Last week, a CMO friend texted me a screenshot of her Google Ads dashboard. "Where did my language targeting go?" she asked. The answer: Google removed it. Starting this month, Search campaigns no longer let you manually select which languages to target. The AI handles it now, matching ads based on your copy and landing pages.
This is not an isolated incident. It's the pattern.
If you've been running B2B campaigns on Google Ads for any length of time, you've watched the control panel shrink while the "AI-powered" label expands. Dynamic Search Ads are auto-upgrading to AI Max starting this month. Enhanced CPC was retired from Search campaigns in March 2025. Performance Max now drives 45% of all Google Ads conversions, and it doesn't let you pick placements, set keyword bids, or exclude audiences the way you used to.
The question every B2B marketer should be asking isn't "How do I fight this?" It's "How do I win inside this new game?"
The Shift Nobody Voted For
Let's be honest about what's happening. Google isn't removing controls because marketers asked for less visibility. They're doing it because their machine learning models perform better when they have more latitude to optimize across signals humans can't process in real time.
Smart Bidding now factors in device type, location, time of day, browser, operating system, and dozens of other contextual signals at auction time. No human team can adjust bids across all those variables for every impression. The math simply doesn't work.
And the results, when the system has good data, are hard to argue with. Google's own benchmarks show that AI Max campaigns see an average 7% lift in conversions when using the full feature suite compared to search term matching alone. Performance Max campaigns report an average 18% CPA reduction versus standard campaign types.
But here's the catch that Google's marketing materials gloss over: those averages hide enormous variance. The algorithm is only as smart as the data you feed it.
Garbage In, Garbage Out (Still True in 2026)
86% of Google Ads campaigns now run automated bidding. Yet the majority of Smart Bidding failures trace to a single root cause: launching a constrained target before conversion data is ready.
For B2B marketers, this is where the game gets tricky. Our sales cycles are long. B2B researchers perform an average of 12 searches before engaging with a specific brand's site. The average deal takes two to six months to close. A form fill in January might not become revenue until July.
Google's algorithm doesn't know that. It sees the form fill and optimizes for more form fills. If you're not telling it which form fills actually turned into pipeline, you're training the algorithm to find you more of whatever converts fastest, not whatever converts best.
This is why enhanced conversions for leads has become non-negotiable for B2B advertisers. The feature connects your CRM data back to Google Ads, matching closed-won opportunities to the exact ad, keyword, and bid that drove them. Without it, you're flying blind. With it, you're teaching the algorithm what revenue actually looks like.
The New Job Description
Here's the mindset shift I've been preaching to my team: we're not campaign managers anymore. We're algorithm trainers.
The old job was about finding the right keywords, writing the right ads, setting the right bids, and excluding the wrong placements. The new job is about feeding the system the right signals and letting it do the optimization work.
What does that look like in practice?
First, fix your conversion tracking before you touch anything else. If your conversion actions are misconfigured, the algorithm will confidently optimize toward the wrong outcomes. I've seen accounts where "page view" was accidentally set as a primary conversion. The AI dutifully drove thousands of page views. None of them were leads.
Second, wait for data maturity. The general rule is 30 conversions per month before Smart Bidding strategies can reliably optimize. For B2B accounts with lower volume, that might mean running manual bidding longer than you'd like, or consolidating campaigns to pool conversion data.
Third, use audience signals as directional hints, not restrictions. In Performance Max, audience signals are starting points, not hard targeting. The algorithm will expand beyond them when it finds additional high-value users. Your job is to give it a good starting hypothesis, then let it learn.
Fourth, invest in creative variety. Asset groups in Performance Max can hold up to 15 headlines, 5 descriptions, 20 images, and 5 videos. The AI assembles these into optimal combinations per placement. More high-quality assets means more combinations to test. Starve the system of creative, and you're limiting its ability to find what works.
The Control You Still Have
I don't want to paint this as total surrender. You still have meaningful levers.
Account-level negative keywords remain essential for brand protection. B2B campaigns should aggressively exclude consumer-intent terms, competitor brand names you don't want to bid on, and job-seeker queries that waste budget.
Value-based bidding lets you assign different conversion values to different actions. A demo request is worth more than a newsletter signup. Tell the algorithm that, and it will optimize accordingly.
Search themes in Performance Max give you directional input for Search inventory. You can specify up to 25 keyword themes per asset group. It's not the same as manual keyword bidding, but it's not nothing.
And perhaps most importantly: you control the landing page experience. Google's AI can optimize which users see your ads and when. It cannot fix a landing page that doesn't convert. The fundamentals of clear value propositions, fast load times, and friction-free forms still matter as much as they ever did.
The Uncomfortable Truth
Here's what I tell marketers who are frustrated by all this automation: the discomfort is the point.
Google is betting that their AI, trained on billions of signals across millions of advertisers, can make better real-time decisions than any individual marketer. For many use cases, they're probably right. The question is whether you're going to be one of the advertisers who benefits from that capability, or one who gets steamrolled by it.
The difference comes down to data quality, creative investment, and strategic patience. Feed the algorithm good signals. Give it enough creative to work with. Let it learn before you panic and start overriding it.
Marketing has always been part art, part science. The science part just got outsourced to a machine. Your job now is to be the artist who gives that machine something worth optimizing.
