Using AI To Predict PPC Results: A Guide for Accurate Forecasting

Google's Native AI Evolution: The Biggest Changes in a Decade

Google Ads has undergone a fundamental transformation in 2024–2025. The platform now uses AI to automate bidding, targeting, creative generation, and performance forecasting across all campaign types. For B2B marketers managing campaigns through platforms like HubSpot, understanding these changes is essential for maintaining attribution clarity and pipeline predictability.

Smart Bidding Exploration: 18% More Converting Queries

Announced at Google Marketing Live 2025, Smart Bidding Exploration allows advertisers to set a ROAS target tolerance (typically 10–30%), enabling Google's AI to bid on queries that might not meet standard ROAS thresholds but show high conversion potential. According to Google's internal data from March–April 2025, campaigns using Smart Bidding Exploration see an 18% increase in unique converting search query categories and a 19% increase in total conversions.

This is particularly valuable for UK B2B companies with long sales cycles. A professional services firm might normally bid only on high-intent terms like "HubSpot implementation partner London." With Smart Bidding Exploration, the AI can identify and test queries like "how to choose a CRM consultant"—capturing prospects earlier in their buying journey.

AI Max for Search: 14–27% More Conversions

Rolling out globally from late May 2025, AI Max enhances existing Search campaigns with three core capabilities: Search Term Matching (connecting ads to user intent beyond explicit keywords), Text Customisation (auto-generating headlines and descriptions from landing pages), and Final URL Expansion (selecting the most relevant landing pages automatically).

Google reports that AI Max delivers 14% more conversions at similar CPA/ROAS for typical campaigns, with campaigns still using mostly exact or phrase match keywords seeing up to 27% conversion uplift. For advertisers who've resisted broad match, AI Max offers a middle ground: enhanced reach with more transparency than Performance Max.

Performance Max: From Black Box to (Slightly) Grey Box

Performance Max remains Google's default campaign type for new advertisers, using AI to automate bidding, targeting, and placement across all Google channels (YouTube, Display, Search, Discover, Gmail, Maps). The 2024–2025 updates have addressed previous transparency criticisms with asset-level conversion reporting, channel performance reporting, and brand guidelines features.

The Power Pack Framework: Google's New Campaign Strategy

Google's 2025 strategy replaces the "Power Pair" with the "Power Pack"—three campaign types working together: Demand Gen (creates awareness and interest), AI Max for Search (captures high-intent searches), and Performance Max (drives conversions across all channels). This framework aligns with how Whitehat structures PPC management campaigns for clients needing full-funnel visibility.

Third-Party Tools That Fill Google's Gaps

Google's native AI is powerful, but it deliberately limits certain capabilities that agencies and in-house teams need for cross-channel management, predictive budget pacing, and impact simulation. Third-party tools fill these gaps—and for UK B2B companies managing integrated campaigns across marketing services, they're often essential.

Enterprise Solutions: Skai and Marin Software

Skai (formerly Kenshoo) leads enterprise solutions with Budget Navigator™ for ML-powered predictive forecasting, Impact Navigator for simulating outcomes before implementation, and "what-if" forecasting engines. The platform is trusted by Procter & Gamble, PepsiCo, and Nestlé for managing £1M+/month multi-channel campaigns.

Marin Software offers AI-based forecasting predicting clicks, conversions, revenue, cost, and profit using proprietary campaign-level models. Their 2024 updates included Marin Advisor—an AI teammate with chat interface for analysing performance. Case study results show 94% of campaigns hitting budget targets and 20% revenue lift within one month.

Agency-Focused Tools: Optmyzr and Adalysis

Optmyzr emerges as the most frequently recommended tool for agencies, offering budget forecasting with projected spend reports, AI Sidekick for natural language diagnostics, and custom automation rules. Whitehat has seen case studies including Colewood Digital achieving 15% revenue increase plus 2,900+ hours saved in 2024, and Mapplinks generating $1M+ revenue on $250K ad spend.

Adalysis (starting from £79/month) provides budget tracking with seasonality forecasting, spend projections, and built-in generative AI for ad groups and keywords. Users report saving 40+ hours of work per month—time that can be redirected to strategy and client communication.

Real Performance Statistics: What AI-Powered PPC Actually Delivers

The data from multiple sources confirms that AI-powered PPC consistently outperforms manual optimisation—though the degree of improvement varies significantly based on implementation quality, data volume, and human oversight.

Aggregate Performance Data

• 30% higher conversion rates for AI-driven PPC campaigns (Premiere Creative 2024)
• 20% increase in ROI from automated bidding versus manual methods (Octoboard PPC Survey 2024)
• 40% decrease in CPA within first month using Smart Bidding (Columbus Agency/Google Marketing Platform)
• Over 70% of marketers report automation has reduced time spent on campaign management

Verified Case Studies

Industry Adoption Rates

Over 80% of Google advertisers now use some form of Smart Bidding (Google Bidding Hub/Market Vantage 2024–2025). The State of PPC Global Report 2024 (1,135 verified respondents) shows 63% expect budget increases for Performance Max campaigns. The industry has clearly moved past the "should we use AI?" question—the focus is now on strategic implementation.

How Generative AI Is Changing PPC Workflows

Frederick Vallaeys, voted #1 most influential PPC expert, identifies effectiveness as "largely divided by those who use GenAI well vs. those who don't." Beyond Google's native AI, tools like ChatGPT, Claude, and Gemini are transforming how marketers approach keyword research, script automation, and campaign analysis.

High-Value Use Cases for Generative AI in PPC

• Keyword development: Generating comprehensive keyword lists from seed terms, brainstorming synonyms and long-tail variations. Multiple experts rank this as the highest-value use case.
• Script automation: Developing Google Ads scripts through natural language prompts for bid adjustments, campaign pauses, and budget reallocations—cited as saving "thousands of hours."
• Data analysis and forecasting: Analysing historical PPC data, identifying patterns, and forecasting seasonal demand shifts without requiring deep statistical expertise.
• Ad copy refinement: Using AI to refine and polish human-generated ideas rather than creating from scratch. Vallaeys emphasises "use AI to refine, rather than create."

Google's Conversational Experience for Search Campaigns allows advertisers to enter a landing page URL whilst AI generates keywords, headlines, descriptions, and images. Small business advertisers using it are 63% more likely to publish campaigns with 'Good' or 'Excellent' Ad Strength.

Critical Limitations That Demand Human Oversight

The expert consensus from Frederick Vallaeys, Brad Geddes, and other practitioners is clear: AI enhances rather than replaces human strategy. Understanding where AI predictions fail is as important as knowing where they excel.

Forecasting Accuracy Constraints

Historical data dependence represents the primary limitation. As Lunio notes: "Results are based on historical data. Things can happen in the future that are not accounted for"—including weather events, new competitors, market changes, and new account managers.

Frederick Vallaeys quantifies the accuracy challenge: "Even the best GenAI models only get things right around 80% of the time, and most models perform much worse than that. LLM outputs must be reviewed and double-checked for accuracy."

Performance Max "Black Box" Concerns

Despite 2025 transparency updates adding channel reporting, advertisers still cannot control budget allocation across channels versus visitor types or assess which Asset Groups perform best. This creates platform dependence—without understanding AI logic, advertisers essentially "start over" when implementing other campaigns.

Situations Where AI Predictions Consistently Fail

Human intervention remains essential for:

• New campaigns and product launches lacking historical data
• Major budget changes that destabilise algorithms
• Niche or low-volume accounts with insufficient data for optimisation
• Website migrations breaking tracking and tag replacements
• Regulated industries where AI frequently misses compliance requirements

Ben Wood (Hallam) cautions: "As more advertisers rely on the same AI systems, we risk seeing a wave of similar ads that prioritise budget over quality. Buyers will learn to tune out these ads." This homogenisation risk is particularly relevant for B2B companies needing to differentiate in competitive markets.

Best Practices for AI-Powered PPC Forecasting

Data Requirements for Accuracy

• Minimum 60 conversions per month for single Performance Max campaign optimal decision-making
• 3–4 weeks lead time before major sales events for automated campaigns to learn
• 30+ conversions monthly minimum (50 for Target ROAS) per Google recommendations
• Historical data period should match the period being forecast

Recommended Forecasting Methodology

AgencyAnalytics recommends mixing three approaches:

Human-AI Collaboration Framework

Brad Geddes' evaluation framework for AI recommendations:

1. Does this fit how we think accounts should be optimised?
2. Should we test it first rather than implement blind modification?
3. If successful, roll out gradually—not account-wide
4. Document: what, why, and when

Implement a 70/20/10 budgeting model: 70% core campaigns, 20% promising new tactics, 10% experimental approaches. This balances proven performance with strategic exploration—the same principle behind Smart Bidding Exploration, but applied at the portfolio level.

What's Changed: Key Developments Making 2023 Content Outdated

If you're reading PPC guides published before 2024, they likely need significant updates. Here are the major changes that have rendered older content obsolete:

1. Enhanced CPC (ECPC) references—deprecated 31 March 2025
2. "Power Pair" strategy—replaced by "Power Pack" framework
3. Performance Max transparency—now includes asset-level and channel reporting
4. AI Max for Search campaigns—entirely new feature launched 2025
5. Smart Bidding Exploration—"biggest bidding update in over a decade"
6. Brand guidelines and High Value Mode—new advertiser controls
7. Modeled conversion improvements—70%+ attribution recovery rates
8. Generative AI integration—Imagen 3, conversational campaign creation

Frequently Asked Questions

The Bottom Line: Strategic AI Implementation Wins

The AI-powered PPC landscape in 2024–2025 represents a fundamental transformation. Google's native AI capabilities have matured significantly, with Smart Bidding Exploration, AI Max for Search, and enhanced Performance Max controls addressing previous transparency criticisms whilst delivering documented performance improvements of 14–27%.

Third-party tools remain essential for agencies and in-house teams requiring cross-channel visibility, predictive budget pacing, and impact simulation capabilities that Google's native tools deliberately limit. For UK B2B companies integrating PPC with HubSpot CRM and marketing automation, these tools provide the attribution clarity that CFOs demand.

The evidence strongly supports AI adoption—case studies demonstrate 40–76% improvements in key metrics and 80%+ industry adoption of Smart Bidding. But the expert consensus emphasises that AI enhances rather than replaces human strategy—particularly for new campaigns, low-volume accounts, and regulated industries.

Amy Hebdon (Paid Search Magic) offers the definitive balance: "There are many ways to use generative AI to enhance your campaigns—and only two ways to get it wrong: Blindly rely on it for everything. Refuse to use it for anything."

References & Further Reading

• Google Smart Bidding Overview: support.google.com/google-ads/answer/7065882
• Google Smart Bidding Exploration Announcement: blog.google/products/ads-commerce/smart-bidding-exploration-ai
• Search Engine Land — Smart Bidding Exploration: searchengineland.com/google-smart-bidding-exploration-455756
• Google Marketing Live 2025 Recap: business.google.com/us/think/ai-excellence/google-marketing-live-2025
• WordStream 2024 Google Ads Benchmarks: wordstream.com/blog/2024-google-ads-benchmarks
• Search Engine Land — AI Tools for PPC: searchengineland.com/ai-tools-ppc-ai-search-social-campaigns-465779
• State of PPC Global Report 2024: swydo.com/blog/state-of-ppc-global-report-2024
• Optmyzr PPC Monitoring Tools: optmyzr.com/blog/ppc-monitoring-tools