Performance Max can look like a shortcut: you connect the products, add materials and let automation find opportunities. But the campaign doesn't inherently know the cost of returns, the stock priorities or why one product is more valuable to the business than another. This information must be turned into coherent goals, data and decisions.
Prepare Performance Max starting from correctly measured purchases, eligible products, relevant materials and a budget tied to contribution. Check the destinations and the controls available in the account, then track the commercial result, not just attributed revenue. Intervene immediately on obvious errors and evaluate performance changes on sufficiently mature data. Automation doesn't remove the need for diagnosis.
Contents
- What the campaign does
- Goals and measurement
- Products and materials
- Destinations and signals
- Budget and examples
- Diagnosis
- Launch plan
- FAQ and checklist
What Performance Max does and where the decision stays yours
Performance Max is a goal-based campaign type, with access to multiple Google channels. It shouldn't be read as a synonym for Shopping. The official Google overview describes the role of automation and the combination of goals, materials and data.
For a store, the useful question is which product selection and which commercial result you want to support. If all products enter the same plan without economic differentiation, the campaign can optimise towards the values it receives, while you track something else in the business.
Start with a definition of success
Write down what a relevant order means, what revenue you use and what contribution you want to keep. Separate acquiring new customers from the return of existing ones, to the extent that you have credible data. Don't judge market expansion only by an average that includes a lot of demand for the store's name.
A campaign can produce revenue without reaching your financial goal. That is why reporting must include the record of retained orders and the products bought. Automating distribution can't replace a coherent definition of value for the business.
Goals and measurement are the first check
Confirm which actions influence optimisation
Check the campaign goals and the actions included. An add-to-cart can be useful for diagnosis, but it doesn't represent the same commercial value as a purchase. Google explains the difference between primary and secondary conversions, including the particularities of custom goals.
Ask for a clear list: the event's name, source, value and role. If two integrations send the purchase, check whether it is counted twice. Don't disable a source at random without understanding the dependencies; document the setup before intervening.
Follow a test order
Choose a product, a variant and a payment method, then check the order ID and value in the relevant systems. Repeat for paths that differ significantly, such as external payment or an order with a discount. A single successful path doesn't confirm all possible combinations.
Keep test orders separate from commercial analysis. Also check what happens when the confirmation page reloads. An event that repeats can change the reporting even if the store created only one real order.
Define periodic reconciliation
The Google report and the store's records don't have to be identical on any given day. Attribution, consent and delay can produce differences. What matters is to explain their direction and size, not to claim a perfect match through arbitrary manual adjustments.
For a mature evaluation, track retained orders and relevant revenue. If the period is recent, mark it as provisional. A conclusion about profitability can be premature before cancellations and returns are clarified.
Prepare products and materials as a whole
Commercial selection comes before technical grouping
Choose products with stock, correct data and a known contribution. Don't assume that all items in the catalogue deserve the same investment. A product can attract many orders but have costly shipping or returns that reduce the final result.
Create groupings you can explain through the customer's need and the store's economics. Don't over-fragment just to get a structure that looks sophisticated. Each separation must help a decision about products, materials or budget.
Check eligibility and the consistency of the data
The data in Merchant Center must describe the real products and match the pages. The official product data specification is the reference point for attributes and conditions. Don't invent identifiers and don't treat a rejection as a bidding problem.
Check a representative sample: a simple product, a product with variants, a discount, limited stock and an important category. Look for mismatches between image, variant, price and availability. An error repeated in the source must be corrected there, otherwise it reappears at the next synchronisation.
Materials must support the product
Texts, images and videos must express real advantages. A generic headline about “premium quality” doesn't explain why the buyer would choose the product. Use verifiable information: compatibility, use, size or commercial terms.
Evaluate the possible combinations, not just each file separately. A text about one category shouldn't be paired with an image that suggests something else. Check the previews and the limits available in the account. The availability of options can change, so avoid instructions based only on the memory of an old interface.
Destinations and signals must be verified before launch
The page the visitor lands on must continue the ad's promise. If the creative shows a specific product or offer, a general page can force the customer to search again. Check the destination settings and the expansion options available for your campaign in the current documentation and interface.
The signals offered to the automation must be interpreted according to their role, without assuming that every audience hint is a rigid boundary. Separate what you entered as guidance from the actual controls of distribution. In the report, say what you can observe and what you can't isolate.
Check the mobile experience
Open the destinations on a real phone. Choose the variant, check delivery and complete a test order. A good impressions report can't compensate for a blocked selector or a price that changes unclearly at checkout.
If work on the Shopify store is needed, prioritise the blockages that prevent choosing and paying. Separate these fixes from aesthetic changes without an observed problem. The development budget must be tied to a verifiable commercial hypothesis.
Budget and three hypothetical examples of interpretation
Hypothetical example 1: cost per order and sessions
In 30 days you have 4,800 RON in media, 2,400 measured sessions and 60 retained orders, one in each of 60 sessions. The rate of sessions with an order is 60 / 2,400 × 100 = 2.5%. The media cost per order is 4,800 / 60 = 80 RON.
These values don't say whether the investment is profitable. You need the contribution of the orders and the additional costs. Don't replace sessions with clicks in the formula without changing the definition of the metric.
Hypothetical example 2: the product mix
Category A produces 12,000 RON of net revenue with a contribution of 40% before media, which is 4,800 RON. Category B produces 12,000 RON with a contribution of 20%, which is 2,400 RON. Together they have 7,200 RON of contribution, before ads and fixed costs.
At 4,800 RON of media, 2,400 RON remains for the other costs and the result. If total revenue stays identical but shifts towards B, the contribution can fall. That is why the aggregate value of conversions isn't enough for the decision to expand.
Hypothetical example 3: a blockage after checkout starts
Of a cohort of 200 sessions that started checkout in a defined period, 70 include a purchase in the tracked path. Completion is 70 / 200 × 100 = 35%. The remaining 130 are not automatically fully recoverable orders.
You check payment methods, delivery cost and errors. If you find a technical problem, fixing it takes priority over increasing the budget. An abandonment percentage indicates where to analyse; it doesn't prove the cause on its own.
Diagnostic table before you intervene
| Symptom | Hypothesis to check | Justified action |
|---|---|---|
| Products with no impressions | Eligibility, data or selection | Check Merchant Center and inclusion |
| Traffic without purchases | Destination, offer or measurement | Test the path and relevance |
| High revenue, low contribution | Mix, returns, costs | Analyse by category |
| Recent drop | Delay, stock, changes | Let the report mature and check the log |
| Orders over-reported | Duplicate events | Reconcile identifiers |
Not all problems call for restructuring the campaign. A major change can make comparison harder if you haven't first established the probable cause. In a Google Ads analysis with BlueAds, ask for an explicit link between observation, hypothesis and intervention.
If you also use Meta advertising, compare the channels' roles on a shared record. Don't add up conversions and don't assume the platform with the last click created the demand alone. A useful evaluation keeps these limits visible.
Launch and review plan
In the first stage, document the goals, data, products and spending limit. Prepare the materials and check the destinations. Launch only after you can explain which result you are tracking and how you will recognise an error that calls for stopping.
After launch, track how things work and the spend without intervening compulsively at every variation. Keep a log of changes, promotions and stock problems. Correct obvious errors immediately; for performance differences, take conversion delay into account.
At review, compare the result with the hypothesis and the products' economics. Decide whether to maintain, correct the selection, change the materials or reduce the investment. There is no fixed duration that guarantees enough data for every store. If the volume is small, the conclusion can remain provisional.
For a check of the campaign's readiness, contact the BlueAds team. A useful conversation starts with products, measurement and contribution, then gets to settings.
Frequently asked questions
Does Performance Max automatically replace all other campaigns?
There is no universal rule. The structure must be evaluated by goals, demand and data. Compare the role of each campaign and avoid decisions based only on one type being more automated.
Can I launch without my own materials?
Check the requirements and options available in the account, but don't confuse the technical possibility of launching with commercial readiness. Relevant materials help you express the offer and check the consistency of messages.
Why can't I see all the information I want?
Reporting depends on the campaign type and the features available. Don't infer details the report doesn't provide. Complete the analysis with the products bought, the store's records and the explicit limits of attribution.
When do I change the budget?
After checking measurement, data maturity and contribution. Also analyse stock and cash. A good result at low volume doesn't prove that the same efficiency holds at any investment.
Is a rejected product solved with more budget?
No. Check the reason and the product data, then follow the relevant fix process. Spend doesn't replace eligibility and doesn't correct a mismatch between the feed and the page.
What does a good ROAS mean?
A ratio compatible with your contribution and costs, on the same revenue basis. Don't use a universal threshold. Separate the platform's metric from the result after returns and full costs.
Can I compare results after a few days?
You can check how it works, but economic conclusions can be premature. Volume, buying cycle and reporting delay matter. Note what is observed and what remains hypothesis.
Final checklist
- Goals and values are verified.
- Products are eligible and available.
- Materials support the real offer.
- Destinations work on mobile.
- The budget has a limit and a clear purpose.
- Reporting includes contribution and retained orders.
- Interventions are documented.
Performance Max is worth evaluating as a system fed with data and decisions. The clearer the preparation, the better you can interpret the result and choose the next step.