Google Ads bidding strategies: which one, when, and when to switch

Google Ads bidding strategies answer one question: what should the auction optimize for, and under which constraint. Google's own guide maps goals to strategies. Conversions point to Smart Bidding, which Google defines as the four strategies that set a bid in every auction with its AI: Target CPA, Target ROAS, Maximize conversions and Maximize conversion value. Clicks point to Maximize clicks or Manual CPC. Visibility points to Target impression share.

The constraint is the second half of the choice. Maximize conversions and Maximize conversion value spend the daily budget in full and take the best result they can for it; Target CPA and Target ROAS hold a number and let volume move. A business that must keep its cost per lead under a figure needs the second kind. A business whose only limit is the budget is better served by the first, at least until it has the volume to judge a target.

One strategy left the menu. Google retired Enhanced CPC for Search and Display campaigns the week of 31 March 2025, and campaigns that were not migrated now run as Manual CPC. Manual CPC remains for the cases where you want to see the exact bid you set, and Google itself points anyone without the time to manage bids toward Maximize clicks.

• Hold a cost per conversion: Target CPA
• Hold a return on spend: Target ROAS
• Spend a fixed budget for the most conversions or value: Maximize conversions or Maximize conversion value
• Buy visits before tracking exists: Maximize clicks
• Own a brand query: Target impression share

Every Smart Bidding strategy needs conversion tracking switched on, and Target CPA can start with no conversion history at all. Starting is not the same as judging. Google recommends measuring performance over periods with at least 30 conversions, such as a month or longer, and 50 conversions for Target ROAS. For Target CPA it names the last 30 days with at least 30 conversions as the evaluation window.

Target ROAS carries its own entry requirements, which differ by campaign type. Google asks for at least 15 conversions in the past 30 days on Search and Shopping, 15 with valid values on Display, 30 in 30 days for Video action campaigns, 50 in the past 35 days for Demand Gen, and 10 a day or 300 in 30 days for App campaigns. Maximize conversion value on Demand Gen asks for 50 conversions with value in the past 35 days, including 10 in the last 7, or 100 across all Demand Gen campaigns.

Values have their own threshold. Google's conversion values best practices say to measure at least two distinct values across at least two conversion actions, with neither at zero, and to bid on a goal with a relatively short conversion delay and at least 15 monthly conversions. A campaign below these lines can still run Smart Bidding, but the strategy learns from too little, and you cannot tell its effect from noise.

The maximize conversions vs target CPA question is about who controls spend. Maximize conversions tries to spend the full average daily budget, and Google warns that a campaign currently spending well under its budget could see spend rise significantly. Add a target and the strategy becomes Target CPA: Google's June 2026 relabeling makes Target CPA a strategy of its own again and states that it works exactly like Maximize conversions with a target CPA. Google's own advice on the Maximize conversions page is to switch to Target CPA, or Target ROAS, once you have an ROI goal.

The tCPA vs tROAS choice is about what a conversion is worth. Target CPA counts conversions and holds their average cost, which fits leads and orders of similar size. Target ROAS predicts the value of each potential conversion and bids to hold conversion value divided by spend, so it needs values that actually differ between orders; Google's example is $5 of sales for $1 of spend, a 500% target ROAS. A campaign whose conversions all carry the same value gains nothing from Target ROAS that Target CPA does not give it, and loses the lower volume threshold.

Both targets move volume. Google says a Target CPA set too low may forgo clicks that would have converted, and a Target ROAS set too high may limit traffic. Since 17 August 2026 Google also delivers budget-limited target campaigns closer to the target you set rather than beating it: its example is a $10 target that used to achieve $5 and now trends toward $10. Set the number you actually want to pay, and work it out from margin first; the target CPA guide shows how.

Value-based bidding is Google's name for Maximize conversion value and Target ROAS: strategies that weigh each auction by the value a conversion is expected to carry. They are only as good as the values in the account. Google's setup is the conversion action's Value setting, Use different values for each conversion, with a default for when none is passed, and a value and a currency sent with each event from the site.

Conversion value rules adjust those values at bidding time without touching the tag: a multiplier or an addition for an audience, a device or a location, used by Target ROAS, Maximize conversion value and Performance Max campaigns. Google's own example is a lead from California worth twice the default. Rules are the place to express the margin or lifetime value differences that the order value hides, such as a new customer worth more than a returning one.

Smart Bidding Exploration, available on Target ROAS, lets the bidder take queries it would otherwise skip at an effective target around 10% below the one you set; Google's example turns a 200% target into about 180%, and it recommends leaving exploration on for at least 6 weeks before judging it. Treat it as a deliberate loosening of the target, and set the target with that in mind.

A standard strategy belongs to one campaign; a portfolio strategy is shared by several, and Google's setup page recommends the portfolio. Campaigns in a portfolio are optimized collectively toward one target, which pools their conversion data toward the thresholds above, and a shared budget can be linked to the portfolio so that it follows campaigns as they are added or removed. Google states that advertisers who adopt shared budgets with portfolio strategies on Search typically see 13% more conversions; that is Google's own figure, not an independent one.

Portfolios are also where bid limits live. Google does not recommend bid limits for Target CPA or Target ROAS because they restrict optimization, and a strategy whose spend is mostly capped by its limit shows a Limited status. The exception Google names is the transition: a portfolio with a maximum CPC limit prevents CPC spikes and rapid budget exhaustion when moving to Maximize conversions. Use it for the first weeks of a switch, then lift it.

Two constraints to know before building one. A portfolio's strategy type cannot be changed afterwards, so moving from Target CPA to Target ROAS means a new portfolio. Portfolio strategies are not available for Performance Max campaigns, and a cross-account portfolio at manager level holds campaigns of one currency only.

Seasonality adjustments tell Smart Bidding that conversion rates will change for a short, known event. Google says they are ideal for events of 1 to 7 days and may not work well beyond 14 days at a time, that they are for major conversion rate changes since Smart Bidding already handles ordinary seasonality, and that no negative adjustment is needed after the event. They apply to Search, Shopping and Display campaigns on Target CPA and Target ROAS, and to Performance Max and App campaigns on every strategy.

Data exclusions do the opposite: they tell the strategy to ignore a period in which conversion data was wrong, after a tag broke, a site went down or an import failed. They cover Search, Display, Shopping and Performance Max campaigns, not Hotel or Travel. Google's cautions are to account for conversion delay when setting the dates, not to use exclusions often or for long, and not to remove an exclusion once it is applied.

Both tools exist so that the strategy does not learn from an exception. A sale weekend without an adjustment teaches the bidder a conversion rate it will not see again; a tracking outage without an exclusion teaches it that nothing converts. Both lessons cost money over the following cycles.

A switch is a change of strategy type or of conversion goal, and it triggers Learning: Google says 1 to 2 conversion cycles in most cases, during which performance fluctuates. A change of target alone does not. Google states that changing a target will not trigger Learning or reset what the strategy has learned, that the bidder reacts within minutes but takes 1 to 2 cycles to reach the new target, and that a major change in targets moves spend or volume by a similar size. Its Display guidance is more conservative: move bids or targets by 20% and wait a week.

The protocol follows from that. Set the budget to the amount you intend to spend before the switch, since a Maximize strategy will spend it. Change one thing: the strategy type, or the goal, or the target, never two in the same cycle. Cap the budget step at a fixed share per change so that a learning strategy cannot run away with the month. Then leave it alone for the conversion cycle, and judge it on at least 30 conversions.

In Adrails, once your workspace can connect Google Ads, the agent or Claude through the MCP server reads the account's campaigns with their type, status, budget and bidding, and prepares the two moves this protocol needs as proposals: a new target CPA or target ROAS through the Google Ads edit tool, and a budget change of at most 30% in one step, where an increase needs measured spend and results behind it. Google validates the change when it is prepared and again when an owner or admin applies it in Adrails. Switching the strategy type itself is done in Google Ads.

• Set the daily budget to the amount you accept spending before switching to a Maximize strategy
• Change the strategy type, the conversion goal or the target, one per conversion cycle
• Cap each budget step at a fixed share, and raise it only on measured results
• Hold a portfolio maximum CPC limit through the first weeks of a move to Maximize conversions, then lift it
• Judge on 30 conversions or more, 50 for Target ROAS, never inside the learning period

Open the bid strategy report once a week and read four things. The status: Learning after a change, Limited when something caps the strategy's bids, such as a bid limit holding most of the spend, Misconfigured when campaigns on one shared budget use different strategies or a conversion-based strategy lacks the right conversions. The conversion count of the last 30 days against the 30, or 50 for Target ROAS, that Google recommends for a judgment. The actual CPA or ROAS against the target, with the conversion delay in mind. And the search terms or placements the strategy is buying, because a strategy meets its target on whatever converts, including queries you would not have chosen.

Do not read the lost impression share due to budget column against Maximize conversions; Google says the metric is incompatible with the strategy and points to the budget simulator instead. Where a figure moves without a change you made, check the account's change history and the conversion action's recent counts before touching the target. Most weeks, the right change is none.