How to Set Smart Bidding Targets for AI Max: tROAS, tCPA, and Maximize Conversion Value Explained

Understanding AI Max Smart Bidding: Which Strategy Fits Your Goal

AI Max campaigns run on Google's most aggressive automation layer, combining broad match signals, automatically created assets, and audience expansion into a single system. The bid strategy you attach to that system acts as its compass — it tells the algorithm what to optimize toward and how aggressively to compete in auctions.

Three bid strategies are available inside AI Max, and they serve fundamentally different purposes:

• Target ROAS (tROAS): Tells the algorithm to generate a specific return on ad spend. Best suited for ecommerce and lead gen accounts where conversion values are tracked precisely.
• Target CPA (tCPA): Instructs the system to acquire conversions at a defined cost per acquisition. Works well when all conversions carry roughly equal value and volume is the priority.
• Maximize Conversion Value: Spends the full budget to extract the highest possible total value from conversions, with no floor on efficiency. Used when you're scaling or when you don't yet have enough data for a meaningful tROAS target.

"The bid strategy doesn't control what the algorithm does — it controls what the algorithm is rewarded for doing. Set the wrong reward, and you'll get the wrong behaviour at scale."

For a broader look at how these strategies fit into a longer-term plan, the Smart Bidding strategy 2026 guide covers layering signals and maintaining control as automation scales across your full account.

Prerequisites Before You Touch a Bid Target

Setting a bid target before your account is ready is one of the most common ways to send an AI Max campaign into a death spiral of under-spending and poor signal quality. Before you configure any bid strategy, confirm all of the following are in place:

If any of these are missing, fix them before setting targets. Launching with an aspirational tROAS on top of broken conversion tracking will teach the algorithm to chase the wrong signals — and it takes weeks to recover from that.

Step 1: Audit Your Conversion History and Establish a Baseline

Your bid target should reflect your actual historical performance, not an internal KPI or wishful thinking. The baseline you calculate here becomes the anchor for everything that follows.

• Open Google Ads and navigate to Tools → Measurement → Conversions to confirm which conversion actions are set as "primary" — these are the only ones that feed Smart Bidding signals.
• Pull a 90-day performance report at the campaign or account level, filtering for the conversion actions you plan to optimise toward.
• Calculate your actual average CPA: total spend ÷ total conversions over the period.
• Calculate your actual average ROAS: total conversion value ÷ total spend, expressed as a percentage or ratio (e.g. 400% or 4:1).
• Identify any seasonal distortions — if your 90-day window includes a sale period or low-traffic month, weight the more representative weeks more heavily in your baseline calculation.
• Note the standard deviation of your CPA or ROAS week over week. If it swings wildly, you need more volume before setting a tight target.

Industry practitioners commonly report that accounts which set their initial tCPA within 10–20% of their historical average CPA see significantly faster exits from the learning period compared to those that set aggressive stretch targets from day one.

Step 2: Choose the Right Bid Strategy for Your Campaign

With your baseline in hand, you can now make an informed strategy selection. The decision tree is straightforward but worth working through deliberately rather than defaulting to what sounds most sophisticated.

• Use tROAS if: You sell products or services at varying price points, you track revenue or conversion value dynamically, and you have at least 50 conversions with value data in the last 30 days.
• Use tCPA if: All your conversions represent roughly equal value (e.g. lead form submissions, app installs, phone calls), you have 30+ conversions in the last 30 days, and efficiency per conversion is your primary goal.
• Use Maximize Conversion Value (no target) if: You're launching a new campaign, you have fewer conversions than the thresholds above, or you're in an aggressive growth phase where efficiency can be secondary to volume for a defined period.
• Navigate to your AI Max campaign in Google Ads, click Settings, then scroll to the Bidding section and click the edit icon.
• Select your chosen strategy from the dropdown. If you're switching from an existing strategy, Google will show a warning about a new learning period — acknowledge it.

The full setup walkthrough for Google Ads AI Max campaigns covers the campaign-level settings in detail, including how bid strategy interacts with asset generation and audience controls.

Step 3: Set Your Initial tROAS or tCPA Target

This is the step where most advertisers make their critical error: they set the target they want rather than the target the data supports. The algorithm is not a negotiation partner — it will simply restrict spend if the target is unachievable, or chase volume indiscriminately if the target is too loose.

• For tCPA: Enter a target that is no more than 10% below your calculated 90-day average CPA. For example, if your historical CPA is £45, start your tCPA at £40–£45. You can tighten it incrementally once the campaign has exited the learning period.
• For tROAS: Enter a target no more than 10–15% above your historical average ROAS. If your actual ROAS is 380%, start at 380–400%. Setting it at 600% when the account has never achieved that will cause severe impression throttling.
• In the bidding settings panel, type your target value into the tCPA or tROAS field — Google accepts tCPA in your account currency and tROAS as a percentage (e.g. enter "400" for a 4:1 ROAS).
• Do not set a bid limit (maximum CPC cap) alongside tROAS or tCPA unless you have a specific floor on profitability. Bid caps frequently conflict with Smart Bidding's auction-time decisions and reduce scale unnecessarily.
• Save your settings and record the date — this is day zero of your learning period.

"A tROAS target set 50% above your actual historical ROAS isn't ambitious — it's a signal to the algorithm to stop spending. Ambition belongs in your creative and product; your bid target belongs in your data."

Step 4: Configure Maximize Conversion Value When You Have No Target Yet

Maximize Conversion Value without a tROAS target is frequently misunderstood as a "beginner" setting — it's actually the correct choice for any campaign that lacks sufficient conversion history for a meaningful target. Used deliberately, it's a powerful data-gathering and scaling tool.

• Select Maximize Conversion Value from the bid strategy dropdown in campaign settings.
• Leave the optional tROAS field blank. Adding a tROAS here converts the strategy into tROAS — only do this once you have the data to support a specific target.
• Set a daily budget that reflects genuine business capacity. Maximize Conversion Value will attempt to spend the full budget every day, so if you set £500/day, ensure you can handle the conversion volume and fulfillment that spending might generate.
• Run the campaign on Maximize Conversion Value for a minimum of 4–6 weeks, or until you accumulate at least 50 conversions with value data. This builds the signal pool the algorithm needs before you layer in a tROAS constraint.
• Monitor your implied ROAS weekly (conversion value ÷ spend) during this period. When it stabilises across two or three consecutive weeks, you have a meaningful baseline from which to set your first tROAS target.
• To graduate the campaign to tROAS, edit the bidding settings and enter a target equal to or slightly above your observed average. The transition will trigger a brief learning period but should stabilise within 1–2 weeks given the existing campaign history.

Step 5: Layer Budget Constraints and Monitor the Learning Period

The learning period is not a passive waiting game — it's an active monitoring phase where your job is to keep the conditions stable so the algorithm can calibrate effectively.

• Keep your daily budget unchanged for the first 2–3 weeks after setting or changing a bid target. Significant budget changes reset or extend the learning period, compounding the disruption.
• Avoid editing ad assets, audience signals, or campaign settings during the learning period. Each substantive change can restart the calibration clock.
• Check campaign status daily in the Status column of your campaign list. A "Learning" badge is expected for up to 7 days; if it persists beyond 14 days, investigate — common causes include budget constraints, very low search volume, or conversion tracking gaps.
• Use the Bid Strategy Report (found under Tools → Bid Strategies, or directly in the campaign's Recommendations tab) to see how often the algorithm is hitting your target and where it's constrained.
• After the learning period ends (typically 7–14 days for established campaigns), review performance against your target for a full week before making any adjustments.
• When adjusting targets after the learning period, move in increments of no more than 10–15% at a time, and wait at least 2 weeks between adjustments to allow the system to respond before you change direction again.

Common Mistakes to Avoid

These are the errors that most consistently derail AI Max Smart Bidding setups, based on patterns reported across accounts at scale:

• Setting a target below your actual CPA or above your actual ROAS from day one. The algorithm interprets an unreachable target as a signal to withhold bids, causing impression share to collapse. Always anchor to historical reality first.
• Switching bid strategies mid-learning period. Each strategy change resets learning. If your tCPA isn't performing after five days, that's not failure — that's the normal calibration window. Intervening early is almost always counterproductive.
• Running tROAS on a campaign with no conversion value tracking. Without dynamic value data, the algorithm has nothing to optimise against. All conversions will appear as zero value, and tROAS will fail silently by under-bidding on every auction.
• Using portfolio bid strategies across campaigns with very different conversion rates. AI Max campaigns generally perform better with campaign-level bid strategies because the signals from one campaign don't contaminate another.
• Ignoring the "Limited by budget" warning alongside a tROAS or tCPA target. A constrained budget tells the algorithm it can't explore enough auctions to hit your target efficiently. Either increase the budget or accept a higher CPA/lower ROAS as a result.
• Treating Maximize Conversion Value as a permanent strategy. It's a ramp, not a destination. Without a value target, the algorithm optimises for absolute value with no efficiency floor — fine during scaling, unsustainable as a long-term efficiency strategy for most accounts.

Expected Results and Timeline

Realistic expectations are what separate accounts that stay the course from those that panic and make changes that compound the problem. Here is a framework for what to expect at each stage of your AI Max Smart Bidding setup:

Practitioners working with AI Max at scale commonly observe that campaigns allowed to complete a full 4-week stabilisation period without interference consistently outperform those where targets were adjusted multiple times in the first two weeks. Patience is a genuine competitive advantage in automated bidding.

Frequently Asked Questions