Our team looked at four stores where product ads did not separate the assortment by line of business. In one, the whole range sat in a single group, and part of the assortment never entered the campaign. In the others, products with different performance were managed together, or there was no split by calculated performance at all. Company names stay out. What remains is the niche and the country.
How to build groups and set priorities across campaigns is already laid out in the article on Shopping product groups and priorities. This page is different. These are the live findings: what was in the account, and why that makes the assortment hard to run.
I have run ads since 2006, and I keep seeing the same shelf. Different lines in one store sit on one bid. The account then cannot tell which product earned that bid.
How one group holds the assortment
Shopping ads, also called Google Shopping, show a photo, a title, and a price. Inside a Shopping campaign you bid on a product group, not on a keyword. The guide to managing a Shopping campaign with product groups says it plainly: a product group is a subset of your inventory that you define, and every product inside it uses the same bid. When you first create the campaign, you get one ad group and one product group called “All products” that holds the whole assortment. You cannot set a bid on a product group that has already been subdivided. The bid sits on the groups that have no further split.
You can subdivide a group only by an attribute from the product data in Merchant Center: item ID, brand, category, product type, or custom label. The ads themselves use that data, not keywords. How that works is in the guide to Shopping ads and in the article on the product feed in Merchant Center. If a product is not in the data the campaign can see, the group has nothing to pull it from.
The guide to the Products tab names product groups in a Standard Shopping campaign and listing groups in a Performance Max campaign. In a Standard Shopping campaign, an undivided group has one bid. In a Performance Max campaign, Google AI sets the bid from the campaign goal. Listing groups sit inside the asset group. Subdividing by category is not available for every target country: Ukraine and Spain are not on the list, and the United Kingdom is. How the types differ is in the article on Performance Max versus standard Shopping. Excluded means the product is not in an active group. That definition is not a store’s finding.
B2B online store for industrial goods in Ukraine: one group and part of the assortment outside the campaign
In the B2B store for industrial goods in Ukraine, our team saw one group on everything advertised. All products were combined into one product-ad group. Part of the assortment was missing from the campaign. That made it hard to distribute product categories and signals across lines of business. Assortment management stayed opaque.
For the owner, these lines are not one product. A category left inside the group shares its bid with the next line. A signal that stays in the same group trains on the mix. The finding does not name the kind of signal. A part of the assortment missing from the campaign and the status Excluded are different checks. Excluded means the product is not in an active group. An inventory filter in the Shopping campaign settings lets only some items into targeting. A product marked disapproved for Shopping ads may no longer be targeted in the product group, and that mark is not the Excluded status. The Products tab shows which items can appear in ads and which cannot. The finding does not name the cause.
Sales and repair of electric scooters and batteries in Spain: different performance in shared standard product advertising
In Spain, electric scooters and batteries with different economic performance were managed together. This was standard product advertising. Combining the groups complicated separate assortment management. Alongside that, it was hard to determine which conversions and value should be passed into optimization.
A scooter and a battery are not the same event for this business: their performance differs, and that is what the finding records. While they share one group, they share one bid. The guide to conversion values describes a second layer. An assigned value shows the effect on the business, not only the number of actions. Which actions and which value to pass into optimization are harder to see in a merged group. Clicks for a brand show up when you filter the products view, with no new product group. The owner then cannot manage the scooter in its own group with its own bid, and the battery in another, and cannot see which event the campaign should learn from.
Online jewelry store in Ukraine: shared standard product advertising and uncontrolled queries
In the jewelry store in Ukraine, different economic performance also sat inside standard product advertising. A second gap sat next to it. Product and search campaign queries were not properly controlled, so traffic relevance stayed inconsistent.
Jewelry with different performance shares one bid until the group is split. A search query the account does not control can bring a person to the wrong product. For product ads, the guide to negative keywords in a Shopping campaign says the campaign does not target by keywords, but a negative keyword can keep the ad off a query where you do not want it. In this account neither the product-ad queries nor the search queries were properly controlled with a negative keyword on the Shopping campaign. The finding does not list those queries. From the outside it looked like unsteady relevance: traffic sometimes reached the piece the owner wants to sell, and sometimes it reached a different piece.
Online store of children’s goods and toys in the United Kingdom: no split by calculated performance
In the United Kingdom, the store of children’s goods and toys had not divided the assortment into groups by calculated performance. Each part of the offering could not be managed separately. Product-ad queries also required optimization. That compounded the campaign-structure issue.
Calculated performance is the store’s own estimate, not a ready-made button in the account. The account splits a group only by an attribute in the product data. A value the store calculated itself can be written as a custom label. Among the examples in the guide to custom labels are sales performance, such as Best Seller, and low margin or high margin. In this store there was no split by calculated performance, so each part of the offering could not be run on its own. Product-ad queries that still required optimization sat on top of that structure. The owner could not manage one part of the assortment on its own, and could not control the queries that part was showing on.
How to check your account
Repeat the checks below in your own account. The All products group is how a campaign starts, not a verdict. A negative keyword and a custom label are help tools, not a conclusion about the four stores above.
- In the page menu, click Campaigns. Click the Shopping campaign, then its ad group. Click Product groups in the page menu. See whether the assortment still sits in All products, and whether one bid covers everything inside it.
- If the group is already subdivided, check the attribute in the menu Subdivide by: item ID, brand, category, product type, or custom label. The bid sits on a group that is not subdivided further. A parent group that has subgroups does not carry a bid.
- Check products with the status Excluded. The help says that product is not in an active group. Next to that, see whether part of the assortment is missing from the campaign entirely. The data a group draws products from lives in Merchant Center.
- In a Standard Shopping campaign, compare lines that do not share the same performance. See whether they sit in one undivided group. If you also have a Performance Max campaign, open Asset groups, then Listing groups.
- Negative keywords. Click the Campaigns icon, open Audiences, keywords, and content in the section menu, then Search keywords and Negative search keywords. A Shopping campaign does not target by keywords. A negative keyword keeps a Shopping ad off a query where you do not want it. If search ads run beside the product ads, review those queries too. For queries that already triggered ads, use the same Campaigns icon, then Insights and reports, then Search terms.
- If you want groups by a performance you calculated yourself, write that value in the product data as a custom label. The help example, not your product name, is Best Seller. In the menu Subdivide by, choose custom label and save the subgroup.
- See which conversions and which value go into optimization. An assigned value shows the effect on the business, not only the number of actions. The question is simple: is the event the campaign learns from the sale that line is there to make?
If you want to walk the same check on your own, the step-by-step version is in my book on Google Ads for business owners.
My e-book “19 Google Ads Secrets”: how to get the most out of Google Ads without draining your budget, drawn from real campaigns and tests.
See the book →When this does not apply
This finding is not about your account if each line already sits in its own product group with its own bid, and the products you sell are inside the campaign.
This finding is not about your account if the store has one product and one level of performance, so there is nothing to split. It is also not about an account that has no product ads and no product groups.
This is an audit finding, not a result after a change.
Questions and answers
Why not run products with different performance in one group?
A product group has one bid for everything inside it. The help does not let you set a bid on a group you have already subdivided: the bid lives on the subgroups. While a scooter and a battery, or jewelry with different performance, or parts of a children’s assortment sit in one undivided group, the account has no separate bid for each part. On the Products tab, metrics are organized by product, category, brand, product type, or custom label. Which event the campaign learns from is checked apart from the group.
What did this look like in the four stores?
In the B2B store for industrial goods in Ukraine, all products were put in one product-ad group, part of the assortment never entered the campaign, categories and signals could not be spread across lines, and assortment management stayed opaque. In Spain, electric scooters and batteries with different performance were run together in standard product advertising, and it was hard to decide which conversions and value to pass into optimization. In the jewelry store in Ukraine, the same shared standard product advertising sat next to product and search queries that were not properly controlled, so traffic relevance was inconsistent. In the United Kingdom, children’s goods and toys were not split into groups by calculated performance, each part could not be managed separately, and product-ad queries also required optimization.
How do you split the assortment in your account?
Open Product groups in the campaign. If you see All products with one bid, the split has not been made. Then choose an attribute that is already in the product data: brand, product type, item ID, or custom label. Category is not available for Ukraine and Spain. A performance you calculated yourself is written as a label first, and only then does the group split on it. Product-ad queries sit next to negative keywords. The conversions and value that go into optimization are checked apart from the group: the group splits the bid, and the value tells the campaign which action weighs more.