Google reported an 18% average lift in conversions for Performance Max advertisers at comparable CPA, up from 13% the previous year. This suggests a strong case for activating every AI feature available. However, this figure conceals a prerequisite that many B2B SaaS teams have yet to meet: the measurement and value architecture underneath.
If your account is optimizing toward form fills while your sales team pursues qualified pipeline, you don’t have a bidding problem; you have a signal problem. AI Max, Performance Max, and Demand Gen can all scale the wrong outcome faster than any human media buyer.
The Maturity Question Nobody Wants to Answer Honestly
Avinash Kaushik recently published a Google Ads maturity model based on two scoring axes: capability (how sophisticated your setup is, 0–4) and depth (how much of your spend runs through that setup, 0–4). This framework encourages honesty. A pilot running AI Max on 8% of non-brand spend isn’t “AI-native”; they’re a tourist.
Kaushik’s anti-deception rule is crucial: if your depth score is 0 or 1, your capability score can’t exceed 2. Score depth before capability to prevent ego from skewing the assessment. Audit the top 80% of spend, not just edge cases.
The model encompasses six dimensions: measurement and value architecture, search operating model, first-party data and audience intelligence, surface breadth and campaign mix, creative and landing page adaptability, and operating cadence and governance. Measurement carries the most weight at 30 points out of 100. This is appropriate; everything else is built on sand if the reward function is flawed.
Measurement First, Automation Second
For B2B, the critical diagnostic is: what percentage of your conversions are tied to actual business value—revenue, closed-won deals, predicted LTV—versus raw lead volume? And what share of those conversions feed back into Google through enhanced conversions or offline conversion imports via Data Manager API?
Google has deprecated the older UploadClickConversions path for many integrations. The recommended approach is now Data Manager API plus Enhanced Conversions for Leads. If your CRM-to-Google pipeline relies on legacy systems, your bidding algorithms are working with incomplete data. They’ll optimize, but toward whatever signal they can see, which might be the cheapest form fill from the least qualified prospect.
Google’s guidance links improving Responsive Search Ads Ad Strength from "Poor" to "Excellent" with about 12% more conversions on average. However, for a B2B team with a 90-day sales cycle, the key question isn’t whether you received more form fills, but whether those fills converted into pipeline 60 days later.
What Changes with AI Max and DSA Retirement
Dynamic Search Ads are set to retire in September 2026, transitioning into AI Max. If your long-tail query capture strategy relies on page-based DSA matching, this is a concrete planning window. AI Max offers broader intent matching, dynamic ad copy generation, and Final URL Expansion, providing more reach but less manual control.
The trade-off is significant: AI Max can expand coverage into queries you wouldn’t have bid on manually, potentially directing traffic to unintended landing pages. Mature teams need landing-page governance and measurement guardrails before relinquishing control. Immature teams will discover the impact on pipeline quality three months later.
Meanwhile, Ask Advisor (Gemini-powered) now integrates Google Ads, Analytics, Merchant Center, and Google Marketing Platform from a single AI interface. Business Agent for Leads places a chat agent inside the ad itself, currently in beta in India and testing in the U.S. for English-language accounts. Both introduce new measurement and routing requirements that most marketing ops teams haven’t yet addressed.
The Honest Self-Assessment
Before activating AI Max or scaling Performance Max, use this checklist for your account:
- Conversion definition: Are you bidding toward qualified pipeline stages or top-of-funnel form fills?
- Offline feedback loop: Is CRM data flowing back to Google via Data Manager API + Enhanced Conversions for Leads?
- Value signals: Does Google differentiate between a $500 lead and a $50,000 deal?
- Landing-page governance: If Final URL Expansion is enabled, do you control which pages are served?
- Depth reality check: What percentage of your spend actually runs through these systems—not just a test campaign?
BCG's maturity research identifies accelerators as first-party data, end-to-end measurement including predictive models, agile test-and-learn loops, and the right skills. This list isn’t aspirational; it’s essential for making AI features effective.
The previous question was whether a skilled media buyer could out-manage the auction. That debate is over. The new question is whether your organization can provide the machine with better truth—better conversion signals, better creative assets, better value definitions—and allow it to learn. Kaushik’s model targets a score of 85 out of 100. Most accounts, when scored honestly, fall between 25 and 45.
The gap between those scores is where the real work lies—not in toggling AI features on, but in building the data architecture that makes them worthwhile.
