospect is not the same as a student asking for a template.
Before AI Max, define:
- What is a qualified lead?
- Which leads should count for bidding?
- Which leads should be ignored?
- Which lead stages should be uploaded?
- Which lead values should be assigned?
- Who will review lead quality?
- How quickly will sales feedback return to marketing?
If the business cannot answer those questions, AI Max may optimize toward the easiest conversions, not the best ones.
Fix 3: Landing pages
Final URL expansion means Google can choose different pages on your site.
That is only good if your pages are worth choosing.
Before AI Max, review the website.
Ask:
- Which pages should paid traffic be allowed to visit?
- Which pages should be excluded?
- Are service pages strong?
- Are location pages strong?
- Are category pages strong?
- Are old blog posts safe for paid traffic?
- Are hiring pages excluded?
- Are outdated pages excluded?
- Are low-intent pages excluded?
- Are irrelevant pages excluded?
- Are pages matched to ad group themes?
Google says AI Max’s Final URL expansion can send traffic to relevant URLs on your website when likely to perform better.
That sounds good.
But “relevant” has to be checked from the business perspective.
A page can be topically relevant and still commercially weak.
For our renovation example, a vague bathroom page or inspiration page would not be the same as a high-intent bathroom remodeling consultation page.
Fix 4: Offer clarity
AI Max uses your ads, landing pages, and assets to create more tailored messaging.
If the offer is vague, the AI has weak material.
Before AI Max, sharpen the offer.
For lead gen, do not rely only on:
- Free quote
- Contact us
- Book now
- Trusted experts
- Quality service
Explain:
- Who the offer is for
- What problem it solves
- What makes it different
- What the next step includes
- Why the business is trustworthy
- What kind of customer is a good fit
- What happens after the inquiry
For premium renovation, the offer should not sound like cheap handyman work.
For interior design, the offer should help serious homeowners understand the process, fit, portfolio, consultation, and transparency.
For ecommerce, the offer should include product difference, shipping, reviews, guarantee, pricing, and value.
AI Max cannot create a great offer from thin positioning.
Fix 5: Search term hygiene
Before AI Max, the existing search terms should already be reviewed.
If the current account is full of irrelevant queries, AI Max is not the first fix.
It may expand the problem.
Review:
- Vague terms
- Research terms
- Job terms
- Course terms
- Free terms
- DIY terms
- Low-price terms
- Wrong-location terms
- Competitor terms
- Brand leakage
- Informational searches
- Service mismatch
- Product mismatch
Google’s AI Max reporting lets advertisers review search terms and landing pages from AI Max, including headlines and URLs.
That reporting is useful only if someone uses it.
We would not turn on AI Max and wait casually.
We would watch it.
Fix 6: Negative keyword systems
AI Max does not remove the need for negative keywords.
It increases the importance of negative keyword discipline.
For renovation, negatives may include:
- jobs
- salary
- course
- school
- free
- DIY
- ideas only
- accessories
- cleaning
- cheap
- rental
- public bathroom
For dentists:
- dental jobs
- dental assistant
- dental school
- free clinic
- insurance only
- salary
- course
For ecommerce:
- free
- template
- used
- wholesale if not relevant
- repair if not relevant
- DIY if not relevant
- manual if not relevant
For SaaS:
- jobs
- salary
- open source
- free template
- definition
- course
- student
Negative keywords should not be random cleanup.
They should reflect what the business refuses to pay for.
Fix 7: Brand and competitor control
AI Max includes brand settings, which is important because brand and non-brand confusion can ruin reporting.
We saw this with Beaver Builder in a different context. A general campaign looked decent because brand synonyms were leaking into it. Once those were separated, the real picture changed.
AI Max needs similar discipline.
Before testing, decide:
- Should AI Max be allowed to match brand searches?
- Should brand be excluded?
- Should competitor brands be included or excluded?
- Should brand traffic live in separate campaigns?
- How will brand/non-brand performance be reported?
- How will AI Max search terms be classified?
If brand traffic is not controlled, AI Max can make performance look better than it is.
Fix 8: URL inclusions and exclusions
Do not let AI Max choose from the whole website unless the whole website is safe for paid traffic.
Most websites are not.
Use URL controls.
Exclude:
- Blog posts that do not convert
- Hiring pages
- Team pages if irrelevant
- Old landing pages
- Thin pages
- Support pages
- Policy pages
- Low-intent guides
- Pages with outdated offers
- Pages that create wrong lead expectations
Include:
- High-intent service pages
- Strong landing pages
- Category pages
- Product pages
- Comparison pages
- Location pages
- Consultation pages
- Offer pages
Google’s AI Max documentation includes URL inclusions and exclusions as available controls.
Use them.
Fix 9: Campaign structure
AI Max should not be layered onto a messy structure without thought.
Before enabling it, check:
- Are campaigns separated by goal?
- Are budgets aligned with priority services?
- Is brand separated where needed?
- Is lead gen separated from ecommerce?
- Are high-value services separated from low-value services?
- Are locations grouped sensibly?
- Is campaign volume enough for Smart Bidding?
- Are ad groups tightly themed enough?
- Are landing pages aligned with ad groups?
PPC Mastery warns that Performance Max can cannibalize Search and go after easy conversions if targets are wrong. AI Max is not Performance Max, but the same automation lesson applies: structure affects what the machine learns.
If structure is bad, automation inherits the mess.
Fix 10: Budget guardrails
Do not test AI Max with money you cannot afford to learn with.
In low-budget, high-CPC accounts, testing needs tighter guardrails.
If the business has under $50/day and clicks cost $15-$40, a few wrong searches can ruin the day.
For AI Max testing, decide:
- Which campaign gets the test?
- What budget is at risk?
- What CPA or CPL is acceptable?
- What lead-quality threshold matters?
- How often will search terms be reviewed?
- How quickly will negatives be added?
- Which URLs are allowed?
- Which features are enabled or disabled?
- How long is the test window?
- What would make us stop the test?
AI Max should be measured as an experiment, not a mood.
The River Stone AI Max testing process
If we were testing AI Max for a client, our first version would usually look like this:
- Audit conversion tracking
- Confirm primary goals
- Review lead quality or revenue quality
- Clean search terms
- Build negative keyword lists
- Review landing pages
- Decide URL inclusions and exclusions
- Decide brand settings
- Choose the safest campaign for testing
- Turn on AI Max only where the account has enough signal
- Review search terms, headlines, and landing pages frequently
- Exclude junk quickly
- Compare AI Max traffic against normal Search performance
- Judge by qualified leads or business value, not only conversions
- Expand only if the test earns trust
That last line is the whole strategy.
AI Max has to earn trust.
When we would not turn on AI Max
We would not turn on AI Max just because Google recommends it.
We would avoid or delay AI Max when:
- Tracking is broken
- The account is already getting bad leads
- The website has weak landing pages
- The budget is too small for exploration
- The service area is extremely narrow
- The offer is vague
- The business cannot review lead quality
- Search terms are already messy
- The account has brand leakage
- The client needs strict legal, medical, financial, or compliance language and text controls are not ready
- Nobody has time to review reporting
AI Max is not a “set it and forget it” feature.
It is a “turn it on only when you are ready to manage what it discovers” feature.
Final view: AI Max is not magic, it is leverage
AI Max is part of where Google Ads is going.
Ignoring it completely is probably not wise.
Trusting it blindly is worse.
The right position is somewhere in the middle:
Understand it.
Prepare for it.
Fix the account first.
Test it carefully.
Judge it by business quality.
Use the reporting.
Control the URLs.
Control the brands.
Review the search terms.
Protect the budget.
For some businesses, AI Max will find profitable demand they were missing.
For others, it will reveal that the account foundation was never strong enough for more automation.
That is not a failure of AI Max alone.
That is a signal.
Before you turn it on, fix the system it will learn from.
About The Author
