Choose Smart Bidding when the conversions and values it receives are the outcome you want to buy. For volume or value from a fixed budget, use Maximise conversions or Maximise conversion value. When a realistic average cost or return matters more than volume, use Target CPA or Target ROAS. Manual CPC and Maximise clicks still have uses, just not as a compulsory first step for new campaigns.
Which Google Ads bid strategy should you use?
Six strategies cover almost every Search campaign decision. Two of them are not Smart Bidding at all.
Strategy | What it seeks | Main decision |
|---|---|---|
Manual CPC | Clicks, with maximum bids you set | Is per click control worth the management effort |
Maximise clicks | As many clicks as the budget allows | Is traffic the real objective, and is a CPC limit appropriate |
Maximise conversions | Conversion volume from the budget | Are the selected outcomes meaningful and comparable |
Target CPA | Conversions around an average acquisition cost | Is the target grounded in mature results and affordable economics |
Maximise conversion value | The most reported value from the budget | Do the values reflect real differences between outcomes |
Target ROAS | Reported value around an average value to spend target | Are the values trustworthy, the campaign eligible and the target realistic |
The tree screens the objective. It does not prescribe a sequence.
Smart Bidding is four of the six: Target CPA, Target ROAS, Maximise conversions and Maximise conversion value. All four use Google AI to optimise for conversions or conversion value in every auction. That's how Google's Smart Bidding overview describes them.
Manual CPC is manual bidding. Maximise clicks is automated bidding outside that list. Smart Bidding is also not Smart campaigns, Performance Max or AI Max. Those are campaign types and a Search campaign feature. This guide stays with Search bidding.
What Smart Bidding does
At each eligible auction the strategy sets a bid. The bid comes from the predicted chance this click produces the selected goal's conversion or conversion value.
A few of the strategy's signals, like device and location, also exist as manual bid adjustments. Smart Bidding adds signals and combinations that manual bidding can't use. Its signals include:
device
location and location intent
time of day
remarketing lists
browser
operating system
the search query itself
The boundaries follow from that. The strategy optimises toward whatever the goal contains. That means a goal counting page views buys page views efficiently. It predicts from the data it's given. So it cannot tell a qualified lead from a time waster unless the account feeds that distinction back.
It bids within the campaign's targeting, budget and schedule. It can't override an exclusion, find extra budget or repair a broken tag.
Because of that, the first question in the tree is about measurement. The campaign around the strategy comes first too. Our guide to setting up Google Ads for a small business covers the build.
One naming change matters when you read older guides. From June 2026, Google shows "Maximise conversions with a target CPA" as Target CPA and "Maximise conversion value with a target ROAS" as Target ROAS. Google's note on how Smart Bidding strategies are organised calls this a purely visual label change. Old and new names may be mixed during the transition. The Target CPA section covers the August change to behaviour.
When Manual CPC or Maximise clicks fits
With Manual CPC bidding, you set the maximum price for a click. It can be an ad group default or set per keyword. Bid adjustments apply on top.
Maximise clicks automates those bids instead, chasing as many clicks as the budget allows. A cap on the maximum CPC is optional. Maximise clicks doesn't optimise toward impression share.
Both control the price of a click. Neither selects the clicks most likely to become qualified leads, because neither is told what a lead is.
That's why a low CPC proves nothing about acquisition cost. A cheap click that rarely converts can cost more per lead than a dear one that often does. Our guide to Google Ads metrics that matter works through the numbers.
Manual control fits three situations. A campaign whose real objective is traffic. A test with a per click ceiling the business chose for cash reasons. An account where one specific control matters more than efficiency.
Outside those, a new account can consider conversion based bidding from day one. It needs valid measurement and a sensible goal.
The reverse also holds. Unresolved tracking is a reason to repair measurement and to limit or pause spend meanwhile. It is not a reason to keep spending under manual bids until a conversion threshold arrives. Manual bidding does not make unmeasured spending safe.
That ceiling, $150 multiplied by 5%, is an economic estimate. It isn't an auction forecast or a recommended keyword bid. Nor does it claim clicks convert at that rate. Our guide to Google Ads cost per click and how to lower it has the full CPC calculation.
The affordable figure comes from job value and margin. Our guide to a good cost per lead for Google Ads works it out. What a small budget can test under any strategy is in our minimum budget for Google Ads guide.
Maximise conversions or Target CPA?
Maximise conversions
Maximise conversions sets bids to get the most conversions it can. It also tries to spend the full average daily budget. Conversion tracking is its only stated requirement. Google's Maximise conversions guide warns a campaign spending much less than its budget could see spend rise significantly.
A $100 average daily budget doesn't mean $100 spent each day. Under Google's spending limits, most campaigns can spend up to twice the average on a busy day. On quiet days they can spend less. No campaign is charged more than 30.4 days' worth of the average in a month. The promise is to use the budget, not to ration it.
The strategy counts whatever the campaign's selected goal counts, not whatever is marked primary somewhere in the account. Primary actions in that goal drive bidding. Google's guide to primary and secondary conversion actions adds one exception. Actions in a custom goal are used for bidding whatever their label. Setting the goal up properly belongs to our guide to Google Ads conversion tracking setup.
Never pad an enquiry goal with page views, button clicks or a duplicate form action. The volume would look healthier. But the strategy will dutifully buy them.
Some goals hold an action that isn't an enquiry or a sale. In our audits, about 1 in 4 accounts had a goal set up that way. Every audit figure in this guide comes from the same sample. That's 180 Google Ads account audits from August 2021 to September 2026.
Target CPA
Target CPA sets a desired average cost per conversion. Some conversions cost more than the target and some less. Google's Target CPA guide warns a target set too low can forgo clicks that would have converted.
Where history exists, Google suggests a starting target from the average CPA of the last 30 days. Without history, Google points to your business goals instead.
A target is an average objective. Three things get mistaken for it: a guaranteed lead price, a click cap and a daily spending ceiling. The ceiling is the budget's job.
A target can also be achievable and still unaffordable. Say mature CPA on the same conversion definition is $180 and the business can afford $120. Setting $120 doesn't repair the offer. It mostly reduces delivery.
Acknowledge the commercial problem and fix the page, the offer or the qualification before the bid. The destination is its own decision. Our guide to the Google Ads landing page vs homepage choice covers it.
Choose the initial target from mature outcomes for the same conversion definition, qualification stage and campaign scope. Put it beside the affordable figure. No rule says the first target must sit above current CPA. No rule says it must step down on a schedule either.
Recent periods are also incomplete by design. Because conversions arrive after their clicks, last week's CPA is inflated until its conversions land. When you compare periods with different targets, use each period's average target. Don't use the number set on the last day.
Set the target at the $120 the business can afford and expect $120 leads
Set the target from mature results and fix the page or the offer until $120 becomes achievable
The second approach is slower. It's also the one that holds up. Google says you can lower a CPA target to focus on efficiency. Set it too low and you may miss clicks that could have converted. And no target fixes the page or the offer behind a $180 lead.
The August change is the one to review now. Since 17 August 2026, budget limited campaigns on target strategies track their set target more consistently. That's according to Google's note on changes to target based bid strategies. Campaigns that had been beating their targets may move back toward them. Google won't adjust targets or budgets for you.
The options are:
keep the target
adjust it in line with recent performance
set a custom number
change strategy
Google's example is a $10 target achieving $5. Left alone, it would start delivering closer to $10.
A stale target is now a live instruction. Our audits reviewed target strategy campaigns from 17 August to 30 September 2026. In about 1 in 2 of them, the target was more than six months old. Raising a budget or loosening a target remains a business decision. No promised profit or preserved CPA comes with it.
We'll check the goal it bids on, how old your targets are and what your enquiries really cost. Then we'll tell you what we'd change first.
Maximise conversion value or Target ROAS?
Maximise conversion value
Maximise conversion value chases the most conversion value the budget can buy. Google's Maximise conversion value guide says it needs transaction specific values. Without a target, it tries to spend the full average daily budget. Add one and it behaves like Target ROAS.
It also leaves the meaning of value to you, such as sales revenue or profit margin.
That last point is the whole discipline. The strategy maximises a number. Whether the number is revenue, margin, a lead score or expected contribution is your decision. The report should say which.
Target ROAS
Target ROAS adds an average return on ad spend to aim at. Google's Target ROAS guide uses $5 of sales for each $1 of ad spend, a 500% target. It warns that a target set too high may limit traffic.
Google suggests a target from actual ROAS over the last few weeks. That's the Conv. value/cost column multiplied by 100. Search and Shopping need 15 conversions in the past 30 days at the conversion tracking level. Other campaign types have their own rules.
A 400% target means $4 of reported value per $1 of media. It means $4 of revenue only if the value is revenue. A ratio built on lead scores or expected contribution should be labelled as that. It isn't realised revenue ROAS.
For an online store, check these before trusting either strategy:
the transaction identifier
the currency
duplicated orders
refunds
the value definition
A strategy fed duplicate orders chases duplicates.
Revenue bidding also has a blind spot. It can favour high revenue orders with thin margins over smaller orders with fat ones. That's because it never sees the margin. Nothing about revenue ROAS optimises profit on its own.
Model your break even ROAS from margin and fees in our ROAS calculator. The full profit calculation sits in our guide to Google Ads ROI.
For lead generation, the question is which stage to bid on. Google's guide to value based bidding for Search and Shopping recommends:
A single stage in the lead to sale journey for bid optimisation.
An earlier goal, like a qualified lead, when the delay to a sale is too long to manage.
At least two different values.
Conversions sent as soon as they are available, with daily offline uploads described as optimal.
No backfilled historical data.
In practice, pick a stage that is timely and meaningful. Define it once and feed it back consistently. Never count a raw lead, its qualification and its sale as three acquisitions in one goal. That triples a single customer. Values can reflect observed differences in expected business value, each with a documented basis and a review date.
Equal lead counts, unequal value. A count strategy treats those 200 leads as identical. Tell a value strategy a Service B lead is worth $450 and a Service A lead $200. It will lean toward B. That's only an improvement if close rates and contributions are measured, not guessed, and reviewed when they drift.
An imported outcome improves the signal only if the stage definition holds and the uploads keep arriving. A stage label on its own is not evidence that a lead is valuable. The tags, imports and consent handling behind it are conversion tracking work, not bidding.
How much conversion data do you need?
There are two separate questions here. The first is whether the strategy is formally available. The second is whether the results can be read with any confidence.
The formal rules are narrow. Target ROAS on Search and Shopping needs 15 conversions in the past 30 days. Target CPA and Maximise conversions need conversion tracking and nothing more.
Target CPA can start with no conversion history at all. Its 30 conversions over 30 days is Google's guidance for judging results, not an entry requirement. Our rule of thumb is 15 to 30 conversions a month before we trust a target. It isn't a Google requirement: Google sets no minimum for Target CPA.
In the same audits, about 2 in 5 accounts had a campaign below that rule of thumb. It ran conversion based Smart Bidding on under 15 primary conversions. The count covers the 30 days before each audit. Smart Bidding at that volume isn't automatically a mistake. But it changes how long the results take to read.
Volume also sets the clock. Google's learning period guidance says performance can take up to one to two conversion cycles to stabilise. It can be faster with more conversion data. For Target ROAS, Google's Smart Bidding overview suggests judging results over at least 50 conversions. Treat 50 as an upper benchmark, not an entry requirement. The chart turns it into weeks at four monthly volumes.
At 15 a month, 50 conversions take more than three months to arrive. At 30 a month, they take about seven weeks. Neither is a reason to avoid Smart Bidding. Both are a reason to judge it over months rather than days.
Confidence depends on more than a count. Check that the goal holds valid actions and nothing else. Check that values mean one thing, where values are used. Check how long the conversion delay runs and whether offline uploads arrive on a schedule.
Then check the context. Did the market or the offer change in the period? Can the budget buy a readable sample? Is the account fragmented into campaigns that each see a handful of outcomes?
A campaign with eight real enquiries a month is not ineligible. It needs a longer evaluation horizon and a careful plan. Pool campaigns into one strategy only when their objectives and economics belong together. Never widen the goal with irrelevant actions to reach a number. That trades a small true signal for a large false one.
Before you switch, value strategies also need reported values through Google's preparation window for that strategy. Collect representative values over full cycles. Keep the conversion definition fixed while you test the bidding decision. Changing both at once makes the result unreadable.
How long does Smart Bidding take to learn?
Learning is a short status, not a verdict on the strategy. Google shows it after a change, when there may be minor performance fluctuations. Its guide to bid strategy statuses lists the reasons: a strategy created or reactivated, or a setting changed. Adding or removing campaigns, ad groups or keywords also counts.
Limited means the strategy is constrained by search volume, bid limits, budget or the strategy itself. Misconfigured flags a setup problem, such as the conversion settings or a budget shared with another strategy.
Google's learning period page says the label never lasts more than seven days. It also says budget, target and targeting changes never trigger it, which narrows the status guide's list. Performance can take up to one to two conversion cycles to stabilise. Conversion data from previous campaigns can help it stabilise sooner. So a cleared label doesn't mean the results are readable.
Target changes run on the same clock. The bidder starts optimising to a new target within minutes. But conversions arrive late. So Google's target adjustments with Search Smart Bidding guide recommends waiting one to two conversion cycles before judging.
It also warns a major target change can have a similarly sized impact on spend or volume. It advises against several ROAS target changes within one conversion cycle, unless there's a true business need.
Each change hands the bidder another version of the desired outcome. Google gives no safe percentage for an edit. Make the change the business needs, once, then leave it for the cycles it takes to read.
The interval between those moments follows the conversion cycle and the business risk, not a day count.
Take the 45 service accounts we managed from May 2024 to September 2026. Across those accounts, the Learning status after a strategy change usually cleared within 1 to 3 weeks. That's when Google stops flagging fluctuations, not when the results are worth reading. Our figure comes from before Google's page stated a seven day limit.
Google says the bidding keeps learning after the label goes. A cleared label doesn't prove the results are commercially acceptable. Waiting silently through broken tracking isn't patience either.
How to change bid strategy and assess the result
Before the change, fill in a change record like the one below. It pins down the baseline, the goal, the targets and the authorised spend before anyone sees a result.
Then change the campaign setting, or the portfolio setting if the campaign belongs to one. Change nothing else that day.
Where volume allows, Google's experiment tools can split a Search campaign. The two strategies then run side by side for a defined period. That's the cleanest reading available.
A small account cannot produce a decisive split test from a handful of conversions a month. Pretending otherwise produces a confident wrong answer. The method there is a sequential comparison over full cycles, with the confounders written down.
Scope and change date: [campaign or portfolio, and the date]
Previous and new strategy: [old strategy and target, new strategy and target]
Selected goal and conversion definition: [the goal, its actions and the qualification stage]
Mature baseline: [period, spend, conversions, actual CPA or ROAS and average target]
Authorised spend and business limit: [budget, affordable CPA or break even ROAS]
Other changes in the window: [offer, tracking, season, targeting, pages]
Review date and stop rule: [when it is read, and what ends the test early]
Define three moments in advance. The review, after the conversion lag has passed and at least one full cycle has completed. The immediate repair, the moment a tag or an import stops. And the stop, the loss limit the owner set before launch, not a feeling in week two.
At the review, compare the same goal and campaign scope, allowing for lag. If the offer, tracking, season or targeting changed in the window, report those beside the bidding change. Don't credit or blame the strategy for everything.
The bid strategy report is where that comparison happens. For Target CPA, its scorecard sets the average target CPA beside the actual CPA and the conversions. A conversion delay section shows how long most people take to convert.
Google's guide to the bid strategy report says to wait until all conversions are reported before evaluating. If the gap between target and actual isn't statistically significant, it says to select a longer date range.
Our first 30 days of Google Ads guide covers a new account's wider review rhythm.
Advanced controls and tracking interruptions
Each of these controls does one bounded job. Each is misused when it is asked to do another.
Control | Use it for | Not for |
|---|---|---|
Portfolio and shared budget | One strategy across campaigns that share a goal, with budget allocation as a separate linked control | Bundling unrelated services to inflate conversion counts |
Device and schedule | Device exclusions and target adjustments where the strategy supports them, schedules for eligibility | Ordinary manual modifiers, which Smart Bidding does not use |
Seasonality adjustment | A major expected short term conversion rate change such as a sale | Routine seasonal demand or a blanket holiday bid boost |
Data exclusion | Erroneous conversion data from a tag or import incident | Erasing a period of real poor results |
Value rules | Evidence based value differences the measured value does not already carry | Counting the same margin or customer difference twice |
Smart Bidding Exploration | Eligible Search Target ROAS campaigns that accept a flexible return for more traffic | Expanding keyword matching |
A portfolio bid strategy runs one strategy across several campaigns. It's one way for a low volume account to pool outcomes that belong together. Google calls linking shared budgets to portfolio bid strategies best practice for campaigns that share the same goals. The linked budget stays with the portfolio. It updates automatically when a campaign is added or removed.
Smart Bidding ignores ordinary manual bid adjustments. But ad schedules still decide when ads can show. Google's guide to bid adjustments lists the exceptions. A device adjustment on Target CPA, for one, changes the target rather than the bid. The rest of the campaign controls are in our guide to Google Ads settings that waste budget.
A seasonality adjustment warns the bidding about a major, short conversion rate change, such as a sale. Google's guide to creating a seasonality adjustment says it suits events of one to seven days. It's for major changes only, because Smart Bidding already manages seasonal events.
Conversion value rules adjust values by audience, location or device. Google's conversion value rules guide says Target ROAS and Maximise conversion value use adjusted values in real time. So a margin difference already built into an uploaded value mustn't be applied again as a rule.
A data exclusion tells Smart Bidding to ignore broken conversion data. The cause might be a tagging problem, a website outage or a failed import. Google's guide to data exclusions says it excludes the period's clicks and their conversions without changing reporting. The range should cover at least 90% of the affected clicks, allowing for conversion delay. Google discourages frequent or prolonged use.
Reported leads collapse after a tag is removed, although the CRM still receives enquiries.
Repair the tag, verify outcomes, identify affected click dates and assess whether a data exclusion is appropriate.
The order matters. A tag fault shows up in Google Ads but not in the CRM. A real fall in demand shows up in both. Excluding it teaches the strategy nothing true.
Exploration is the newest option and the easiest to misdescribe. It's an opt in beta for Search campaigns on Target ROAS. It uses flexible ROAS targets to win more of the searches the campaign is already eligible for. Google's guide to Smart Bidding Exploration says it doesn't expand broad match reach. It measures success by traffic diversity.
A flexible target means some of that traffic may come in below your usual return. So judge Exploration as its own change, apart from targets, budgets and matching.
Four practical bidding scenarios
These are teaching scenarios with assumed figures, not client results. The differences between the businesses drive the decisions.
A new Melbourne plumber
No conversion history. A $1,500 monthly test budget the owner can afford to lose. Emergency work that arrives by phone.
What can be measured on day one decides it. Call tracking that records answered calls. A form that reaches an inbox someone reads. A weekly check that the recorded enquiries were real jobs.
With that in place, Maximise conversions on a goal of verified enquiries is a reasonable first strategy. Does the owner want per click control while the phone handling is proven? Then Manual CPC or a capped Maximise clicks test is reasonable too.
Neither path has to come first. A target can wait until the enquiries reach our 15 to 30 a month rule of thumb. The review date is set before launch. The stop rule is the budget.
Dental appointment enquiries
Assume $3,000 of media and 25 leads in a mature month, a $120 raw cost per lead. The first question is what a lead is. An enquiry, a booked appointment or an attended one are three different numbers with three different values.
Say bookings are recorded reliably and the practice can afford $150 a booking. Then Target CPA on a bookings goal is defensible, with the target set from the mature booking cost.
If only enquiries are measured, the target has to be an enquiry target. The attendance gap stays a business problem. And nothing walks the target down on a schedule. It moves when mature booking costs say it can.
An ecommerce store
Reliable purchase values, a known margin by category and enough orders for Target ROAS to be eligible. Compare two options. Budget led value bidding spends the budget for the most reported value. A return target is set from the recent Conv. value/cost figure. Then it's checked against the break even return for the margin.
A lower target does not always improve profit, because it can buy revenue below margin. A higher one can starve volume. The switch waits until values have been reported through Google's preparation window for that strategy. The review compares margin after returns, not revenue.
Low volume B2B
Five to eight enquiries a month, each potentially worth $5,000 to $15,000 of revenue. Deal size is not lead value, because most enquiries do not close.
The useful moves are a timely qualification stage and CRM feedback on a regular schedule. Consolidate campaigns so one strategy sees all the outcomes. Measure the evaluation horizon in cycles rather than weeks. Controlled bidding is an option if the owner wants predictability while that data accumulates.
Nothing about five to eight leads makes the account ineligible for Smart Bidding. Nothing about it makes a four week verdict meaningful either. At eight a month, 50 conversions are about six months away.
What to monitor after switching
Monitor the outcome the business cares about, beside the figures the strategy was given. Allow for the lag. A falling CPC or a higher conversion count isn't the success criterion on its own.
The same audits include accounts we take over. Among those, about 3 in 5 were still on their main campaign's original bid strategy. The history showed no change and no recorded reason. That's the strongest argument for the review record below.
Outcomes
Qualified leads or customers from the business's records for the period, not the platform count alone
Lag
The data through date, the typical conversion delay and whether the period is mature enough to read
Target
Actual CPA or ROAS against the average target for the period, with any target changes dated
Spend
Pacing against the authorised budget, days the budget limited delivery and any Limited status reason
Signal
The goal's actions, value meaning, upload regularity and any tracking incident in the window
Handling
Calls answered, forms responded to and enquiries qualified, because the strategy cannot fix a missed call
Review period and conversion lag: [dates and delay]
Selected goal and value meaning: [definition]
Spend and mature conversions: [amount and count]
Qualified leads or customers: [count and source]
Actual CPA or ROAS and average target: [values]
Other changes or data incidents: [details]
Business limit and next action: [keep, change or stop]
Next review and owner: [date and person]
If an agency runs the account, this review belongs in its monthly report. That's what our Google Ads agency page shows.
What we recommend at Elev8d
Fix the goal before you pick the strategy. We strip the campaign's goal back to verified enquiries or sales. Wherever the phone brings in work, we make sure calls are counted. Everything after that depends on it.
Then we start new Search campaigns on Maximise conversions. We set a daily budget the owner is happy to see spent in full. That's because the strategy will try to spend it. We don't make anyone serve time on Manual CPC first.
A target comes later, once the campaign reaches our 15 to 30 a month rule of thumb. We add Target CPA or Target ROAS when the mature cost or return is known. We set it from those results and check it against what the business can afford. The change record gets filled in before anything moves. Then we leave the target alone for full conversion cycles.
We'd do something else in three cases:
Tracking is broken. We limit or pause spend and fix it before choosing any strategy.
Volume sits well under 15 a month. We pool campaigns that share a goal into one portfolio. Our guide to Google Ads account structure covers the approach. Then we judge the results over months.
The owner needs a hard ceiling on what a click costs, for cash reasons. Manual CPC or a capped Maximise clicks test fits that better.
FAQ: remaining bidding questions
Can I cap the cost per click on Target CPA or Target ROAS?
Only on a portfolio strategy. Google allows maximum and minimum bid limits on portfolio Target CPA and Target ROAS strategies, not standard ones. It applies them in Search Network auctions only. It also recommends against them, because a limit can stop the bidding from reaching the target you set.
Does a hard click ceiling matter more than the average cost? Then Manual CPC or a capped Maximise clicks campaign is the honest choice.
Does Smart Bidding count phone calls?
Only the calls you measure and include in the campaign's goal. Google's calls from ads conversion counts a call that lasts longer than a minimum duration you set. It needs call reporting switched on.
Calls to the number on your website need their own setup. Our guide to call tracking for Google Ads covers it. A call the account can't see is a lead the bidding never learns from.
Should I switch to broad match when I move to Smart Bidding?
Not as part of the same change. Matching decides which searches a campaign is eligible for. Bidding decides what it pays for them. Our guide to Google Ads match types explains this from the matching side.
Judge a matching change on the search terms report and a bidding change on mature outcomes. Avoid changing both in the same window.
Next steps
Check the goal before the strategy
Open the campaign's selected conversion goal and list every action inside it. Remove anything that isn't the outcome you want to buy.
Write the change record
Fill in the baseline, authorised spend, business limit, review date and stop rule before the strategy changes.
Review stale targets now
Compare each Target CPA or Target ROAS campaign's target with mature recent results and the affordable figure. Since 17 August 2026 the target is what the campaign delivers toward.
