Google Ads in 2025: What AI Features You Need (and How)

Google Ads in 2025: What AI Features You Need (and How)

AI Turns Google Ads into a Black Box… Or Opportunity?

The Google Ads advertising landscape is undergoing a profound transformation. With the major announcements at Google Marketing Live 2025, artificial intelligence is no longer simply an assistant optimizing your bids – it’s becoming a co-strategist making complex decisions for you.

For advertisers, this transformation raises a fundamental question: how can you remain competitive by harnessing the power of AI, while retaining the necessary strategic control? How can you prevent your campaigns from becoming a “black box” where you lose all visibility over the decisions made by the algorithm?

This comprehensive guide explores the new AI features being deployed in 2025, their concrete budgetary impacts, and the strategies needed to cope with more automation without losing control of your advertising spend.

1. Panorama of Major Additions in 2025 | Google Ads

1.1 AI Max for Search: The New Standard for Search Campaigns | Google Ads

AI Max for Search represents the biggest evolution of Search campaigns in a decade, introducing a comprehensive suite of targeting and creative enhancements that integrate the best of Google’s AI. Unlike traditional Search campaigns that rely solely on your keywords, AI Max uses artificial intelligence to identify search opportunities that you wouldn’t have targeted manually.

The key components of AI Max :

Enhanced Search Term Matching: The system uses broad match and keywordless technology to find relevant, high-performing queries you’d otherwise miss, by learning from your current keywords, creative assets and URLs. For example, if you’re targeting “running shoes”, AI Max might identify “best sneakers for trail hiking” as a relevant opportunity.

Dynamic ad personalization (Text Customization): The system automatically generates new text assets, such as titles and descriptions, based on your landing page, ads and keywords, adapting creative in real time to match emerging intent. This feature, formerly known as “automatically created assets”, allows your ads to adjust instantly to the nuances of each query.

Final URL Expansion: This advanced technology, an evolution of Dynamic Search Ads, automatically directs users to the most relevant landing page based on their search intent, even if it’s not the URL you initially configured.

Observed performance :

Advertisers who activate AI Max in their Search campaigns typically see a 14% increase in conversions or conversion value at similar CPA/ROAS. For campaigns based primarily on exact match or phrase keywords, the typical increase is even higher, reaching 27%.

Availability: AI Max completed its global beta launch in May 2025, with full availability scheduled for early Q3 2025.

1.2 Performance Max 2.0: More Control and Transparency at Last | Google Ads

Performance Max has long been criticized for its lack of visibility and control. Google listened and rolled out a series of major improvements in 2025 that radically transform the advertiser experience.

Negative keyword lists at campaign level :

The campaign-level negative keyword functionality, introduced in beta in 2024, is now rolled out to all advertisers. This evolution finally makes it possible to exclude specific queries for reasons of relevance or brand compliance, and to apply these exclusions to multiple PMax campaigns simultaneously.

Search Themes improved :

Search themes allow you to indicate the queries you know your customers are using to find you, and are additive to the queries and placements that Performance Max predicts will perform well. Google has doubled the limit from 25 to 50 themes per asset group and introduced a usefulness indicator that shows whether your themes are generating incremental traffic beyond what PMax would find on its own.

Brand exclusions by format :

For retail advertisers with product feeds, brand exclusions now apply differently according to format: you can apply brand exclusions only to Search text ads, while retaining branded traffic for Shopping ads. This granularity prevents cannibalization between your PMax campaigns and your dedicated branded Search campaigns.

Demographic and device controls :

Two new betas are coming soon: age-based demographic exclusions to exclude age brackets such as “18-24” or “65+”, and per-device control to customize targeting for computer, mobile or tablet.

Transparent reporting :

Search term reporting completed its roll-out across all Performance Max campaigns in 2025, providing complete transparency on the user queries that trigger your ads. In addition, the ability to segment and download asset group performance has been rolled out, enabling granular analysis of conversions, conversion value, impressions, clicks and costs across YouTube, Display, Search, Discover, Gmail and Maps.

Investment exclusions via API :

In a major revelation from January 2025, Google has acknowledged that Performance Max campaigns are controlled via API placement exclusions, contradicting months of documentation and official support guidance. This capability now allows advertisers to block specific websites, apps and YouTube channels on a large scale, solving a major brand safety issue.

1.3 Smart Bidding Exploration: The Biggest Auction Update in Over a Decade | Google Ads

Smart Bidding Exploration represents Google’s biggest bidding update in over a decade, pursuing less obvious but potentially high-performance research.

How it works :

Traditionally, your bids focus on high intent queries where you have confidence. Smart Bidding Exploration enables AI to test more exploratory queries – searches where the user is still in the discovery phase rather than in immediate purchase mode.

With flexible ROAS targets, your campaigns can now move beyond narrow, high-intent terms and start capturing more exploratory searches. For example, instead of being limited to “mortgage rate”, your ad could appear for “how to buy a house” – capturing the user much earlier in their journey.

Observed results :

Campaigns using Smart Bidding Exploration see an average 18% increase in unique query categories generating conversions, and a 19% increase in conversions.

1.4 Demand Gen: The multiplatform replacement for Video Action Campaigns | Google Ads

Demand Gen replaces Video Action campaigns and offers extended coverage across YouTube, Discover and Gmail with mixed formats optimized for demand generation.

New control capabilities :

Advertisers now have greater control over where their Demand Gen ads appear, with new channel controls allowing placements to be specified across YouTube, Discover and Gmail. One particularly attractive option enables exclusive targeting of YouTube Shorts, Google’s answer to short-form video platforms like TikTok and Instagram Reels.

Product flow integration :

Demand Gen adds new product feed integrations allowing users to see full product details directly from ads, provided your account is set up in Google Merchant Center.

1.5 Ads in AI Overviews and AI Mode | Google Ads

Google extends ads in AI Overviews to the desktop, and introduces ads in AI Mode to create new opportunities.

AI Overviews on desktop :

Google is now displaying Search and Shopping ads directly inside AI Overviews on desktop, starting in the US, enabling businesses to connect with potential customers much earlier in their search journey. If you’re already using campaigns like Performance Max, Shopping or Search en broad match, you’re automatically eligible for these new placements.

AI Mode :

AI Mode is a distinct conversational search experience, powered by Gemini. In AI Mode, users don’t simply search, they “brief” with queries 2 to 3 times longer, pointing to a growing reliance on search to manage complexity. Ads will soon appear contextually in these extended conversations.

1.6 AI Creation Tools: Veo, Imagen and Asset Studio | Google Ads

Google has made incredible progress in improving the quality of its creative tools powered by its cutting-edge video and image generation models, Veo and Imagen, which will soon be available in Google Ads and Merchant Center.

Video generation from images :

The new Veo model, already available in Merchant Center and soon in Google Ads, transforms static product images into dynamic videos. This capability democratizes video creation for brands without a full production team.

AI Outpainting:

This feature allows marketers to extend video frames to create more immersive and dramatic content.

Asset Studio :

All this is brought together in a new Asset Studio, where brands can easily generate and test high-quality visuals that look polished and aligned with the brand, without requiring a full production team.

1.7 Agentic capabilities: Marketing Advisor and Google Ads Expert | Google Ads

Google introduces “agentic” tools that go beyond suggestions to actively create, optimize and solve problems in real time.

Marketing Advisor :

Marketing Advisor, an AI agent integrated into the Chrome browser, will help advertisers manage all kinds of marketing tasks, such as suggesting product promotions based on seasonality, examining your website and solving a conversion tracking problem, or assisting with complex multi-site tagging work.

Google Ads Expert :

This tool aims to streamline campaign creation and provide specific recommendations based on your existing campaign and company data. Google says it will proactively identify and correct problems before they impact your ads.

1.8 First-Party Measurement and Data: Data Manager | Google Ads

The new Data Manager tools make it possible to collect first-party data from more sources (such as Salesforce, HubSpot, etc.), store it securely and use it more effectively from a “one-stop data shop”.

Extended incrementality testing :

Google has lowered the spending threshold for its incrementality tests to $5,000, with an improved testing methodology for better results. This democratization of testing enables more advertisers to rigorously measure the real incremental impact of their campaigns.

2. What It All Means For Your Budget & Strategy | Google Ads

2.1 The Risk of Control Dilution: Less Micromanagement, More Strategic Monitoring

Increasing automation is transforming the role of the advertiser. You move from “campaign manager” to “strategic supervisor”. This evolution presents both opportunities and risks.

The challenge: With AI Max, Performance Max and Smart Bidding Exploration, your campaigns make decisions you could never make manually – because they process billions of signals in real time. But this power comes at the price of reduced visibility into individual decisions.

The opportunity: By delegating tactical decisions to AI, you free up time to focus on what really matters: your positioning strategy, your message angles, your value proposition, and your customer understanding.

The key: Define clear safeguards. Brand exclusions, campaign-level negative keywords, demographic exclusions and “URL contains” rules become your essential tools for framing AI without stifling it.

2.2 Budget impact: between increased efficiency and the risk of uncontrolled expansion

Scenario 1: Increased efficiency

In the best case, AI identifies high-performance audiences and queries that you would never have discovered manually. Your cost per acquisition decreases, your conversion volume increases at constant budget, and your ROI improves significantly.

Case in point: An advertiser using AI Max for Search saw a 14% increase in conversions at a similar CPA. On a monthly budget of $50,000, this represents $7,000 in additional value generated at no extra cost.

Scenario 2: Margin dilution

The opposite risk exists: AI could identify less qualified but more numerous audiences. Your volume of conversions increases, but your conversion rate beyond the first purchase decreases, and your customer lifetime value deteriorates.

Case in point: Smart Bidding Exploration captures queries higher up the funnel (“how to choose a CRM” vs. “buy Salesforce”). These conversions have a lower immediate value, even though they can generate long-term value.

The key: Segment your metrics. Don’t just track overall CPA and ROAS. Analyze:

• Multi-stage conversion rate (first conversion vs. purchase)
• Customer lifetime value by acquisition cohort
• Retention rate by campaign source
• Profitability by audience segment

2.3 Auction volatility and learning cycle

AI needs data to optimize. This learning period can create initial volatility.

Phase 1: Exploration (weeks 1-2) The algorithm tests different audiences, queries and placements. Your costs may temporarily increase, and your ROAS may fluctuate. This is normal – the AI collects the necessary data.

Phase 2: Stabilization (weeks 3-4) Performance begins to converge towards a stable average. You begin to see real trends emerging.

Phase 3: Continuous optimization (beyond) AI continues to refine, but with less volatility. Adjustments become gradual rather than abrupt.

Recommended budget for the test phase: Allow 20-30% above your usual budget for the first 3-4 weeks. This margin absorbs learning volatility without compromising your other channels.

2.4 Coordination between investment in Data First-Party and Automation

AI is only as good as its access to quality data. Your first-party data – customer lists, CRM data, historical purchasing behavior – is the fuel that enables AI to make intelligent decisions.

Priority investments :

1. Enriched Customer Match: Segment your customer lists by lifetime value, purchase frequency, categories of products purchased. This enables AI to distinguish high-value audiences.
2. Advanced conversion tracking: Track not only conversions, but also micro-conversions (downloads, newsletter subscriptions, time on site). AI uses these signals to predict lead quality.
3. CRM-Google Ads integration: Use the new Data Manager to connect your CRM systems (Salesforce, HubSpot) directly to Google Ads. This enables a complete feedback loop: the AI sees which leads become real customers, and adjusts accordingly.

ROI on data investment :

An investment in the quality of your first-party data generates a multiplier return. If your AI has access to 2x better quality signals, it will make 2x better decisions – and your overall performance can improve by 30-50%.

2.5 Adjusted budget scenarios: Test phases vs. mature phases

Test phase (months 1-2) :

• Budget: 120-130% of normal budget
• Objective: Learning and validation
• KPIs monitored: CPA volatility, conversion quality, audience signals
• Frequent adjustments : Weekly

Transition phase (months 3-4) :

• Budget: 110% of normal budget
• Objective: Stabilization and optimization
• KPIs monitored: ROAS, volume, query diversification
• Adjustments : Bi-weekly

Mature phase (month 5+) :

• Budget: 100% with flexible allocation for scaling
• Objective: Performance maximization and scaling
• KPIs monitored: Volume growth at target ROAS, share of voice, geographic expansion
• Adjustments : Monthly or based on business events

3. Recommendations & Best Practices | Google Ads

3.1 Getting started : Pilot approach and Test Accounts | Google Ads

Never switch all your campaigns at once. The pilot approach is essential:

Step 1: Identify the pilot Choose a representative but non-critical campaign:

• Sufficient budget (minimum $3,000-5,000/month) to enable AI learning
• Performance history to establish a baseline for comparison
• Not your most critical campaign (to limit risk)

Step 2: A/B test setup Create a clear test structure:

• Control campaign: maintain your current approach
• Test campaign: active AI Max, Smart Bidding Exploration, or Performance Max
• Split budget 50/50 or 60/40 (60% control for safety)
• Minimum duration: 4-6 weeks to enable complete learning

Step 3: Metrics for success Define your decision criteria in advance:

• ROAS improvement of at least X%.
• Maintain volume of qualified conversions
• No increase in return/cancellation rate
• Positive query diversification (new high-performance keywords)

Step 4: Go/no-go decision If the test is successful, deploy gradually (one campaign per week). If the test fails, analyze why: insufficient data, poorly calibrated objectives, or the tool is not adapted to your context.

3.2 Maintaining Safeguards: Exclusions, Alerts and Thresholds | Google Ads

Multi-layer exclusion system :

Level 1: Negative keywords

• Account-level master list for global protection (competitors, irrelevant terms, career searches)
• Campaign-level lists for specific sectors
• Monthly review and progressive addition

Level 2: Investment exclusions

• Using API investment exclusions for PMax
• Systematically block low-performance or problematic investments
• Maintain an evolving list based on performance reports

Level 3: Demographic and geographic exclusions

• Exclude irrelevant demographic segments (e.g. 18-24 for a senior B2B product)
• Refining geographical areas to concentrate budgets

Automated alert system :

Set up alerts to detect anomalies quickly:

• Alert if CPA increases by >25% over 3 consecutive days
• Alert if prints suddenly drop >40%.
• Alert if a new request consumes >10% of the budget in 24 hours
• Alert if conversion rate drops below X% (your minimum threshold)

Budget thresholds per campaign :

Set clear limits:

• Maximum daily budget per campaign (to avoid slippage)
• Maximum cost per conversion accepted
• Minimum ROAS required before scaling
• Automatic pause threshold if conditions deteriorate

3.3 Asset Quality : The Fuel of AI | Google Ads

AI is only as good as the materials it has at its disposal. Your assets – titles, descriptions, visuals, landing pages – are the fuel that drives performance.

Text assets (Responsive Search Ads, PMax) :

• Provide at least 10-15 varied titles (not minor repeats)
• Include different angles: benefits, features, price, urgency, social proof
• Test different lengths (short and punchy vs. descriptive)
• Include main keywords in at least 3-4 titles
• Vary your calls to action

Visual assets (Display, Demand Gen, PMax) :

• Multiple formats: square, landscape, portrait (the AI will choose the best according to placement)
• Professional quality: sharp, well-lit images aligned with your brand identity
• Test different styles: lifestyle, single product, contextualization
• Include videos if possible (generally superior performance)

Landing pages :

• Optimized loading time (<3 seconds)
• Message consistent with the ad (avoid disappointment)
• Clear, visible call to action
• Mobile-first: 60-70% of traffic will be mobile
• Complete tracking to enable AI to learn

Cooling principle :

New AI features. Refresh your assets:

• Titles/descriptions: every 2-3 months
• Visuals: quarterly
• Videos: every 6 months
• Landing pages: continuous optimization based on heatmaps and analytics

3.4 Close tracking of metrics: Beyond ROAS | Google Ads

Essential monitoring dashboard :

Performance metrics :

• ROAS by campaign, by audience segment, by channel (Search, Display, YouTube)
• CPA by conversion type (lead, purchase, registration)
• Volume of qualified conversions (not just total)
• Average conversion value

Health metrics :

• Conversion rates by funnel stage
• Landing page bounce rate
• Average time on site post-click
• Brand search rate (brand lift indicator)

Control metrics :

• Share of budget consumed by unknown requests (AI Max, broad match)
• Number of new placements tested by AI per week
• Asset match rate (% of assets actually used vs. downloaded)
• Average quality score by ad group

Long-term quality metrics :

• Customer lifetime value by acquisition cohort
• Retention rates at 30, 60, 90 days
• NPS or customer satisfaction by source of acquisition
• Return/refund rate by campaign

Revision frequency :

• Daily: Anomalies and alerts
• Weekly: Performance vs. objectives, tactical adjustments
• Monthly: Strategic analysis, scaling/pause decisions
• Quarterly: Comprehensive review, strategic realignment

3.5 Periodic Review : Human vs. Machine Arbitration | Google Ads

Automation is not binary. It’s not “all manual” or “all automated”. It’s a spectrum where you have to continually arbitrate.

Arbitration questions to ask yourself quarterly :

1. Does AI capture opportunities that I wouldn’t have found?
   • If yes: give it more space (increase AI Max budgets, widen audiences)
   • If not: tighten controls, refine exclusions
2. Is the quality of conversions in line with my business objectives?
   • If yes: scaling possible
   • If not: recalibrate your conversion signals, enrich your first-party data
3. Are acquisition costs rising without value justification?
   • If yes: return to more manual control, test different audiences
   • If not: continue and explore scaling
4. Does my team understand why campaigns perform the way they do?
   • If yes: you have the right level of transparency
   • If not: invest in training, demand

At Koanthic, we help companies implement and orchestrate AI-driven Google Ads (AI Max, Performance Max, Smart Bidding, Demand Gen, Data Manager, etc.). Our specialty: turning automation into measurable competitive advantage, thanks to clear safeguards (brand safety, negative lists, exclusions, URL rules), high-quality creative assets, and first-party measurement integrated with your CRM.

– Express account audit (controls & transparency)
– AI pilot deployment (A/B clear, business KPIs)
– Data integration (CRM, Customer Match, feedback loops)
– Governance & dashboards (ROAS, LTV, retention, lead quality)