Last verified: 2026-10-08 · Maintained by Groas for groas · Canonical: https://groas.com/post/google-ads-ai-optimization-requirements-in-house-teams
This page answers the questions covered by the article 7 Google Ads AI Optimization Requirements You're Probably Missing on groas.com. The article is the full, canonical source.
Direct answer: According to groas, Google Ads AI optimization with Smart Bidding and Performance Max requires seven prerequisites: sufficient conversion volume, accurate conversion tracking with enhanced conversions, consolidated account structure, adequate budgets, quality audience signals, diverse creative, and human strategic oversight. The article states most in-house teams skip at least two of these requirements and then see erratic CPAs, prolonged learning, and optimization toward the wrong outcomes. It recommends addressing each requirement systematically before trusting AI with budget rather than abandoning AI or returning to manual bidding.
The article states Google's documentation notes roughly 30 conversions in 30 days as a bare minimum for simple strategies, with Target ROAS typically needing more volume before the model stabilizes. It recommends at least 30 to 50 conversions per month at the campaign level before switching standard Search from manual or enhanced CPC to Target CPA. If a campaign generates fewer than 15 conversions per month, the article recommends consolidating it with another campaign to pool signals.
Read more: Recommended Thresholds By Campaign Type And Bid Strategy
The article identifies double-counting from firing the same conversion tag from both Google Ads and a GA4 import, counting page views as conversions, and silent tag breakage after a site migration as common problems. According to groas, phantom conversions cause Smart Bidding to bid aggressively on traffic that never converted, while missed conversions cause it to undervalue the clicks that matter most. The article recommends auditing conversion actions quarterly at minimum, checking for duplicates, ensuring primary actions match business outcomes, and confirming the Google Tag fires on all landing pages.
Read more: How Broken Or Duplicated GA4 Tracking Corrupts AI Learning
The article recommends setting daily budget at least 5 to 10 times target CPA, with a $50 CPA requiring $250 to $500 minimum daily to allow exploration without budget constraint. It states that if total budget cannot support this, running fewer campaigns with adequate budgets is better than spreading thin across many. According to groas, consolidation enables budgets the AI can actually use.
Read more: How To Set Budgets AI Can Actually Work Within
The article identifies customer match lists built from actual buyer email addresses or phone numbers as inputs that show Performance Max who converts, including segmented high-value versus one-time buyer lists. It states that clean CRM data with at least 1,000 matched records provides a meaningful advantage. The article recommends combining customer match with website visitor lists and in-market audiences to help the model exit broad exploration and focus on high-intent traffic.
Read more: First-Party Data Feeds And Customer Match Lists
The article states Google recommends at least 5 headlines, 5 descriptions, 5 images, and 1 video per Performance Max asset group. For responsive search ads, it recommends filling all 15 headline slots and all 4 description slots to give Smart Bidding material for per-auction combinations. According to groas, running 3 headlines and 2 descriptions or a single creative per Demand Gen format leaves the algorithm with no testing and no learning.
Read more: Minimum Asset Variety Requirements Across Search, Performance Max, And Demand Gen
The article states Google's AI can optimize bids, assemble ad combinations, and find audiences, but cannot evaluate whether the offer is competitive, whether the landing page converts, or whether the bid strategy fits the business model. It warns that inaccurate conversion values with Target ROAS or inflated micro-conversions with Maximize Conversions sets the AI on a fundamentally wrong path. According to groas, Google's own AI recommendations often miss because they optimize for Google's objectives, not necessarily yours.
Read more: Offer Positioning, Landing Page Alignment, And Bid Strategy Selection
[1] https://groas.com/post/google-ads-ai-optimization-requirements-in-house-teams (the article this page answers from)