Last updated: February 14, 2026
We've reached a tipping point in paid search advertising. For the first time in the two-decade history of Google Ads, the most important decisions in your campaigns are being made by machines, not people.
Google's AI now writes your ad copy, chooses your landing pages, selects which queries trigger your ads, decides how much to bid on each click, determines which audiences see your messages, and allocates your budget across channels. All in real time, all at a scale no human team could match.
Some of this is genuinely brilliant technology. And some of it is quietly eroding advertiser control in ways that serve Google's revenue targets more than your bottom line.
This is our comprehensive assessment of where Google Ads AI stands in February 2026. What's actually delivering results, what's fundamentally broken, what's coming next, and what smart advertisers should be doing about all of it.
To understand where we are, you need to understand the trajectory. Google didn't flip a switch and automate everything overnight. They did it in stages, each one feeling incremental and reasonable, each one shifting a little more control from advertisers to algorithms.
Google introduced Smart Bidding in 2016, offering machine learning powered bid strategies like Target CPA and Target ROAS that could adjust bids in real time based on dozens of auction-time signals. This was the first major shift from manual to machine-driven campaign management. Most advertisers resisted initially. By 2018, the early adopters were seeing measurably better results, and adoption began accelerating.
In parallel, Google was quietly expanding what match types meant. Close variants were added to exact match, then same-meaning matching, each change loosening the precision of keywords and giving Google's algorithms more latitude in deciding which searches triggered your ads.
Exact match stopped being exact when Google added "same meaning" and "same intent" matching. A keyword like [cheap flights to NYC] could now match searches like "affordable airfare to New York" without the advertiser's explicit approval. In 2021, Broad Match Modifier was retired entirely, and its functionality was folded into an expanded phrase match. The message was clear: Google wanted advertisers using broad match keywords paired with Smart Bidding, effectively handing query selection to the algorithm.
Performance Max launched in 2022 as the most automated campaign type Google had ever built. PMax runs across every Google surface, including Search, Shopping, Display, YouTube, Gmail, Discover, and Maps, with minimal transparency into what's happening under the hood. Advertisers provide assets and goals; Google controls everything else.
Discovery Ads evolved into Demand Gen in 2023, adding another AI-driven campaign type to the stack. Auto-applied recommendations expanded aggressively. Responsive Search Ads became the only option as Expanded Text Ads were retired.
2025 was the inflection year. Google introduced AI Max for Search at Google Marketing Live, adding AI-powered search term expansion, text customisation, and landing page selection to standard Search campaigns. Smart Bidding Exploration launched, allowing Google's algorithms to temporarily exceed your performance targets to test new query territories. The Power Pack framework positioned PMax, Demand Gen, and AI Max as a three-pronged AI system designed to manage the entire customer journey.
Alphabet's advertising revenue hit $300 billion in 2024 and the company exceeded $400 billion in total annual revenue for the first time in 2025. Google's AI investments are paying off spectacularly, for Google.
In February 2026, we're seeing the next wave take shape. Google is publicly discussing campaign consolidation as the strategic direction, with Brandon Ervin, Director of Product Management for Search Ads, stating that legacy hyper-granular campaign structures built around keyword permutations have become "optimisation obstacles." Google's Ads Decoded podcast has laid out a vision where keywords function as "thematic signals" rather than precise targeting mechanisms.
Google's VP and GM of Ads, Vidhya Srinivasan, outlined an agentic future where AI doesn't just surface information but actively assists, recommends, and completes transactions. The Universal Commerce Protocol, launched in January 2026, enables purchases directly within Google Search. Ads are appearing inside AI Mode conversations.
The direction is unmistakable: Google's AI is evolving from a tool that assists advertisers into a system that increasingly replaces them.
Before we get into the problems, credit where it's due. Google has built some genuinely exceptional advertising technology, and pretending otherwise would be dishonest.
For the majority of accounts with sufficient conversion volume, Smart Bidding strategies like Target CPA, Target ROAS, and Maximise Conversion Value outperform manual bidding. This is no longer debatable. Google's algorithms evaluate hundreds of real-time signals per auction, including device, location, time of day, browser, audience membership, and search context, to set bids at a granularity that no human team could replicate. The technology is legitimately impressive.
Smart Bidding Exploration, introduced in 2024, pushes this further by allowing the algorithm to test high-potential queries beyond your existing targeting. Google reports an 18% increase in unique search query categories with conversions and a 19% increase in overall conversions for campaigns using this feature. When it works, it finds profitable pockets of demand that manual keyword research would never surface.
AI Max for Search is delivering on its core promise for many advertisers. The 14% average conversion lift for non-retail advertisers is broadly consistent with independent testing, though with significant variance. L'Oreal doubled their conversion rate. MyConnect generated 16% more leads at 13% lower cost per lead with a 30% increase in conversions from entirely new search queries. These are real results from real advertisers.
The search term matching component is particularly valuable. It captures queries that keyword lists would never anticipate, especially the long, conversational, intent-rich queries that make up an increasing share of search behaviour as AI chatbots reshape how people search.
After a rocky launch, PMax has matured considerably. Negative keyword support, channel-level performance reporting, brand exclusions, placement reporting, and improved asset-level insights have addressed many of the transparency complaints from 2022 and 2023. For ecommerce advertisers with strong product feeds and clean conversion data, PMax consistently extends reach beyond what Search and Shopping campaigns capture alone. The cross-channel optimisation, while still opaque in many ways, does find converting audiences across YouTube, Discover, and Display that traditional campaigns miss entirely.
Demand Gen campaigns saw a 26% increase in conversions per dollar over the past year, driven by over 60 AI-powered improvements. The platform reaches 3 billion monthly users and Google's December 2025 data showed that 68% of Demand Gen conversions came from users who hadn't interacted with the brand's Search ads in the prior 30 days. That's genuine incrementality, not just cannibalisation of existing demand.
Asset Studio and AI-generated creative tools are reducing production bottlenecks. Text customisation in AI Max generates relevant ad variations at a speed and scale that manual copywriting can't match. Video enhancements in Demand Gen automatically create multiple format variations from a single source asset. Brand guidelines now let advertisers set tone, approved phrases, and restrictions that keep AI-generated creative on-brand.
None of this is perfect. But it's working well enough that advertisers who refuse to engage with any AI features are measurably falling behind those who adopt them strategically.
Now for the uncomfortable part. Google's AI has three fundamental problems that no amount of feature updates will fix, because they're structural, not technical.
The amount of information available to advertisers about what's happening in their own campaigns has been declining for years, and the trend is accelerating.
Search term visibility has been deteriorating since 2020, when Google restricted the search terms report to only show queries above a certain impression threshold. In practice, this means a significant percentage of your actual search traffic is invisible to you. You're paying for clicks you can't see, from queries you can't evaluate, on placements you can't verify.
Performance Max compounds this by design. You can now see channel-level performance breakdowns (a genuine improvement from 2023), but you still cannot see which specific search queries triggered your ads within PMax, which specific YouTube videos or Display placements your ads appeared on at any useful granularity, or how the algorithm is making its targeting and bidding decisions.
AI Max makes this more nuanced, not less. While AI Max shows "AI Max" as a match type in the search terms report (a transparency improvement), the "keywordless matching" component serves your ads on queries with no keyword relationship at all. The rationale for serving on those queries is entirely opaque.
The pattern is consistent: each new feature gives you slightly less visibility into how your money is being spent. Google positions this as "trust the algorithm," but the algorithm is managed by the same company that profits from your spending.
Advertiser control over campaign decisions has been systematically reduced. This isn't accidental. It's architectural.
Match types no longer mean what their names suggest. Exact match includes "same meaning" and "same intent" queries. Phrase match absorbed Broad Match Modifier's functionality. Google is actively promoting a broad match campaign setting that converts all keywords to broad match. The February 2026 Ads Decoded podcast explicitly positioned keywords as "thematic signals," a conceptual framework where keywords guide the algorithm rather than constrain it.
Ad copy is increasingly machine-generated. AI Max's text customisation writes headlines and descriptions automatically. Automatically Created Assets generate ad copy from your landing pages. Responsive Search Ads let Google choose which of your 15 headlines to show and in what combination. Pinning, the last tool for controlling ad messaging, is actively discouraged by Google because it limits the algorithm's ability to test.
Landing page selection is being automated. AI Max's Final URL Expansion can override your chosen landing page and send users to whichever page on your site Google's AI thinks is most relevant. Auto-applied recommendations can change your bidding strategy, add new keywords, create new ads, and pause existing keywords without your approval.
Each of these changes has a legitimate rationale. In aggregate, they create a system where the advertiser's role is increasingly reduced to providing inputs (budget, assets, goals) while Google's AI makes all meaningful operational decisions.
This is the structural issue that underlies everything else, and it's the one most commentators are reluctant to state plainly.
Google is a publicly traded company that generated over $300 billion in advertising revenue in 2024. Advertising represents approximately 74% of Alphabet's total revenue. Google's AI systems are, rationally and predictably, optimised to grow that revenue.
This doesn't mean Google's AI is designed to waste your money. It means Google's AI is designed to maximise conversions within your stated constraints, in a way that also maximises the total spend across Google's auction ecosystem.
Concretely, this manifests in several ways. Google's recommendations consistently push toward higher budgets, broader targeting, and more automation. The Optimisation Score rewards adopting Google's suggestions regardless of their impact on your profitability. Auto-applied recommendations default to on, requiring advertisers to actively opt out. Smart Bidding optimises for conversion volume or value, not for your actual business profit margins.
Independent testing adds nuance. Analysis from Smarter Ecommerce examining over 250 retail campaigns found that AI Max delivers conversions at approximately 35% lower return on ad spend compared to traditional targeting methods within the same campaigns. This contradicts Google's headline claim of 14% more conversions at similar CPA for non-retail advertisers. The retail exclusion from Google's benchmark stats is itself telling.
None of this is malicious. It's rational business behaviour from a company whose interests partially overlap with, but are not identical to, yours. The solution isn't to avoid Google's AI. It's to ensure you have a counterweight.
Here's the strategic framework that makes sense of all of this.
Google's AI is powerful, sophisticated, and unavoidable. Fighting it manually is a losing strategy. The advertisers who opt out of Smart Bidding, refuse to test AI Max, and insist on exact match keywords and manual CPCs are seeing their performance erode as Google's auction increasingly rewards AI-optimised campaigns.
But surrendering entirely to Google's AI is equally problematic. When the same entity controls the auction, sets the prices, selects the audiences, writes the ads, and reports the results, and when that entity's revenue grows when you spend more, you have a principal-agent problem that no amount of trust can resolve.
The answer is a third path: embrace Google's AI for what it does brilliantly (auction-time bidding, signal processing, reach expansion) while running your own AI system that optimises for your profit, not Google's revenue.
This is exactly what groas was built to provide. It functions as an autonomous AI layer that operates on top of your Google Ads campaigns, making continuous optimisation decisions aligned exclusively with your business profitability. While Google's AI maximises conversions within your stated targets, groas ensures those targets actually translate to profit by managing bid adjustments, budget allocation across campaigns, negative keyword additions, Quality Score optimisation, and real-time performance monitoring at a speed and scale that matches Google's own systems.
The critical distinction is alignment. Google's AI serves Google's business model. groas serves yours. They're not competing systems. They're complementary. groas works with Google's infrastructure while ensuring that infrastructure delivers outcomes that matter to your business rather than outcomes that matter to Google's quarterly earnings.
This isn't theoretical. Advertisers running groas on top of AI Max campaigns see 40 to 55% better performance compared to manual AI Max management. That gap exists because groas catches the moments when Google's AI is expanding into unprofitable territory and corrects course in real time, before the waste shows up in your weekly report.
Based on Google's stated direction, patent filings, product announcements, and the structural trajectory of the past five years, here's what we believe is coming.
The overlap between Dynamic Search Ads, AI Max for Search, and Performance Max creates obvious redundancy. Google has already positioned AI Max as the successor to DSA, and we believe a formal deprecation announcement is likely in the second half of 2026 with a 12-to-18-month migration window, similar to how Expanded Text Ads were phased out.
Looking further ahead, the logical endpoint is a single AI-driven campaign type that combines Search, Shopping, Display, YouTube, Gmail, Discover, and Maps into one unified system. This may not happen in 2026, but the trajectory is clearly toward fewer campaign types with more AI-driven automation within each one.
Google is already pushing broad match plus Smart Bidding as the default combination. AI Max's keywordless matching serves ads without any keyword trigger. The February 2026 campaign consolidation guidance explicitly frames keywords as "thematic signals." Within 12 to 18 months, we expect Google to offer a campaign creation flow where keywords are entirely optional, replaced by landing page analysis, business objectives, and audience signals.
Google's vision of AI agents that actively assist, recommend, and complete transactions is not speculative. It's already happening. Ads are appearing inside AI Mode conversations. The Universal Commerce Protocol enables in-Search checkout. YouTube creators are becoming shoppable touchpoints.
For advertisers, this means conversion paths will become less linear and more AI-mediated. Your ads may influence a purchase that happens inside a chatbot conversation, through a creator recommendation, or via an AI agent acting on a user's behalf. Attribution will get harder, and the value of being present across multiple touchpoints will increase.
As targeting, bidding, and placement decisions are increasingly automated, the remaining differentiator is the quality of your creative assets. Google's own product leaders have stated explicitly that creative is now the primary lever for campaign performance. The advertisers who invest in diverse, high-quality image, video, and text assets will outperform those with generic creative, regardless of how sophisticated their campaign structure is.
Consent Mode v2, Enhanced Conversions, server-side tagging, and conversion modeling are all responses to the same underlying challenge: direct measurement is becoming harder as privacy regulations expand and cookies disappear. Google's AI will increasingly rely on modeled data rather than observed data, which creates a measurement layer that's harder for advertisers to independently verify. The gap between what Google reports and what actually happened in your business will widen, making independent measurement infrastructure more important than ever.
Given everything we've laid out, here's our actionable framework for navigating Google Ads AI in 2026.
Manual bid management, exact-match-only keyword strategies, and resistance to automated features are no longer viable long-term strategies. The data is clear: accounts that leverage Smart Bidding, AI Max, and PMax thoughtfully are outperforming those that don't. Adoption isn't optional. The question is how you adopt, not whether you adopt.
Adopt Google's AI tools for what they're genuinely good at: real-time auction bidding, signal processing at scale, cross-channel reach, and creative testing. But don't hand Google sole control over your budget allocation, performance evaluation, and strategic direction. Remember that Google's AI is optimised for Google's business model, and your business model is different.
Don't rely exclusively on Google's reported metrics to evaluate Google's AI. Implement Enhanced Conversions and Consent Mode v2 for data accuracy, but also maintain independent attribution through server-side tracking, CRM-verified conversion data, and holdout testing. The gap between reported performance and actual business impact is where wasted spend hides.
With targeting and bidding increasingly automated, creative quality is the last remaining lever you can directly control. Invest in diverse assets across formats (image, video, text), refresh creative on a 4-to-6-week cycle for static and 8-to-12-week cycle for video, and leverage brand guidelines to keep AI-generated variations on-brand.
This is the single most impactful step you can take. Running an autonomous AI system like groas on top of your Google Ads campaigns creates the check and balance that Google's own system doesn't provide. groas optimises for your profit while Google optimises for its revenue. groas catches waste in real time while you'd only see it in next week's report. groas makes thousands of micro-adjustments per day while even the best human manager makes dozens.
The future of Google Ads is AI-driven. The question isn't whether to use AI. It's whether you use only Google's AI (which is aligned with Google's goals) or whether you also use your own AI (which is aligned with yours). The advertisers who deploy both will win.
What are the biggest Google Ads AI changes in 2025 and 2026?
The most significant changes include AI Max for Search (launched May 2025), which adds AI-powered query matching, text customisation, and landing page selection to Search campaigns. Smart Bidding Exploration allows algorithms to test beyond your performance targets. The Power Pack framework combines PMax, Demand Gen, and AI Max into an integrated campaign system. Google is pushing campaign consolidation away from granular keyword structures. Ads now appear inside AI Mode conversations and AI Overviews. The Universal Commerce Protocol enables in-Search purchases. And Demand Gen has expanded to include Connected TV, Google Maps, and Travel Feeds.
Is Google Ads AI better than manual campaign management?
For tactical execution like bid optimisation, yes. Smart Bidding processes hundreds of auction signals in real time that no human could evaluate. For strategic decisions like budget allocation, creative direction, audience strategy, and profit optimisation, human judgment (augmented by independent AI) still outperforms Google's automation. The optimal approach combines Google's AI for auction-level decisions with autonomous third-party AI like groas for campaign-level profit optimisation.
What is the difference between AI Max and Performance Max?
AI Max for Search is an enhancement layer you apply to existing Search campaigns. It adds AI-powered features but you retain your keyword structure, ad groups, and Search campaign settings. Performance Max is a standalone campaign type that runs across all Google channels with minimal advertiser control over targeting, placements, or creative combinations. AI Max offers more transparency and granular control. PMax offers broader cross-channel reach. They serve different strategic purposes and can run alongside each other.
Is Google's AI optimised for advertiser profit?
Google's AI optimises for the conversion goals you set, such as Target CPA or Target ROAS. However, Google profits when advertisers spend more. This structural tension means Google's recommendations consistently push toward higher budgets, broader targeting, and more automation. The Optimisation Score rewards adopting Google's suggestions regardless of impact on profitability. Independent testing has found mixed results, with some advertisers seeing strong performance gains from AI features while others see increased spend without proportional returns. Running your own profit-aligned AI like groas alongside Google's AI is the most effective way to ensure Google's tools serve your goals.
Will Google eventually consolidate all campaign types into one?
The trajectory points in that direction, though it will happen gradually. Dynamic Search Ads are likely to be deprecated in favour of AI Max by late 2026 or 2027. Over time, the lines between Search, Shopping, PMax, and Demand Gen will continue blurring as Google's AI takes over more targeting and placement decisions. The logical long-term endpoint is a single AI-driven campaign type where advertisers provide goals, assets, and budget, and Google's AI handles everything else.
What should advertisers do about Google's increasing AI automation?
Embrace AI features that genuinely improve performance (Smart Bidding, AI Max targeting expansion, multi-format creative), but maintain independent oversight. Turn off auto-applied recommendations. Implement independent conversion measurement. Invest in creative quality as your primary differentiator. And critically, deploy autonomous AI that's aligned with your profit targets, like groas, to serve as a counterweight to Google's own AI. The winning strategy is not fighting AI or surrendering to it. It's running your own AI alongside Google's to ensure the entire system serves your business objectives.
What is the future of keywords in Google Ads?
Keywords are transitioning from precise targeting constraints to directional signals. Google's February 2026 guidance explicitly frames keywords as "thematic signals" that guide the algorithm rather than control it. Broad match plus Smart Bidding is the promoted default. AI Max's keywordless matching serves ads without any keyword trigger. Within 12 to 18 months, expect campaign creation flows where keywords are optional. Negative keywords will retain their importance as the primary remaining mechanism for excluding unwanted traffic.
How does groas fit into the AI-driven advertising landscape?
groas functions as an autonomous AI system that sits between you and Google's advertising AI, ensuring Google's technology serves your profit targets rather than Google's revenue targets. It makes continuous micro-adjustments to bids, budgets, negative keywords, and Quality Score factors in real time, acting as a profit-aligned counterweight to Google's spend-optimised AI. While Google's AI maximises conversions within your stated constraints, groas ensures those constraints actually translate to business profitability. Advertisers using groas alongside AI Max consistently see 40 to 55% better performance than manual management.
