Performance Max is not magic: it learns from the signals you give it. Tell it that success is a purchase and it will look for users who resemble people who buy. Tell it that success is an add to cart and it will look for users who resemble people who add products to a cart.
The trouble is that plenty of eCommerce businesses look at the "Conversions" column without checking what is inside it. And sometimes you find everything in there: purchases, carts, checkout starts, calls, product views, forms, events badly imported from GA4 and secondary actions filed under the wrong logic.
The outcome can be dangerous: the dashboard looks positive, the cost per conversion looks low, but orders are not growing. The campaign is not learning to sell. It is learning to produce micro-conversions.
Performance Max follows the goal you give it
Google Ads uses conversion goals to optimise campaigns. In the official documentation, Google distinguishes primary conversion actions, which are included in the "Conversions" column and can be used for bidding, from secondary actions, which are useful for observation and reporting.
For an eCommerce site, that distinction is fundamental. A completed purchase is a macro-conversion. An add to cart is a micro-conversion. Both are useful, but they are not worth the same.
If you set "Add to cart" as a primary action, the campaign can treat it as a result worth chasing. And since a cart is far easier to obtain than a paid order, the numbers can start looking good very quickly.
Except that carts do not empty your warehouse. Orders do.
Why the cart is a weak signal
Tracking the cart is the right thing to do. In GA4, events such as add_to_cart, begin_checkout and purchase exist precisely to help you understand the eCommerce funnel. The problem is not measuring them. The problem is giving them all the same weight.
A cart can mean a lot of different things:
- a user who is interested but not ready to buy;
- a user comparing price and shipping;
- a user who abandons once they see the final costs;
- a user treating the cart as a wish list;
- a user who adds products and then buys through another channel;
- bots, poor-quality traffic or unqualified clicks;
- checkout, payment or trust problems.
A purchase, on the other hand, is far clearer: somebody paid. That is why, in an eCommerce campaign, the purchase event has to sit at the centre.
The symptom: plenty of conversions, few orders
The classic case looks like this: the account shows hundreds or thousands of conversions, but the back office tells another story. Real orders are far fewer. The cost per conversion looks fantastic, ROAS looks acceptable or confusing, and the agency talks about the "learning phase", the "algorithm", "insufficient budget" or a "competitive market".
Before you accept those explanations, run a very simple check: segment conversions by conversion action. You need to know how many are genuine purchases and how many are carts, checkouts, calls or secondary events.
We showed a real example of this in the article 978 conversions but only 103 orders: the total looked excellent, but most of those conversions were "Add to cart".
A clean setup for an eCommerce site
A healthy setup does not have to be complicated. It has to be hierarchical.
- Completed purchase: primary conversion, used for bidding and for the "Conversions" column.
- Order value: the real amount passed correctly, not an invented value.
- Add to cart: secondary conversion or analysis event.
- Begin checkout: secondary conversion or diagnostic event.
- Product view: a useful metric for funnel and remarketing work, not a final goal.
- Calls or forms: to be used only when they genuinely form part of the sales process.
If the eCommerce site sells online, the campaign needs to know that the goal is selling online. It sounds obvious, but many accounts are configured to reward the easier and less profitable actions.
Primary and secondary conversions: the difference that changes everything
The distinction between primary and secondary is one of the most important checks to run in Google Ads.
Primary conversions are the ones that sit at the heart of evaluation and optimisation. Secondary ones can stay visible in reporting, but they must not drive bidding as if they were sales.
That means you can carry on measuring the cart — in fact you should. You just must not confuse it with an order.
A serious report should show at least:
- number of orders from Google Ads;
- order value;
- spend;
- ROAS;
- cost per purchase;
- carts;
- checkouts started;
- cart → purchase rate;
- returns or cancellations, where available;
- margin or margin band, if the feed allows it.
Conversion value: another place where damage gets done
Performance Max can also work on value-based strategies, such as maximising conversion value or target ROAS. But if the value passed to Google Ads is wrong, the strategy becomes wrong too.
The frequent mistakes are:
- passing a value for carts as well and adding it to order values;
- importing the same purchase twice, from GA4 and from the Google Ads tag;
- passing gross value without accounting for returns and cancellations;
- failing to distinguish high-margin from low-margin products;
- leaving a fixed value across different conversions;
- counting coupons, shipping or taxes inconsistently with respect to margin.
ROAS is only useful if the conversion value is credible. Otherwise it is a nice percentage built on weak data.
When does it make sense to use micro-conversions?
There are cases where a micro-conversion can be watched carefully or used temporarily. New accounts, for example, or very few orders, catalogues with a long purchase cycle, or early phases where you need to gather data.
But it has to be an explicit, temporary and documented choice. It must not become the way you tell the client that the campaign is working.
The rule is simple: if you use the cart as a proxy, you have to say so. You have to show the figure separately from orders. You have to explain when you will switch to purchases as the main goal. And you have to check whether those carts really are turning into orders.
How to check your account in 10 minutes
You do not need to be a specialist to run the first check.
- Open Google Ads.
- Go to Goals → Conversions → Summary.
- Check which actions are marked as primary.
- See whether "Add to cart" or "Begin checkout" are primary.
- Open the campaign report.
- Add the segment by conversion action.
- Compare purchases, carts and checkouts.
- Check the conversion value for each action.
- Compare Google Ads orders with the real orders in the CMS.
- Ask the agency why each action is primary or secondary.
If the agency cannot answer, or answers with vague formulas, you have already found a problem.
What to ask your agency before increasing budget
Before you put more money into Performance Max, ask these questions:
- What is the campaign's primary conversion? It has to be clear whether you are optimising for purchases or for micro-actions.
- Is order value being passed correctly? Counting orders is not enough; you need to know what they are worth.
- Are conversions segmented by action? The aggregate total is not enough.
- Are GA4 and Google Ads duplicating purchases? Double tracking means distorted data.
- Are cart and checkout primary or secondary? If they are primary, there needs to be a strong reason.
- Does the report compare Ads data with the back office? Ad platforms must not be the only source of truth.
- Are you looking at margin or only at revenue? Not every product deserves the same budget.
- Are any products excluded or segmented? PMax should not push everything the same way when the catalogue is uneven.
- What is the learning plan? There needs to be a path, not an eternal "we just have to wait".
- What happens if orders do not grow? Hypotheses are there to be tested, not defended.
The problem is not Performance Max. It is feeding it the wrong data
Performance Max can work very well on an eCommerce site, but it needs clean data, a well-maintained feed, correct conversions, reliable values and a strategy that matches your margins and catalogue.
If instead you feed it carts, checkouts, pointless calls or wrong values, you cannot be surprised when it brings you the wrong results. Automation amplifies the signal it receives. If the signal is dirty, it scales the dirt.
It is the same reason why real Google Ads work does not stop at switching campaigns on. It has to connect platform, tracking, catalogue, feed, margins and real data. We covered this too in the article on why you need a genuine Google Ads expert for eCommerce.
How we handle it at BitHub
When we work on Google Ads for eCommerce, we start from the figures that matter: orders, value, margin, traffic quality and consistency with the catalogue.
We check primary and secondary conversions, GA4, Tag Manager, Merchant Center, product feeds, landing pages, facets and category pages. If the site is ours or runs on a custom platform, we can also connect catalogue data, margins, products worth pushing and real sales signals more tightly.
That is one of the advantages of bespoke eCommerce development: marketing is not separate from the site. Tracking, feed, URLs, facets and orders all speak the same language.
Frequently asked questions
Does Performance Max work for eCommerce?
Yes, it can work very well, but it depends on data quality. If conversions, order value and product feeds are wrong, automation can optimise towards results that are not profitable.
Should add to cart be a conversion?
It can be tracked as an event or as a secondary conversion for funnel analysis. As a rule it should not be the primary conversion used to judge the commercial success of an eCommerce site.
Which conversion should drive Performance Max?
For an eCommerce site, normally the completed purchase with the correct order value. Carts and checkouts help you diagnose the funnel, but they should not replace orders.
Why do I see plenty of conversions but few orders?
The conversions column probably includes micro-conversions such as carts, checkouts, calls or other events. You need to segment conversions by action and compare them with the site's real orders.
Can ROAS be distorted?
Yes. If conversion value is duplicated, incomplete or also assigned to micro-conversions, ROAS can look better or worse than reality.
Can I use carts if I have few orders?
In some cases they can be used temporarily as a secondary signal or proxy, but this has to be stated and monitored. The goal must remain getting to the point where you optimise on real purchases.
How do I tell whether Google Ads is duplicating purchases?
Check the conversion actions imported from GA4 and those generated by the Google Ads tag. If the same order is counted by two sources without correct deduplication, the data can inflate.
What should I ask the agency every month?
Real orders, order value, spend, ROAS, cost per purchase, a breakdown by conversion action, a comparison with the CMS and, where possible, estimated margin.
Do you have Performance Max campaigns running without knowing whether they are bringing in real sales? We can check conversions, GA4, Tag Manager, Merchant Center, feeds and reports to work out whether Google is learning from orders or from carts.
Get in touch for a Google Ads eCommerce audit.
Useful sources: the Google Ads documentation on primary and secondary conversion actions, the guide to conversion goals, best practices for Performance Max and the GA4 documentation on eCommerce events.