When Google introduced AI Max for search campaigns, the headline features got most of the attention - text customisation, expanded reach, smarter query matching. What got far less attention was what it does to your keyword match types and how it changes the data you can actually see. Both of those things matter a great deal if you're running campaigns where lead quality and cost per acquisition are the primary metrics.
Broad Match by Default, Whether You Want It or Not
Here's the core issue: when you enable AI Max on a search campaign, it treats all your keywords as broad match - including keywords you've deliberately set to exact or phrase. You don't change the match types yourself. Google's systems simply start operating as if you had. For advertisers who have spent time building tightly controlled keyword sets, this is a significant shift in how those campaigns function.
This isn't a subtle edge case. Exact match keywords exist for a reason - they give you predictability around the queries that trigger your ads, tighter quality control, and cleaner attribution. Phrase match adds controlled flexibility. Broad match, used intentionally, can work well alongside Smart Bidding with enough conversion data behind it. But having broad match behaviour imposed on your entire keyword set by enabling a feature is a different situation. It changes the risk profile of your account without an explicit decision to do so.
The practical consequence is that your ads may start appearing for a much wider range of queries than your match types would previously have permitted. For campaigns targeting specific service lines or high-value verticals, that expanded reach can introduce irrelevant traffic, inflate impressions, and - if Smart Bidding doesn't have strong enough conversion signals - pull CPA in the wrong direction before you've noticed anything is wrong.
The Reporting Change That Makes This Harder to Catch
Compounding the match type issue is a change to search term reporting. AI Max does not surface the same level of search term data that standard campaigns have historically provided. The search terms report becomes less granular - meaning you have less visibility into the specific queries that are triggering your ads and generating conversions or wasted spend.
This creates a real problem. Broad match behaviour expands the query pool significantly, but reduced reporting makes it harder to audit what's in that pool. The two changes work against each other from an advertiser's perspective. You're operating with wider reach and narrower insight at the same time. That combination makes it genuinely difficult to identify poor-performing queries, build meaningful negative keyword lists, or understand which search intent categories are driving results.
For lead generation accounts in particular, this matters more than it might for ecommerce. A lead from a mismatched query doesn't just waste the click budget - it goes into your CRM, gets worked by a sales team, and can corrupt your downstream performance data if you're feeding offline conversion signals back into Google Ads. Messy query matching at the top can cause real problems further down the funnel.
Rethinking Your Keyword Workflow for AI Max
The response to this isn't to avoid AI Max entirely - the reach expansion and query matching capabilities may well deliver value, particularly in competitive markets where exact and phrase match have become increasingly limited by Google's own matching behaviour over recent years. But you do need to adapt your keyword management workflow if you're enabling it.
Negative keywords become your primary control mechanism. With match types effectively neutralised as a filter, negatives are doing more of the work than they were before. That means your negative keyword lists need to be more comprehensive and more proactively maintained. Waiting for bad queries to appear in the search terms report and then adding negatives reactively is a workflow built for standard campaigns - it's too slow when reporting visibility has been reduced.
Front-load your negatives before enabling AI Max. Think about the adjacent queries your keywords are likely to attract at broad match - competitor brand terms you don't want to show for, irrelevant job titles or industries, informational queries that won't convert. Build those exclusions in before the campaign goes live in AI Max mode, rather than waiting to see what surfaces. You'll also want to review your campaign-level and account-level negative lists to make sure they're current.
Using Search Term Data More Strategically
Even with reduced granularity in the search terms report, the data you do have access to should be reviewed more frequently than was necessary under tighter match type controls. Weekly reviews are a sensible minimum when AI Max is running. Look at which query categories are generating conversions versus which are consuming budget without measurable outcomes.
Where possible, layer in offline conversion data to help Smart Bidding understand lead quality, not just lead volume. If AI Max is expanding your reach into queries that generate high volumes of low-quality leads, bidding purely on form fills or call conversions won't correct for that. Feeding qualified pipeline or revenue data back into Google Ads via the offline conversions import gives the algorithm a better signal to optimise against - and helps counteract the quality dilution that wider query matching can introduce.
It's also worth segmenting your performance by audience where you can. AI Max's text customisation features can tailor ad copy based on audience signals, which gives you some ability to differentiate messaging for different user profiles even within the same campaign. Used carefully, that can improve relevance at the ad level even when query-level control has been reduced.
What This Means for Account Structure
If you're running AI Max alongside standard search campaigns in the same account, you need to be clear about which campaigns are doing which job. AI Max campaigns operating with broad match behaviour will compete with your standard campaigns for the same queries unless you're deliberate about separation. That can lead to cannibalisation or confusion in your attribution data.
Consider treating AI Max as an expansion layer - something you run in addition to a tightly controlled core of exact and phrase match campaigns, rather than a replacement for them. That way you retain predictable performance from your core campaigns while testing whether AI Max's expanded reach generates incremental volume at an acceptable CPA. Running both and comparing performance over a meaningful time window gives you real data on whether the trade-off in control is worth the reach.
The broader pattern here is one advertisers have seen before with Performance Max - Google's AI features tend to expand reach and reduce granular control simultaneously. The right response isn't to reject the tools, but to be specific about what you're giving up, build compensating controls around what you still have access to, and measure outcomes rigorously rather than relying on the platform's own reporting to tell the full story.