What determines performance in shopping ads is often not the bid strategy but the quality of the product data. The most common feed errors and their effect on ads.
In shopping ads you don't choose keywords; Google and Meta decide which product to show for which search by looking at your product data. So in these campaigns most of the performance is determined not on the campaign screen but in the product feed.
1. Titles not written in the language of search
No one searches for the product you named "Classic Model 3-Pack" inside your store by that name. The title in the feed should follow the order people actually type: brand + product type + defining feature + size/color. The first words of the title carry the most weight, because the part trimmed in the list is the end.
2. Category assignment wrong or missing
The product category determines which competition pool the platform puts your product in. The wrong category means never reaching the right buyer. Choosing a general top category harms similarly: the more specific you are, the more accurate the match.
3. Images not following the rules
Images with "50% OFF" written on them, carrying a watermark or framed are rejected on most platforms. A rejected product never goes live, and this usually goes unnoticed — the campaign looks like it's working, but part of the catalog isn't live.
4. Stock and price data not reflecting reality
If the feed updates once a day, a product that sells out in the morning stays in the ad until evening. The clicking user meets an "out of stock" page. This is both a budget loss and, if repeated, lowers the account's credibility. Price mismatch is more serious; if the price on the site differs from the price in the feed, the product can be suspended.
5. Distinguishing fields left empty
Fields like GTIN, brand, condition, size, color, material are skipped as "not mandatory." Yet these fields directly increase the accuracy of the match. In particular, products without size and color information can't be shown to users searching for that feature.
6. Variants sent as a single product
When different colors and sizes of the same product are sent as a single record, the user can't find the variant they're looking for and lands on an irrelevant page. Variants should be sent as separate records, with a shared item group id.
The order in which to fix the feed
Instead of trying to fix everything at once, this order works:
- See the rejected products and close their reasons one by one — a product not live benefits from no optimization.
- Overhaul the title and image of the top 50 highest-revenue products.
- Increase the frequency of stock and price updates.
- Only then deal with the bid strategy.
Feed health must be monitored regularly
A product feed is not something set up once and forgotten. Small changes on the site side — a theme update, a change in category structure, a new variant field — can silently break the feed. The breakage usually shows itself not by the campaign stopping but by gradually decreasing impressions; that's why it's noticed late.
There are three numbers to check weekly: the number of products live, the number of products rejected and the number of products flagged with a warning. If there's a sudden change in one of these three, the cause should be investigated at once. Warnings are not rejections but are often the herald of rejection.
Frequently asked questions
Should I send the feed manually or automatically?
If the number of products is small and rarely changes, a manually uploaded file does the job. If stock and price change often, an automatic connection is mandatory; manual updates always stay behind reality.
Should I put all products into ads?
Usually no. Separating out products with low profit margin, uncertain stock or high return rate makes the budget more efficient.
The product description doesn't appear in the ad, is it still important?
Yes. Even if the user doesn't see it, the platform reads the description for matching. Descriptions you write yourself provide a better match than copy-paste manufacturer texts.
