Why Performance Max concentrates its budget on a few products

The concentration is a consequence of the bidding logic, not a defect

Performance Max does not spread budget across the range, it bids in every single auction. According to Google, target ROAS predicts the value of a possible conversion on every search and bids high if a valuable conversion is likely, otherwise low (Google Ads Help). A product with many sales has a robust forecast. A product without sales has none, gets low bids, loses the auctions and therefore collects no data either. Optmyzr describes exactly this loop: without volume, Google does not know a product's conversion rate and therefore does not keep testing it (Optmyzr).

How pronounced this is shows in an agency analysis by Lyonic of 156 accounts with at least 50 products: at the median, 74% of product costs went to the 10% of products with the highest costs, and 44% to products that brought no sale in three months (Lyonic). Concentration is therefore not the same as efficiency.

The usual reaction, pushing the neglected products into a campaign of their own, skips one question: does this product, at its price and margin, deserve paid traffic at all?

The calculation: what click price a product can afford

The highest click price a product can bear without a loss is roughly:

Price × contribution margin rate after returns, before ad spend × expected conversion rate

The contribution margin rate is the same figure from which you derive break-even ROAS. The result is the break-even click price. If you want to earn something after advertising, subtract your minimum contribution margin from the rate beforehand. A worked example with invented, round numbers:

Under these assumptions, Product A cannot advertise profitably, whichever campaign it sits in. If Performance Max ignores it, that is commercially correct. Product B, by contrast, has room. If it still gets no impressions, there is a problem worth solving.

You do not know the conversion rate of a product without history. Use the rate of comparable products from the same category and price band, not the account average. You estimate the click price of the search queries from comparable products in the account or from Keyword Planner. Both are estimates. For the decision they are enough as long as the gap is clear. If the affordable and the paid click price are close together, a small test decides.

Order of checks before you change anything

1. Can the product be served at all? According to Google, problems with listing groups or the product feed can be the cause if budget is not spent or impressions are missing (Google Ads Help). Google advises listing only approved products and using the Product issues column to spot missing attributes and offers that perform weakly. The Diagnostics tab on the Products page shows problems with individual products (Google Ads Help). A disapproved or incomplete product is not a structural problem.

2. Is the price position right? Merchant Center has a price competitiveness report (Google Ads Help). A product that is clearly more expensive than the same product at other retailers loses clicks and purchases. More on this in price position and Shopping reach.

3. Does the product pass the click price calculation? If not, the right measure is price, purchasing or no action at all. Advertising does not change the calculation.

4. Only then: structure. Only products that pass steps 1 to 3 are candidates for a campaign of their own.

Finding products without impressions

In Google Ads, the Products page and the campaign's Listing groups tab show the figures per product. Filter for zero impressions over a period in which the product was continuously approved and in stock. When comparing, note this: if several products are shown in one ad, Google records an impression for each product. Product totals therefore differ from the campaign figures (Google Ads Help). For the question of whether a product runs at all, that does not matter. For shares of the budget, you should look at cost per product instead.

Then lay this list next to price and contribution margin. Only the combination tells you whether you are looking at a delivery problem or a profitability problem.

A campaign of its own has a price: less data per campaign

The common countermeasure is a Performance Max campaign without assets that contains only products without clicks or impressions. Of three accounts that JumpFly shows with it, two reached their ROAS target (JumpFly).

Against that stands Google's own statement: performance improves if you run fewer, larger campaigns that get more conversions (Google Ads Help). For new Performance Max campaigns, Google recommends a structure as simple as possible with a unified budget (Google Ads Help). A campaign made up of products that have so far sold nothing starts with almost no conversions. Its bidding model has little to learn from. Roughly how many conversions a value-based strategy needs is covered in how many conversions target ROAS needs.

There is also a second effect. If you give the leftovers campaign a lower target ROAS so that it serves at all, then according to Google you buy more auctions and volume (Google Ads Help). For products that fail the click price calculation, that is paid-for loss. And if the budget is tight, Performance Max may, according to Google, look for cheaper inventory such as YouTube or Display (Google Ads Help). The leftovers campaign then delivers reach, but not necessarily the Shopping clicks you wanted for the product.

When a campaign of its own makes sense

Google itself names cases for separate campaigns: new products, best sellers, seasonal events, when different budgets or ROAS targets are needed (Google Ads Help). This can be implemented with custom labels in the feed, which mark products for campaigns or asset groups.

I would only separate when two conditions are met at the same time:

• The product passes the click price calculation by a clear margin.
• There is a retailer reason why it needs visibility now: a new item without history, a clearance of tied-up capital or a line that is to be built up strategically.

The campaign then gets a fixed test budget and a stop criterion set in advance, for example: if after a certain number of clicks the contribution margin after ad spend is negative, the products go back or out of advertising. PPC Hero too, in its model, deliberately takes products with little potential out of the ad spend (PPC Hero).

Are products that do not serve a problem?

Not automatically. A shop does not need every product in paid advertising. Cheap add-on items, accessories with a small contribution margin or niche variants often sell via the basket, on-site search or organic traffic. It is a problem if a product with a good contribution margin and a competitive price never gets a chance for lack of data, or if, as in the Lyonic analysis, a large part of the budget lands on products without a sale.

The limit of this view: it assesses each product individually. If a cheap product typically triggers first purchases that come back later, its affordable click price may be higher than the simple calculation shows. But that applies only if you can prove the repeat purchases, not as a hope. For an overview of what Performance Max otherwise conceals, see Performance Max for online shops.