Setting Target CPA and Target ROAS in Google Ads correctly is the difference between a bidding strategy that scales profitably and one that bleeds budget while Google's algorithm flails in learning phase limbo. Target CPA (cost per acquisition) tells Google the maximum you want to pay per conversion, while Target ROAS (return on ad spend) tells Google the minimum return you expect per dollar spent. Both are powerful smart bidding strategies, but only when they are set from real account data, not guesses, not industry benchmarks, and not whatever number "feels right."
By the end of this guide, you will know exactly how to set Target CPA and Target ROAS in Google Ads using a five-step framework grounded in your own conversion data. You will also learn when to use each strategy, how to scale targets without resetting learning, and which mistakes to avoid.
Prerequisites: You need a Google Ads account with active campaigns, conversion tracking properly configured (with conversions actually firing), and at least 30 days of historical data. If your conversion tracking is unreliable, fix that first. No bidding strategy can optimize toward broken signals. If you suspect signal quality issues, this guide on fixing conversion tracking and signal quality is worth reading before you proceed.
Before You Start: What You Need In Place
Before touching any bidding strategy, confirm three things. First, your conversion actions are set correctly in Google Ads. Only primary conversions should be included in the "Conversions" column that smart bidding optimizes toward. Secondary or observation-only conversions should not be feeding into the algorithm. Second, you have enough historical data to derive meaningful baselines. Campaigns with fewer than 15 conversions in the last 30 days will struggle with any target-based strategy. Third, your attribution model is set to data-driven attribution (or at minimum, position-based). Last-click attribution will distort your CPA and ROAS baselines and set you up with targets that do not reflect real performance.
Step 1: Establish Your Baseline With Maximize Conversions First
The single biggest mistake advertisers make when setting Target CPA or Target ROAS in Google Ads is skipping the baseline phase. Before you can tell Google what to aim for, you need to know what your account naturally delivers without constraints.
Start your campaigns on Maximize Conversions (for CPA-oriented goals) or Maximize Conversion Value (for ROAS-oriented goals) with no targets set. This lets Google's algorithm bid freely, finding the conversions or conversion value available at whatever cost the auction demands. The data you collect during this phase becomes your baseline.
How Long To Run Maximize Conversions Before Setting A Target
Run Maximize Conversions or Maximize Conversion Value for a minimum of two to four weeks. The goal is not a specific number of days but a specific volume of data. You need enough conversions for the average CPA or ROAS to stabilize, meaning day-over-day fluctuations settle into a recognizable range rather than swinging wildly.
What Conversion Volume You Need Before Switching
Aim for at least 30 conversions before switching to a target-based strategy. Google's own documentation suggests 15 conversions over 30 days as a minimum, but in practice, 30 or more conversions give you a far more reliable baseline. If your account converts fewer than 30 times per month at the campaign level, consider using portfolio bid strategies that pool conversion data across multiple campaigns.
How To Read Your Baseline CPA Or ROAS From The Data
Pull your campaign-level data for the baseline period. Look at the "Cost / conv." column for your baseline CPA or the "Conv. value / cost" column for your baseline ROAS. Use the full baseline period, not just the best or worst days. Write these numbers down. They are the foundation for everything that follows.
Step 2: Calculate A Realistic Starting Target
Your first Target CPA or Target ROAS should never be aspirational. It should be a slight improvement over what your account already delivers. The goal is to give the algorithm a constraint it can hit while still pushing performance in the right direction.
How To Derive Your First Target CPA From Historical Data
Take your baseline CPA from Step 1. Your starting Target CPA should be 10% to 20% above that baseline. If your baseline CPA is $50, set your initial Target CPA between $55 and $60. This gives the algorithm room to operate without immediately choking off volume.
Yes, you are setting a target that is slightly worse than your current average. That is intentional. The algorithm needs headroom. Setting a target exactly at your current CPA forces it to bid conservatively on every auction, which typically results in lost impression share and fewer conversions at roughly the same CPA, or worse.
How To Derive Your First Target ROAS From Historical Data
Take your baseline ROAS from Step 1. Your starting Target ROAS should be 10% to 20% below that baseline. If your baseline ROAS is 500%, set your initial Target ROAS between 400% and 450%. The logic is identical: give the algorithm space to find conversions across a wider range of auctions before you tighten the constraint.
The 10-20% Buffer Rule And Why It Matters
The buffer exists because smart bidding operates on averages, not guarantees. Google aims to hit your target on average over time, not on every individual conversion. If your target leaves no room for variance, the algorithm restricts bids so aggressively that volume collapses. You will see fewer impressions, fewer clicks, and often a CPA or ROAS that is actually worse than your target because the reduced volume makes each individual conversion more volatile.
Step 3: Choose The Right Strategy For Your Account
Google Ads smart bidding offers both Target CPA and Target ROAS, and choosing between them is not arbitrary. The right choice depends on your conversion data structure, your margins, and your account's conversion volume. Choosing the wrong strategy for your account profile is one of the fastest ways to waste budget.
When Target CPA Outperforms Target ROAS
Target CPA works best when your conversions have roughly equal value. Lead generation accounts where every lead is worth approximately the same amount are the classic use case. If you are a B2B SaaS company generating demo requests, or a home services business generating phone calls, and each conversion is worth a similar amount, Target CPA gives the algorithm a clean, simple signal to optimize toward.
Target CPA also tends to outperform in accounts with lower conversion volume because it requires less data granularity than Target ROAS. The algorithm only needs to learn cost-per-conversion patterns, not cost-per-dollar-of-value patterns.
When Target ROAS Is The Better Choice
Target ROAS is the right strategy when your conversions have meaningfully different values. Ecommerce is the most common example: a $30 order and a $300 order are both "conversions," but they are not equally valuable. If you use Target CPA in this scenario, the algorithm treats both equally, often gravitating toward the cheaper, lower-value conversions. Target ROAS solves this by optimizing for total value returned per dollar spent.
For Target ROAS to work, you must be passing accurate conversion values back to Google Ads. If your conversion values are wrong, static placeholders, or missing, Target ROAS will optimize toward garbage data. Fix your value tracking before switching to this strategy. For ecommerce brands already running Performance Max, this connects directly to how asset group structure and budget allocation affect ROAS outcomes.
When Neither Works And What To Use Instead
If your account has fewer than 15 conversions per month at the campaign level, neither Target CPA nor Target ROAS will perform reliably. The algorithm simply does not have enough signal to learn from. In these cases, stay on Maximize Conversions or Maximize Conversion Value without a target, or use Manual CPC with Enhanced CPC enabled. Build volume first, then add targets.
The Special Case Of Accounts Under 50 Conversions Per Month
Accounts generating 15 to 50 conversions per month sit in a gray zone. Target-based strategies can work, but they require wider buffers (20% or more) and longer monitoring periods. Portfolio bid strategies that pool data across campaigns are particularly effective here because they give the algorithm a larger dataset to learn from. If your account sits in this range and you are struggling with bidding strategy selection, this is exactly where the groas engine excels. It is trained on over $500 billion in profitable ad spend, which means it draws on patterns your individual account could never produce on its own, making smart bidding decisions more reliable even at moderate conversion volumes.
Step 4: Set The Target And Monitor The Right Metrics
Setting the target is the easy part. The hard part is knowing what to watch afterward and resisting the urge to change things too quickly.
In Google Ads, navigate to your campaign settings, open the bidding section, and enter your calculated target. If you are using portfolio bid strategies, set the target at the portfolio level. Confirm and save.
What To Watch In The First 14 Days After Switching
The first 14 days after setting a target are critical because Google's algorithm enters a "learning" phase. During this period, expect higher volatility in CPA or ROAS, fluctuating impression share, and day-to-day performance that looks worse than your baseline. This is normal.
Track these metrics daily but evaluate them weekly: conversions, cost per conversion, conversion value, ROAS, impression share, and the campaign status label (which will show "Learning" if the algorithm is still calibrating). Do not make changes based on a single day's data.
The Learning Phase Budget Buffer You Must Protect
During learning, do not reduce budget. If anything, ensure your daily budget is at least 5x to 10x your Target CPA (for Target CPA strategies) or sufficient to generate the volume the algorithm needs. Cutting budget during learning extends the learning phase or restarts it entirely, which wastes the data the algorithm has already collected.
When To Adjust The Target And By How Much
Wait a full two weeks after exiting the learning phase before making any adjustments. If performance is within 10% of your target after this period, the strategy is working. Let it run. If performance is consistently 20% or more off-target for two or more weeks post-learning, adjust the target by no more than 10% to 15% at a time. Larger adjustments reset learning.
Step 5: Scale The Target As Performance Improves
Once your campaign has stabilized on a target-based strategy, you can begin tightening. The goal is to gradually lower your Target CPA or raise your Target ROAS to improve profitability without collapsing volume.
The Safe Increment For Tightening CPA Or Raising ROAS Targets
Adjust in increments of 10% to 15%, maximum. If your Target CPA is $55 and performance is stable, lower it to $49 or $47, not to $35. If your Target ROAS is 400% and performance is consistent, raise it to 440% or 460%, not to 600%. Each adjustment triggers a new learning cycle, so smaller moves mean shorter disruptions and more predictable outcomes.
Wait at least two full weeks between adjustments. Track whether each change results in maintained volume at better efficiency, or reduced volume with no efficiency gain. If volume drops more than 15% to 20% after a target change, you have likely tightened too aggressively. Revert.
How To Scale Budget Alongside Target Changes Without Resetting Learning
Budget and target changes should not happen simultaneously. Change one variable at a time. If you want to tighten your Target CPA and increase budget, tighten the target first, wait for stabilization, and then increase budget. Changing both at once gives the algorithm two new constraints to learn, which extends the learning period and produces unreliable data.
When increasing budget, raise it by no more than 15% to 20% at a time. Google's algorithm can absorb gradual increases without resetting learning. Large jumps (doubling budget overnight, for example) almost always trigger a full learning reset.
Common Mistakes And How To Avoid Them
Setting ROAS Targets From Industry Benchmarks Instead Of Account Data
Industry benchmark articles love to say things like "the average ROAS for ecommerce is 400%." That number is meaningless for your account. Your margins, your AOV, your competitive landscape, and your conversion tracking setup are unique. Always derive targets from your own baseline data. Benchmarks are useful for directional context but dangerous as inputs to your bidding strategy. This applies equally to auditing your account's overall performance: your numbers are the only numbers that matter.
Changing Targets Too Frequently
Every target change triggers learning. If you are adjusting weekly, your campaigns never exit learning, and the algorithm never has stable ground to optimize from. Set a cadence: review every two weeks, adjust no more than once per month unless something is clearly broken.
Ignoring Seasonality When Setting Targets
A Target CPA set during your peak season will be too aggressive during your slow season. A Target ROAS set during a promotional period will be unachievable at full price. Account for known seasonal patterns by loosening targets ahead of down periods and tightening them as demand increases. If you do not proactively adjust for seasonality, the algorithm will do it for you by cutting volume when it cannot hit an unrealistic target.
Applying One Target Across All Campaigns
Different campaigns serve different intent levels, different products, and different margins. A branded search campaign and a broad-match prospecting campaign should not share the same Target CPA. Use portfolio bid strategies to group campaigns with similar performance profiles, and set different targets for each portfolio.
Stacking Target Changes With Other Major Edits
Do not change your target on the same day you restructure ad groups, add new keywords, or launch new landing pages. Each of those changes introduces its own variables. Isolate your changes so you can attribute results accurately. When you change everything at once, you learn nothing about what worked.
How groas Handles All Of This For You
Every step in this framework requires judgment calls: when your baseline is stable enough, how much buffer to add, when to tighten, how to handle seasonality, whether to use Target CPA or Target ROAS. These are not one-time decisions. They are ongoing, campaign-by-campaign, week-by-week calibration exercises that compound in complexity as your account grows.
This is precisely what groas is built for. The proprietary engine trained on over $500 billion in profitable ad spend handles bidding strategy selection, target calculation, learning phase management, and scaling decisions around the clock. It draws on patterns from a dataset orders of magnitude larger than any single account could produce, which means it identifies optimal targets faster and scales them more reliably than manual management.
For in-house teams running their own Google Ads (DWY), the engine handles the heavy lifting while a senior strategist works alongside your team, providing a strategy call every other week and weekly reports on exactly what was done. Your team stays in control, but the execution runs on a different level. For businesses that want Google Ads fully handled (DFY), a dedicated strategist owns every decision end to end, from bidding strategy to landing pages to offer structure. Nothing to log into or manage. Agencies running client accounts (DIY) get direct access to the groas engine and run it themselves across unlimited client accounts, keeping their brand and margin while the execution underneath becomes dramatically more powerful.
No onboarding fees. Month-to-month commitment, cancel anytime. groas earns the next month by performing.
If you are an agency looking to scale client performance across accounts, start your 7-day free trial. If you have an in-house team and want the engine plus a strategist, get started today. If you want Google Ads fully managed, apply for DFY access.
The Bottom Line
Setting Target CPA and Target ROAS in Google Ads is not a set-it-and-forget-it exercise. It is a structured process: establish a baseline, calculate a realistic starting target with a buffer, choose the right strategy for your account, monitor through learning, and scale incrementally. Skip any step and you risk either strangling volume with aggressive targets or burning budget with conservative ones.
The framework above gives you a repeatable system for getting it right. But the reality is that bidding strategy management is one of dozens of optimization levers that need constant attention in a Google Ads account. If you would rather spend your time on the parts of the business only you can run, groas puts every one of these decisions into the hands of an engine and a strategist that do this all day, every day, across thousands of accounts. That is the difference between following a framework and having one built into everything your campaigns do.
Frequently Asked Questions
What Is The Difference Between Target CPA And Target ROAS In Google Ads?
Target CPA tells Google the maximum amount you want to pay per conversion. Target ROAS tells Google the minimum return you expect per dollar of ad spend. Target CPA treats all conversions equally, making it ideal for lead generation where each conversion has similar value. Target ROAS accounts for varying conversion values, making it the right choice for ecommerce or any account where different conversions produce different revenue. Both are smart bidding strategies that use machine learning to set auction-time bids, but they optimize toward fundamentally different goals.
How Many Conversions Do I Need Before Setting Target CPA Or Target ROAS?
You need a minimum of 30 conversions in your baseline period to set a reliable Target CPA or Target ROAS. Google's documentation suggests 15 conversions over 30 days as the floor, but 30 or more conversions produce a significantly more stable baseline. If your campaigns generate fewer than 15 conversions per month, stay on Maximize Conversions or Maximize Conversion Value without a target until volume builds. Portfolio bid strategies that pool data across campaigns can help accounts in the 15 to 50 conversion range.
Should I Set Target ROAS Based On Industry Benchmarks?
No. Industry benchmarks reflect averages across thousands of accounts with different margins, products, tracking setups, and competitive landscapes. Your Target ROAS should always be derived from your own baseline data collected during a Maximize Conversion Value phase. Use your actual ROAS over two to four weeks as the starting point, then apply a 10% to 20% buffer below that number. Benchmarks can provide directional context but should never be used as direct inputs to your bidding strategy.
How Often Should I Change My Target CPA Or Target ROAS?
Adjust no more than once per month under normal conditions. Every target change triggers a new learning phase, and campaigns that never fully exit learning produce inconsistent results. Wait at least two full weeks after exiting learning before evaluating whether an adjustment is needed. When you do adjust, change by no more than 10% to 15% at a time. If performance is within 10% of your target, the strategy is working and does not need changes.
What Happens During The Google Ads Learning Phase After Setting A Target?
During learning, Google's algorithm experiments with different bid levels to calibrate toward your target. Expect higher volatility in CPA or ROAS, fluctuating impression share, and day-to-day swings. This typically lasts one to two weeks. Do not reduce budget, change the target, or restructure campaigns during this period. Interrupting learning wastes the data already collected and often extends the phase. Protect your budget and let the algorithm stabilize.
Can I Use Target CPA And Target ROAS On The Same Account?
Yes. Different campaigns within the same account can use different bidding strategies. A branded search campaign might use Target CPA because every brand conversion is worth roughly the same, while a shopping campaign might use Target ROAS because product values vary widely. Use portfolio bid strategies to group campaigns with similar performance profiles and set appropriate targets for each group.
What Is The Best Way To Handle Seasonality With Target CPA And Target ROAS?
Proactively loosen targets ahead of known slow periods and tighten them as demand increases. A Target CPA set during peak season will be unachievable in a slow month, causing the algorithm to choke off volume. Review your year-over-year data to identify seasonal patterns and build a calendar for planned target adjustments. If managing seasonality across multiple campaigns sounds complex, groas handles this automatically. The engine draws on patterns from over $500 billion in ad spend data, adjusting strategy in real time as demand shifts.
Is It Better To Use Maximize Conversions Or Target CPA In Google Ads?
Maximize Conversions without a target is the right starting point for new campaigns or campaigns without enough conversion data. Target CPA is the next step once you have 30 or more conversions and a stable baseline. Think of Maximize Conversions as the data collection phase and Target CPA as the optimization phase. Skipping straight to Target CPA without baseline data means you are guessing at the target, which usually leads to either strangled volume or wasted spend.
How Does groas Handle Bidding Strategy Decisions Automatically?
groas uses a proprietary engine trained on over $500 billion in profitable ad spend to handle bidding strategy selection, target calculation, learning phase management, and scaling decisions continuously. Instead of relying on one account's limited data, the engine identifies optimal targets using patterns from thousands of accounts. Depending on the product, a senior strategist either owns every decision end to end (DFY), works alongside your team (DWY), or the agency runs the engine directly (DIY). This eliminates the manual calibration cycle described in this guide and produces more reliable results faster.
.png)
