Performance Max can spend the full budget, generate orders, and report a healthy ROAS while the business makes very little money from those sales.
That does not necessarily mean the campaign is broken.
It may be doing exactly what it was told to do.
Performance Max uses the conversion goals and values available in Google Ads to decide which opportunities are worth bidding on. If the account sends purchase revenue but the business judges success by contribution profit, Google and the finance team are working with two different definitions of value.
That is where many Performance Max problems begin.
Performance Max optimises the value you give it
Performance Max is a goal-based campaign type that can serve across Google's advertising inventory, including Search, Shopping where eligible, YouTube, Display, Discover, Gmail, and Maps.
The campaign uses automated bidding and Google's prediction systems to decide which auctions may produce the results defined in the account.
That last part matters most.
If the account is optimising for purchases, Google looks for purchase opportunities.
If bidding uses conversion value, it looks at the values attached to those conversions.
If those values represent revenue, the system is primarily being taught which orders generate more reported revenue.
If the values have been adjusted to better represent margin or another business outcome, bidding can use that information instead.
The campaign cannot infer your commercial model from a ROAS target alone.
Revenue, ROAS, and profit answer different questions
Suppose Google Ads reports:
€1,000 ad spend → €4,000 conversion value → 400% ROAS
That looks efficient.
But the €4,000 may represent sales revenue before subtracting:
- Cost of goods
- Discounts
- Payment fees
- Shipping subsidies
- Fulfilment costs
- Returns
- Other variable costs attached to the order
If only a small part of that €4,000 remains after those costs, the reported 400% ROAS tells a much less impressive business story.
This becomes even more important when product margins vary.
Imagine two products each generate €200 in revenue.
Product A leaves €90 before advertising costs.
Product B leaves €25.
If Google receives only the €200 transaction value for both orders, those two sales look equally valuable from a revenue perspective.
The business knows they are not.
Google Ads can work with better value data
This is where the original “Performance Max does not optimise for profit” argument needs an important qualification.
Google Ads can use value-based bidding with values that better represent the business outcome you care about. Google also supports conversion value rules, and Shopping-related setups can provide cost-of-goods data for gross-profit reporting.
The practical problem is that many eCommerce accounts still send the transaction total and then use ROAS as a substitute for profitability.
That is not a Performance Max calculation error. It is a measurement decision.
Know your break-even point before choosing a ROAS target
A target ROAS only becomes commercially useful when you understand what return the business actually requires.
A simplified break-even calculation can start with contribution margin.
For example, if €100 of revenue leaves €30 available before advertising costs, spending more than €30 to generate that €100 would push that order below break-even under that simplified model.
The exact calculation depends on what the business includes in contribution margin.
That is why there is no universal “good ROAS.”
A 500% ROAS can be poor for a low-margin business. A 250% ROAS might be attractive for a product with stronger economics.
Before setting a target, know what that target is supposed to protect:
- Gross margin
- Contribution margin
- Customer acquisition economics
- Cash-flow requirements
- Another clearly defined business target
Otherwise, the target is just a number that looks efficient inside Google Ads.
Target ROAS cannot repair the wrong conversion value
Target ROAS tells Google to optimise conversion value while aiming for an average return relative to advertising cost.
It does not check your product costs before deciding whether that return is profitable.
If the values entering Google Ads represent revenue, the target is effectively asking for a particular relationship between ad spend and reported revenue.
Increasing the target does not magically convert that revenue metric into profit.
It can also change volume.
A more restrictive ROAS target can reduce the number of auctions the campaign is able to enter profitably according to Google's predictions. A looser target can give the system more room to spend and pursue additional conversion value.
Neither direction is automatically correct.
The target needs to make sense relative to:
- Current conversion data
- Product economics
- Budget
- Volume requirements
- Market conditions
Frequent target changes after a few weak days also make performance harder to evaluate. Conversion delays and bidding adjustments need time to show up in the data.
Before changing bids, verify the purchase data
Automated bidding depends on conversion data.
If the data is wrong, Google can optimise very efficiently towards a result that did not actually happen.
Check for problems such as:
- A purchase firing twice
- Several conversion actions recording the same order
- Cancelled or failed orders remaining in the reported value
- Incorrect currencies
- Incorrect transaction values
- Test purchases entering production reporting
- Micro-conversions being treated as primary optimisation goals
A duplicated €200 order is not a small reporting detail.
If Google receives €400 of value from one real €200 purchase, the campaign sees stronger performance than the business actually achieved.
That can affect future bidding decisions.
Google Ads, GA4, and the store do not need to match perfectly
Different systems use different attribution logic, reporting windows, identity methods, and event processing.
Small differences are therefore expected.
What deserves investigation is a large, unexplained, or sudden difference.
For eCommerce, we prefer to reconcile campaign reporting against the system that actually records the order.
The question is not:
“Does GA4 show exactly the same revenue as Google Ads?”
It is:
“Does the advertising data represent the commercial reality closely enough to make bidding decisions?”
Do not optimise all customer actions as though they were purchases
An add to cart is useful information.
It is not the same business outcome as a completed purchase.
The same applies to beginning checkout, viewing a product, signing up for a newsletter, or performing another micro-conversion.
If several actions with very different business value are included in campaign optimisation without an intentional value model, the bidding system receives a mixed objective.
For a normal eCommerce purchase journey, the completed purchase will often be the main optimisation event.
Other actions can still be measured for funnel analysis without automatically giving them the same role in bidding.
There are exceptions. A store with very low purchase volume or a more unusual conversion journey may need a different approach.
The important part is that the conversion hierarchy reflects the actual business hierarchy.
Brand demand can make Performance Max look more incremental than it is
Someone searching directly for your company name is different from someone discovering the company for the first time.
Both can produce valid sales.
But they answer different marketing questions.
If Performance Max captures a meaningful amount of existing branded demand, reported ROAS can be strong partly because those customers were already close to purchasing.
That does not mean branded conversions should be discarded.
It means they should not automatically be interpreted as demand created by Performance Max.
Google provides brand exclusions for advertisers who want to restrict Performance Max from serving on specific branded queries. This can be useful when brand demand is managed separately and the business wants a clearer view of non-brand performance.
Brand exclusions are not a universal requirement.
The right setup depends on how the account is structured and what the business is trying to measure.
The important question is:
How much of the result came from customers already looking for us, and how much came from demand we needed advertising to capture or create?
Returning customers can hide weak acquisition economics
The same issue appears with existing customers.
A returning customer may convert more easily because they already know the product, brand, checkout, delivery experience, and returns policy.
That revenue is valuable.
But if the business is judging Performance Max primarily as a customer-acquisition channel, total ROAS can hide what it costs to acquire someone new.
Google Ads currently provides customer lifecycle goals that can help advertisers value, prioritise, or in some setups focus specifically on new customers.
These controls still depend on customer data and Google's ability to identify customer status.
We therefore prefer to compare advertising reports with backend, CRM, or eCommerce customer data where possible.
A useful report separates questions such as:
- How much revenue came from new customers?
- How much came from returning customers?
- What did a new customer cost to acquire?
- What was the first-order contribution from those customers?
That is much more informative than one blended ROAS number.
Product revenue should not be confused with product value
Performance Max may find it easier to sell some products than others.
That can be perfectly rational from the bidding system's point of view.
But the easiest product to sell is not necessarily the product the business most wants to advertise.
Products can differ in:
- Margin
- Return rate
- Stock position
- Average order value
- Seasonality
- Repeat-purchase behaviour
- Promotional priority
- Market availability
If those differences matter to the business, they need to appear somewhere in campaign structure, product data, conversion value, or reporting.
For feed-based eCommerce campaigns, product data is particularly important. Titles, identifiers, availability, pricing, images, and product grouping all affect what Google can understand and advertise.
The answer is not automatically “create more Performance Max campaigns.”
Extra separation makes sense only when there is a real business reason to manage groups differently, such as different margins, budgets, markets, targets, stock situations, or strategic priorities.
Do not ask one blended ROAS to answer every product question
Imagine a campaign generates:
| Product group | Revenue | Ad spend | ROAS | Commercial context |
|---|---|---|---|---|
| Group A | €10,000 | €2,500 | 400% | Low margin, frequent discounting |
| Group B | €6,000 | €2,000 | 300% | Higher margin, stronger contribution |
A campaign-level report makes Group A look more efficient.
The business may still prefer Group B once product costs are included.
This is why product-level or category-level economics matter.
Automation becomes more useful when the value supplied to it resembles the value the business actually cares about.
Creative determines what Performance Max has to work with
Performance Max is not simply a Shopping or Search campaign with extra automation.
Its assets can be used across several Google channels and formats.
If the campaign receives generic headlines, similar images, weak video, and one broad message for very different products, automation cannot invent a strong positioning strategy from those ingredients.
Asset groups should have a clear reason to exist.
That reason might be a category, product family, customer need, use case, or offer.
Useful assets can communicate:
- What the product does
- Who it is for
- Important product differences
- The offer
- Use cases
- Common objections
- Evidence or social proof where appropriate
More assets are not automatically better.
More useful variation gives Google more relevant material to work with.
Final URL expansion needs a website worth expanding into
Performance Max can use Final URL expansion to select another relevant page from the advertiser's domain rather than always sending traffic to the originally supplied URL.
Google currently enables this behaviour by default in Performance Max.
That can be useful on a well-structured eCommerce site because Google has more commercial pages available when matching intent.
It can also expose weak parts of the site.
For example, you may not want paid traffic landing on:
- Outdated category pages
- Low-stock or unavailable product ranges
- Blog content with no commercial purpose
- Old campaign pages
- Internal or support content
Google provides URL exclusions for this reason. Advertisers that need tighter landing-page control can also turn Final URL expansion off and use the eligible URLs supplied through campaign assets or page feeds.
The setting is not the strategy.
The real question is whether every page Google can choose supports the commercial intent behind the click.
Search themes and audience signals are guidance, not a profitability switch
Search themes can give Performance Max additional information about queries that may be relevant to customers.
They are not traditional exact targeting keywords.
Audience signals serve a similar strategic purpose. They provide information that can help Google understand useful audience patterns, but they do not permanently restrict the campaign to those audiences.
Good first-party data and well-chosen signals can give the system better context.
They cannot compensate for:
- Incorrect conversion tracking
- Revenue values that ignore the economics you care about
- A poor offer
- Weak landing pages
- Unclear product priorities
If the economic model is wrong, adding more audience signals simply gives the campaign more information while it pursues the wrong definition of success.
Use channel reporting to diagnose, not micromanage
Performance Max used to provide much less visibility into how results were distributed across Google's channels.
Google now provides a channel performance report showing campaign delivery and contribution across its channel inventory, with additional format-level data and diagnostics.
That is useful when a blended campaign result needs more context.
It can help answer questions such as:
- Which channels are actually receiving meaningful delivery?
- Where are conversions appearing?
- Are particular formats contributing differently?
- Are there diagnostics limiting a channel?
But Performance Max remains an automated cross-channel campaign.
The channel report is better used to understand behaviour and identify problems than to try to manually recreate separate Search, Shopping, Display, and YouTube campaigns inside Performance Max.
Performance Max should have a defined role in the account
Google describes Performance Max as complementary to keyword-based Search campaigns.
That distinction is useful.
Search gives advertisers direct control over keywords, search intent, ads, and landing pages.
Performance Max provides automated delivery across several Google properties.
An eCommerce account might therefore use separate campaign types when the business needs separate control over things such as:
- Brand demand
- Important non-brand searches
- Specific product groups
- New-customer acquisition
- Remarketing or demand generation
Not every account needs all of those.
The point is to know what job Performance Max is expected to perform.
“It can access most Google inventory” is not a campaign objective.
How we diagnose an unprofitable Performance Max campaign
When a Performance Max campaign spends but the commercial result is weak, we would not start by moving the target ROAS up and down.
We would work through the problem in this order.
1. Verify what the campaign is optimising for
Check the primary conversion goals and the values attached to them.
Are purchases the real business result? Are micro-conversions influencing bidding? Does conversion value represent revenue, margin, or something else?
2. Reconcile reported purchases with actual orders
Compare Google Ads with the store backend.
Look for duplicated transactions, missing values, incorrect currencies, cancelled orders, and unexplained differences.
3. Calculate the economics outside Google Ads
Know the approximate contribution available to pay for advertising and what ROAS or acquisition cost that implies.
Do this for meaningful product groups rather than assuming every SKU has the same economics.
4. Separate customer types
Understand how much performance comes from new customers, returning customers, and existing brand demand.
5. Look at what the campaign is selling
Compare spend and revenue with margin, stock, discounts, return behaviour, and strategic product priorities.
6. Review the post-click experience
Check the pages actually receiving traffic.
Make sure the offer, product, availability, shipping information, mobile experience, and checkout support the promise made before the click.
7. Only then adjust bidding and structure
Once the inputs are trustworthy, bidding decisions become much easier to interpret.
The change might involve the ROAS target, budget, product segmentation, customer-acquisition settings, brand exclusions, landing-page controls, creative, or the wider Google Ads account structure.
The data should tell you which lever deserves attention.
What a useful Performance Max report should contain
A monthly report that says:
Spend: €20,000
Revenue: €80,000
ROAS: 400%
does not tell us enough to judge profitability.
Where data is available, we would rather add context such as:
| Question | Why it matters |
|---|---|
| How much reported revenue matches backend orders? | Checks whether bidding data is reliable |
| Which products generated the value? | Shows whether low- or high-margin products drive the result |
| What share came from new customers? | Separates acquisition from retention where that distinction matters |
| How much came from branded demand? | Adds context to apparent campaign incrementality |
| What contribution did those orders generate? | Connects advertising revenue to business economics |
| Where did the traffic land? | Identifies weak commercial pages or URL-expansion issues |
| Which channels and formats contributed? | Provides context for the blended Performance Max result |
Not every business will have all of this data.
Use what is reliable rather than manufacturing precision.
Give the algorithm a commercial definition of success
Performance Max is very good at automating decisions at a scale no campaign manager could reproduce manually.
But automation still needs an objective.
If the account says that €200 of revenue is worth €200 regardless of product margin, customer type, returns, or business priority, Google has no reason to interpret that value differently.
If the business wants something more specific, that information needs to reach the advertising system through better conversion values, campaign structure, customer goals, product data, or the way performance is evaluated outside the platform.
That is the real role of human management.
Not deciding every auction manually.
Deciding what outcome is worth winning the auction for.
At Creative Brackets, we look at Performance Max together with conversion tracking, product economics, customer acquisition, creative, CRO, and the post-click experience.
Because spending the budget is not the difficult part.
The difficult part is making sure the value being scaled is valuable to the business.