AI Max Smart Bidding: When to Use Target CPA vs Target ROAS and How to Set Initial Targets

Smart Bidding strategy selection is one of the first decisions in AI Max campaign setup, and it's one that's difficult to reverse cleanly once the algorithm has accumulated learning data. Getting the initial strategy and target right avoids the learning phase penalty of making major bidding changes later.

Target CPA vs Target ROAS: The Fundamental Difference

Both Target CPA and Target ROAS are automated bidding strategies that use Google's machine learning to set bids at each auction. The difference is in what they optimize toward:

Target CPA (Cost Per Acquisition):

• Optimizes to get as many conversions as possible at or below your target cost
• Best when each conversion has approximately equal value
• Requires: clear conversion definition, consistent conversion value

Target ROAS (Return on Ad Spend):

• Optimizes to maximize conversion value relative to spend
• Best when conversions have variable value (e-commerce with different product prices)
• Requires: conversion value tracking (revenue values passed to Google Ads)

The practical test: if all your conversions are worth approximately the same (lead form submissions for a service business, consultation bookings at a fixed price), use Target CPA. If conversions have different values (e-commerce purchases of $20 or $500 depending on what was bought), use Target ROAS.

When Target CPA Works for AI Max

Target CPA is the right choice when:

• You're generating leads (form fills, phone calls, consultation bookings)
• You're optimizing for app installs at a specific cost
• Your product has a fixed or near-fixed price point
• You don't have revenue value data available to pass to Google Ads

Setting the initial CPA target:

The common mistake is setting the target too aggressively (too low) at launch. An unrealistically low CPA target forces the algorithm to either restrict volume significantly or underdeliver.

Framework for initial CPA target:

1. Calculate your acceptable economics (what's the maximum you can spend per acquisition while remaining profitable)
2. Set initial target 20-30% above your minimum acceptable CPA
3. After 4-6 weeks with sufficient conversion volume (30+ conversions), evaluate: if actual CPA is consistently below target, lower the target by 10-15%; if actual CPA is at or above target with low volume, the target may need to increase or the campaign structure needs review

For new campaigns with no historical data: start with a "Maximize Conversions" bid strategy for the first 4-6 weeks to accumulate conversion data, then switch to Target CPA once you have 30+ conversions to inform the target.

When Target ROAS Works for AI Max

Target ROAS is the right choice when:

• You're running e-commerce campaigns with variable purchase values
• Your conversions have different economic value (premium vs. standard purchases)
• You have revenue tracking properly implemented in Google Ads

For AI Max Shopping (https://yositeup.com/blog/google-ai-max-shopping-replacing-performance-max-2026), Target ROAS is almost always the appropriate strategy because product prices vary and shopping campaigns typically measure purchase revenue directly.

Setting the initial ROAS target:

ROAS is expressed as a multiplier or percentage. A 300% ROAS target means for every $1 spent, you want $3 in conversion value.

Calculate from your economics:

• Gross margin on products: e.g., 40%
• Overhead/ops cost as % of revenue: e.g., 15%
• Target net margin: e.g., 10%
• Maximum marketing cost as % of revenue: 40% - 15% - 10% = 15%
• Minimum ROAS needed: 1 / 0.15 = 667% (or 6.67x)

Set initial target at your minimum viable ROAS. If you want to grow volume and are willing to accept lower margin initially, set it lower. If you need to maintain profitability immediately, set it at or above minimum viable.

The DSA Migration Context for Bidding

The shift from DSA to AI Max (https://yositeup.com/blog/google-ads-dsa-ai-max-migration-february-2027) often involves a bidding strategy change as well. DSA campaigns commonly used Target CPA. When migrating to AI Max:

• If your DSA campaign was performing at a stable CPA for 6+ months, use that historical CPA as the starting point for your AI Max Target CPA
• Add a 15-20% buffer to account for the learning phase — AI Max will be less efficient initially as it builds audience and asset serving data
• Don't set AI Max Target CPA to exactly match your DSA CPA — the learning phase will trigger underdelivery if the target is set too tightly from day one

Bidding Adjustments and Targets Over Time

After the initial learning phase, bidding targets should be adjusted based on actual performance data, not arbitrary schedules.

Triggers for lowering CPA target (increasing efficiency demand):

• Consistent 4+ week window where actual CPA is 15%+ below target
• Campaign is hitting budget limits consistently (sign that there's more volume available at lower cost)
• You've improved landing page conversion rate

Triggers for raising CPA target:

• Conversion volume is lower than needed despite budget being available
• Campaign is not spending full budget (bidding constraints may be too tight)
• Business can profitably absorb higher acquisition cost for growth

Change magnitude: Adjust targets 10-15% at a time. Large jumps (>20%) trigger another mini-learning phase and can destabilize well-performing campaigns.

Wait minimum 2 weeks after each adjustment before evaluating impact.

July 2026 ToS Compliance with Smart Bidding

The July 2026 ToS update (https://yositeup.com/blog/google-ads-tos-july-2026-ai-automation-what-changed) includes requirements around automated bidding disclosures in certain industries. For financial services, healthcare, and other regulated verticals using Smart Bidding, verify that your bidding configuration doesn't create disclosure issues with your compliance team.

The update also clarified requirements around third-party bidding tools and scripts. If you're using external bid management software alongside AI Max's native Smart Bidding, ensure your configuration doesn't violate the ToS requirements for automated systems.

Portfolio Bidding for Multiple AI Max Campaigns

Portfolio bidding strategies apply a shared Target CPA or Target ROAS target across multiple campaigns. The algorithm distributes spend toward whichever campaign within the portfolio is currently hitting the best performance.

Portfolio bidding is appropriate when:

• You have 3+ campaigns with similar conversion types
• You want the algorithm to dynamically shift budget toward the best-performing campaigns
• Your campaigns have seasonal variation where different campaigns peak at different times

Portfolio bidding is not appropriate when:

• Campaigns have fundamentally different conversion values or types
• You need to maintain minimum spend in specific campaigns regardless of performance
• You're in a regulated industry where per-campaign bidding control is required for compliance

Reporting and Attribution Impact on Bidding

The June 2026 reporting data deletion (https://yositeup.com/blog/google-ads-reporting-data-deleted-june-2026) affected historical conversion data for some accounts. If your CPA or ROAS targets were set based on historical averages that are now missing, re-derive your targets from current data windows.

Use a minimum 60-day window for calculating reference CPA/ROAS targets. Shorter windows can include seasonal anomalies or promotional periods that skew the target calculation.

Data-driven attribution (available for accounts with sufficient conversion volume) provides more accurate CPA/ROAS reporting than last-click by distributing credit across the full conversion path. If you're making bidding target decisions based on Google Ads conversion reports, ensure you understand which attribution model your reports are using.