ANALYSIS
How many asset groups should you have in Performance Max?
There is no universally correct number of asset groups. Armin Memić starts with the business’s products, landing pages, data volume and economic differences.
There is no universally correct number of asset groups in Performance Max. I start with the business: which products, services or features are different enough that Google needs to treat them differently?
An asset group is a collection of creative assets around a theme or audience. Google often recommends basing groups on website categories or themes. That is a good starting point, rather than a rule that every category needs its own group. Google’s explanation of asset groups
Start with the business, rather than how many groups you can create
I start with a few simple questions:
- What are the business’s most important products or services?
- Which are the best sellers?
- Which products or features define the business?
- Where is it important to secure visibility and budget?
My basic principle is that asset groups should reflect how the business works, rather than how many groups Google Ads lets you create.
If a SaaS business has ten services but two core features clearly define it, I might start with three asset groups:
- Core Feature 1
- Core Feature 2
- A catch-all for the rest
The two most important features should ideally have strong dedicated landing pages. For SaaS and lead generation, I also sometimes use asset groups to compare landing pages in practice. If the homepage works well for a product and a dedicated landing page also exists, two similar asset groups can sometimes help compare approaches.
This is not a scientific A/B test. It is an approach I sometimes use, rather than a universal Google best practice. It requires enough data for a meaningful comparison.
When using two asset groups to compare landing pages, I also want control over where traffic goes. Otherwise the comparison becomes hard to interpret. Final URL expansion can replace the specified final URL with another site page based on the user’s query and intent.
In this setup, I therefore normally prevent Final URL expansion from selecting other destinations. With the feature disabled, PMax sends traffic to URLs specified in asset groups or a page feed. This makes the comparison more practical and understandable. It does not mean asset groups themselves provide a controlled experimental design. Google’s documentation on Final URL expansion
Budget determines how much structure makes sense
I want to consolidate data where possible and avoid overstructuring small accounts. Budget sits at campaign level and is not distributed evenly between asset groups, so there is no official rule assigning each group a fixed daily amount.
As a very rough personal rule of thumb, when budgets become very small, I sometimes think in terms of around SEK 50 per day per asset group. This does not mean each group receives exactly SEK 50, and it is not a Google requirement.
It is a check: have I created more parts than the campaign can reasonably support with traffic and data?
With a campaign budget of SEK 250 per day, I would personally rather start with perhaps three to five clear asset groups than ten small ones. Give the core product or service a real chance to collect data first.
For e-commerce: product categories, feeds and listing groups
For e-commerce, I focus mainly on:
- product category
- feed structure
- similar products
- relevant product or category pages
- creative that fits the products
T-shirts, shoes and accessories are simple examples of groups that can make sense when the range, landing pages and creative differ.
In PMax, listing groups sit within an asset group. They organise products and collections using Merchant Center attributes and determine which products are included. Google Ads can select products by category, product type, brand, item ID and custom labels, among other attributes. Google’s documentation on asset groups and listing groups
Check that products do not unintentionally take overlapping roles. When copying asset groups linked to Merchant Center, Google recommends targeting different product sets to avoid listing group overlap. Google’s instructions for copying asset groups
Naming a group correctly does not make margin a bidding signal
Margin is another important factor. I would not automatically split asset groups by margin simply because products have different margins if the campaign still optimises towards standard revenue values.
If product group A has a 60% margin and group B has a 20% margin, Google does not automatically know that SEK 100 in revenue from A is more valuable than SEK 100 from B because an asset group is called High Margin.
For margin to affect bidding, it needs to be reflected in the value or signal Google optimises towards. Splitting asset groups by margin alone may otherwise be cosmetic.
That is my assessment of how value-based bidding works in practice. The specific implementation of margin data needs reviewing in each account before it affects conversion values or bidding.
For lead generation: CPA targets and customer value may require more separation
For lead generation, I think closely about landing pages and CPA targets.
A cleaning business might offer deep cleaning, end-of-tenancy cleaning, pre-viewing cleaning, regular home cleaning and window cleaning. Deep cleaning, end-of-tenancy and pre-viewing services could share a PMax campaign with separate asset groups if the client accepts roughly the same CPA for them.
Regular home cleaning may have much greater value because the customer can become recurring. Window cleaning may have lower value and a very different acceptable CPA target.
We should not consolidate so aggressively that those economic differences disappear.
I generally try to keep campaign numbers down so Google has enough data to work with. Consolidation is not a goal in itself. It is a balance between enough data for the algorithm and enough structure for bidding to reflect the business.
My final question is therefore: Which parts of the business are different enough that Google needs to treat them differently?
For the broader approach to goals, signals and profitability, see Performance Max: my approach to tracking, structure and profitability.