Google Ads bidding strategies: which to choose and how to set targets

Google Ads bidding strategies decide who sets your bid in each auction and what that bid aims for. Three questions narrow the choice. Do you want more conversions, more conversion value, more traffic or more visibility? Does the account collect enough conversion data to feed that goal?

The third question applies if you set a target: does it come from your own margin and from how the account performs today? This guide orders the strategies by those decisions and works out a target with an example. It closes with how each common mistake shows up in your own account.

Platform facts here come from the Google Ads Help Center and Google's developer documentation. Google's pages were accessed on November 5, 2026.

Google separates bidding strategies by what they optimize, how many campaigns they cover and which campaign types offer them. Knowing these three splits makes the options in the interface easier to read.

Smart Bidding and everything else

Google's Smart Bidding page names four Smart Bidding strategies. They are Target CPA (cost per action), Target ROAS (return on ad spend), Maximize conversions and Maximize conversion value. Each sets a separate bid in every auction for conversions or conversion value. Google calls this auction-time bidding.

The rest chase clicks or visibility. Maximize clicks and Target impression share are automated, but they do not use conversion data. With Manual CPC (cost per click) you enter the maximum click bid yourself.

Enhanced CPC (ECPC) is gone from Search and Display campaigns since March 2025, according to Google's ECPC page. Campaigns that were not moved now run on Manual CPC. Some guides still list ECPC as an option.

Standard and portfolio strategies

A standard strategy applies to one campaign. A portfolio bid strategy groups several campaigns, ad groups and keywords under one target and lives in the Shared library. Google's portfolio definition says the four Smart Bidding strategies, Maximize clicks and Target impression share can all run as portfolios. Performance Max campaigns cannot use one.

What sets a portfolio apart is the shared target, not shared data. The Smart Bidding page says Smart Bidding can optimize based on data from all of your campaigns, so that needs no portfolio. In a portfolio, the Target CPA page says, campaigns, ad groups and keywords are optimized "collectively for a single target". The same page says separate ad group targets can restrict Smart Bidding.

What each campaign type offers

Not every strategy exists in every campaign type, and Google's pages do not always agree on which. The table shows the two value-based options in four campaign types, each cell based on a Google page.

The Performance Max conflict works like this. Google's Maximize conversion value page says the strategy is "only available for Search, Shopping, and Demand Gen campaigns". The Performance Max bidding step page recommends Maximize conversion value for Performance Max if you track values, and lists Target ROAS too. You see which options your own campaign offers when you build it.

The Shopping conflict sits inside the Target ROAS page. One note says a Target ROAS portfolio is not applicable to Shopping campaigns. Another paragraph says bid limits "are available for Search and Shopping portfolio bid strategies". Target impression share works on the Google Search Network only, excluding search partners.

Two changes in 2026

From June 2026, Google is renaming "Maximize conversions with a Target CPA" to "Target CPA". "Maximize conversion value with a Target ROAS" becomes "Target ROAS". Google says only the labels change and bidding behaviour stays the same. This guide uses the new names.

The second change affects behaviour. From August 17, 2026, budget-limited campaigns on a target-based strategy deliver closer to the target you entered. The example below works out what that means for your choice of target.

Choosing a bidding strategy rests on two questions. What do you want more of, and does the account produce enough data for that goal? The four goal types come first, then the data each strategy needs.

Conversions: Maximize conversions and Target CPA

If every conversion is worth about the same to you, there are two options. Lead forms with similar sales potential are a typical case. Maximize conversions spends the whole budget to get as many conversions as possible. If you have a specific cost goal, Google points you to Target CPA.

With Target CPA you enter the average amount you want to pay per conversion. Google's Target CPA page says some conversions may cost more than the target and some less. Altogether, it tries to keep cost per conversion equal to the target.

So the target is an average, not a ceiling. You can turn Target CPA on without any conversion history.

Conversion value: Maximize conversion value and Target ROAS

If orders differ in revenue or profit, value-based bidding fits better. Maximize conversion value spends the whole budget to get the highest total value. Target ROAS chases that value while trying to hit the average return you set. The Target ROAS page requires values on your conversions before you can apply it.

The same page suggests a sequence before you switch. Ideally you already bid to the same conversion goal with Target CPA. Then you report values for four weeks or one to two conversion cycles, whichever is longer. Adding a target ROAS to Maximize conversion value makes it behave like Target ROAS.

Traffic: Maximize clicks and Manual CPC

If the goal is visitors to your site, the click strategies remain. Maximize clicks sets click bids itself to get the most clicks within budget. Google's goal-based bidding page says it can suit new campaigns without enough conversion data. It is not a good fit when you need a specific cost per conversion.

With Manual CPC you set the maximum click bid per ad group or keyword. If you know which keywords are profitable and have time to watch bids, you keep the control. The CPC calculator helps you track what each click costs.

Visibility: Target impression share

Target impression share aims to show your ad at a share you set. The position can be the absolute top, among the top ads or anywhere on the results page. Google says it can be useful for campaigns with brand terms.

It looks at visibility only, not at the cost of a conversion. Google also warns that a low maximum CPC bid limit can stop the strategy reaching its impression share goal.

The data each strategy needs

The table compares the strategies by what they increase, what they require and what is most often missed.

The 15 conversions in the Target ROAS row apply to Search and Shopping campaigns. They are counted over the past 30 days at the conversion tracking level. Demand Gen and App campaigns have higher thresholds over different periods.

Evaluation has its own threshold. The Smart Bidding page recommends periods of a month or longer with at least 30 conversions, or 50 for Target ROAS. If a campaign stays below those numbers in a month, lengthen the evaluation period.

The quality of your conversion values

Value-based bidding is only as accurate as the values you send. Google says conversions with a value of zero do not count toward Target ROAS eligibility. It also says the value can be sales revenue or profit margin.

The strategy bids on the conversion goals and actions the campaign uses for bidding. In standard goals these are primary actions, the ones in the "Conversions" column. Google's primary and secondary conversion actions page names one exception. Actions in a custom goal are used for bidding whether they are primary or secondary.

So check what the value stands for. If you send the gross order amount at checkout, refunds, cancellations and tax stay inside it. A value-based strategy will then treat unprofitable orders as valuable. If you change how values are reported, Google suggests waiting one to two conversion cycles until the new values arrive at a similar rate.

What you base Target CPA or Target ROAS on is the decision that sets contribution after ad spend. Every figure in this section is hypothetical. The example uses the same brand as the performance marketing metrics guide. Its average order is $80 net of sales tax and returns, and each order contributes $25 before advertising.

Break-even sets the upper limit

Each order contributes $25, so the most you can spend on ads per order is $25. Break-even CPA is therefore $25. You can find your own contribution with the gross margin calculator. Average order value (AOV) comes from the AOV calculator.

The metrics guide's section "Set the bid target to match your data volume" covers how these thresholds enter the platform. This section goes one step further and asks which target makes total contribution largest.

The ROAS target depends on the value you send

Break-even ROAS changes with the value you send to the platform. If you send the $80 sales revenue, break-even ROAS is $80 divided by $25, which is 3.2. If you send the $25 contribution before advertising, the same economic limit is $25 divided by $25, which is 1.0.

Google Ads takes target ROAS as a percentage. The Target ROAS page says to multiply conversion value per cost by 100 to get it. Here the break-even target is 320% if you send revenue and 100% if you send contribution. The $15 CPA target used below equals 533% on revenue and 167% on contribution.

What happens when the target is too low

Google's Target CPA page says a target that is too low may forgo clicks that could convert, leaving fewer total conversions. The table shows the effect with a volume curve. The volumes are the kind of estimate a bid simulator or an experiment could show.

The model rests on three assumptions. The strategy hits its target on average, extra conversions bring extra orders to the business, and contribution per order stays the same across targets. Contribution after ad spend is conversions times ($25 minus the target CPA), and it is not net profit because fixed costs are not deducted.

The lowest target leaves the most per order, but 60 orders stop the total at $900. The break-even target brings the most orders and zero contribution after ads. Among the five targets examined, $15 gives the most. The table does not show whether a value in between would do better.

The reason is marginal CPA: what each extra conversion costs when you raise the target one step. The table divides extra spend by extra conversions at each step.

Only the first step keeps marginal CPA below the $25 contribution. At every later step, an extra order costs more than it contributes.

The rule follows from this. While the marginal CPA of extra conversions stays below contribution per order, raising the target grows the total. Once it rises above, raising the target shrinks it. The metrics guide's rule of keeping 30% of contribution as profit gives a $17.50 target here.

That target earns $1,200, below the $1,300 at $15. A fixed profit share works as a starting point, and the response of volume decides the final target.

Which assumptions change the answer

Two assumptions drive the result: contribution per order and how volume responds to the target. Below, order value and the assumption that the target is hit on average stay fixed. The first three rows keep the same volume curve and change only contribution.

The last row keeps $25 contribution with a flatter volume curve. Its cell lists the monthly conversions at the five target steps.

The target that gives the most rises with margin. When volume barely responds, a low target is enough. On the flatter curve, the first step's marginal CPA is exactly $25, equal to contribution, so the total does not change.

The table holds only under these assumptions. In your own account, only a bid simulator or an experiment that compares targets shows how volume responds.

The economic limit is not the starting target

Margin tells you what you can pay, not what target the account can reach today. Say the same brand's actual CPA over the last 30 days is $40. Each order then loses $15 after ad spend, but cutting the target straight to $15 may forgo clicks and cut volume hard, as Google warns.

So tie the target to three inputs. The first is the business's economic limit, $25 here.

The second is past performance with conversion lag complete. Google bases the recommended Target CPA on the last 30 days' average, adjusted for conversion delays. The Target ROAS page says to set the target from business goals and historical ROAS.

The third is how volume and cost respond when the target changes. A bid simulator or an experiment shows that. If actual CPA sits above the limit, the problem may not lie in the target alone. Keywords, ads and the site belong on the checklist too.

The August 17, 2026 update and a break-even target

Writing break-even into the target as a safety ceiling can cost money after the August 17, 2026 update. Google's page on changes to target-based strategies gives an example. A budget-limited campaign with a $10 target CPA and a $5 actual CPA now delivers closer to $10. Google does not change your targets for you.

Apply this to the example brand. The target is set at $25, the campaign is limited by budget and the monthly budget is $1,950. The table shows what the same budget buys when delivery moves toward the target.

Once CPA reaches break-even, each order's contribution goes to ads and contribution after ads falls to zero. So on budget-limited campaigns, the target should be the CPA you are willing to pay. App, Video reach and Video view campaigns keep the previous behaviour.

Lead accounts: form value times the chance of a sale

In an account that collects forms, the same logic links a form's value to its chance of becoming a sale. Say 20% of forms become sales and each sale contributes $250 before advertising. A form is then worth 0.20 times $250, which is $50, and break-even CPA per form is also $50.

If the close rate falls to 10%, a form is worth $25, and a target set from 20% pays twice too much per form. Work the rate out from forms whose sales cycle has finished, and track the steps from form to sale with the funnel analysis calculator. If forms differ in value, Google says transaction-specific conversion values can be used for leads too.

Setting up a strategy change as an experiment first lets you measure the result. Google offers campaign experiments, the learning status, seasonality adjustments and data exclusions for this.

Compare in a campaign experiment

Google's custom experiment page says an experiment shares the original campaign's traffic and budget to compare a change. In Search campaigns you can split by cookie or by search. Custom experiments exist for Search, Display, Video and Hotel campaigns, not for App or Shopping. In Target CPA or Target ROAS settings, "Save as experiment" starts a new target directly as an experiment.

Set up a strategy experiment in this order:

1. Pick one variable, such as only the target CPA or only the strategy. Google recommends doing this on a large-volume, long-running campaign.
2. Write down the success metric first. If you aim for contribution, measure total order contribution minus ad spend in each arm. With constant contribution, that is conversions times contribution per order minus ad spend. If the arms get different traffic shares, compare them per equal share of traffic.
3. Set the duration by two separate needs. One is enough observations, with conversion lag complete, to read performance. The other is a sample large enough to show the difference you expect. Google's 30 conversions guide evaluation, not a sample size for telling two strategies apart.
4. Change nothing else in either arm while it runs. Google says changes during an experiment make the results harder to interpret.
5. Pick the winner on the size of the difference, its uncertainty and its economic result together. Google lets you choose directional results or require significance, but significance alone does not say whether the difference pays.

A rough calculation shows why sample size is its own question. Take two arms with equal spend and identical true performance at about 30 conversions each. The 95% range for the gap between them is about plus or minus 15 conversions, half the average.

Showing a 20% difference at 95% confidence and 80% power needs about 430 conversions in the control arm. The calculation treats conversions as Poisson counts. For your own figures, use the sample size calculator.

The two arms work like a test group and a control group. The control group guide explains the logic. The statistical significance guide covers when a result is more than chance.

When the learning status starts

Google's bid strategy status page says the "Learning" status never lasts more than 7 days. It names three triggers. The first is a new or reactivated strategy, and the second is campaigns or keywords added to or removed from it. The third is a significant change to the conversion goals being optimized.

The same page says changes to budgets, targets, creatives or targeting never remove historical data or trigger "Learning". Hover over the status in your account to see which reason applies. A period with at least 30 conversions gives a firmer reading than the end of learning does.

Seasonality adjustments and data exclusions

For a short sale, a seasonality adjustment can replace a temporary target change. Google's seasonality adjustments page recommends it only when you expect major changes in conversion rate. Smart Bidding already handles seasonal events.

The tool suits short events of 1–7 days and may not work as well beyond 14 days. No reverse adjustment is needed after the event.

Data exclusions handle a past measurement fault instead. A removed tag, a site outage or a broken data import are typical cases. In those cases the data exclusions page lets you drop the affected clicks from Smart Bidding's data.

Exclusions affect bidding data only, and the conversions still appear in reports. The tool covers Search, Display, Shopping and Performance Max campaigns.

Do not limit the date range to the days the tag was broken. Exclusions work on clicks, so cover the click dates that could have led to the affected conversions, allowing for conversion delay. In Google's example, uploads had issues from October 15 to 18 with a 5-day delay, and the exclusion covers October 10 to 18.

Bidding mistakes usually leave a trace in a column or a status label, which a Google Ads audit also reads. Each check below says where the mistake shows and what that data cannot show.

Comparing actual CPA with the target you typed

Google recommends comparing actual CPA with the "Avg. target CPA" column, not with the target you entered. That column averages device adjustments, ad group targets and target changes over the period, weighted by traffic. Add it to the Campaigns page and set it beside actual CPA for the same date range.

This shows whether the strategy hit its target. It does not show whether the target is profitable, which needs a comparison with contribution.

Setting the target too tight

If Target CPA is too low or Target ROAS too high, Google may mark the campaign "Limited by bid strategy". Even without that status, the signs match. Spend stays well under budget, conversions fall and actual CPA sits under the target. A rise in impression share lost to rank can also be expected.

This data shows volume is being held back. It does not show whether a looser target would bring profitable extra orders. Marginal CPA and an experiment show that.

Setting targets on thin data and judging during learning

In a campaign with fewer than 30 conversions a month, changing the target each week on weekly CPA means reacting to noise. Read the status and the last 30 days of conversions together in the bid strategy report. If conversions arrive with a delay, Google suggests leaving the most recent delay period out of the evaluation. Otherwise recent conversions have not arrived yet and CPA looks higher than it is.

Sending the wrong value to value-based bidding

Before you check the value, pin down what it is. Does the platform receive sales revenue, contribution before advertising or the expected value of a lead? How do tax and returns enter it, and which orders and which date basis does the comparison use?

Comparing platform conversion value with net sales for the same period is only a rough first check. If you send contribution, do not expect it to match net sales. If you send revenue, company-wide sales and Google Ads-attributed sales still differ in scope, so match the same orders on the same date basis. The GA4 and Google Ads mismatch guide covers why the scopes differ.

An intermediate action such as add to cart may sit in a goal used for bidding, and bidding then treats it as valuable too. Check both your primary actions and the contents of any custom goal. The ROAS calculator checks the arithmetic with your own figures, but it does not prove the attribution or the value is right.

Spend jumping after a switch to Maximize conversions

Google says Maximize conversions could increase spend significantly in a campaign that spends well under its budget. The strategy is designed to spend the full budget, so this is expected behaviour, not a fault.

Before switching, compare the last 30 days of average daily spend with the budget. If the gap is large, bring the budget closer to actual spend first. Also skip the "Search lost IS (budget)" column with this strategy. Google says the column is incompatible with Maximize conversions and points to the budget simulator instead.

In the end, you judge whether a strategy works by contribution per order. The conversion rate calculator helps you compare conversion rate across periods. If conversion rate falls while bids stay put, the cause may lie on the site rather than in bidding, for example in landing page speed. The steps in the conversion rate optimization guide come before any change to the target.