In Shopping and Performance Max, the product feed is the targeting. Here is what to fix in it, in what order, and why bid tinkering rarely moves the number.
In Shopping and Performance Max campaigns, your product feed is the targeting. Titles, attributes, images and availability decide which queries you appear for, and no bid strategy can buy relevance the feed has not supplied. Fixing the feed is the highest-return work available in ecommerce paid search, and it is the work almost nobody wants to do.
The reason is understandable. Bid strategies and campaign structures are visible, discussable and satisfying. Feed attributes live in a spreadsheet somebody exported from the platform in 2023. But when we take over an underperforming ecommerce account, the feed is where the problem is far more often than the bidding is.
Why the feed is the targeting
In a text search campaign you tell the platform which queries to match. In Shopping and Performance Max you do not. The platform reads your product data and decides which queries each product is relevant to. Your keywords, in effect, are your product titles and attributes.
That has two consequences worth sitting with. First, a product with a poor title is not being under-bid, it is being under-matched, and no amount of budget will surface it against the right query. Second, when the platform cannot tell products apart, it spends the budget on whichever ones it can understand, which are rarely your best margin lines.
Automated bidding is genuinely good now. It is good at optimising within the relevance the feed gives it, and helpless outside it.
What to fix, in order
1. Titles
The single highest-leverage field. Most feeds inherit whatever the merchandising team typed into the CMS, which is written for a customer already on the page rather than for someone searching cold.
A title that works loads the front with what people search: brand, product type, then the differentiating attributes that appear in queries. Size, colour, capacity, material, compatibility, pack quantity. The pattern that performs varies by category, and the way to find yours is the search terms report rather than a template from a blog.
The test is simple. Read a title without the image. If you cannot tell what the product is and who it is for, neither can the platform.
2. Product type and category
Google product category tells the platform what shelf the item belongs on. Your own product type field tells it how your business organises the range, and it is the field most often left empty or filled with a single top-level value. Populated properly, with a full hierarchy, it becomes the structure you use to segment campaigns and report on margin.
3. Attributes that appear in queries
Colour, size, material, gender, age group, condition, GTIN and MPN. These are dull and they are how the platform matches long-tail queries where intent is sharpest and competition is thinnest. Missing GTINs in particular can quietly restrict eligibility on the products you most want shown.
4. Images
Feed images are judged on clarity at small sizes, not on art direction. Watermarks and promotional overlays cause disapprovals. Lifestyle shots that hide the product cost you the click. The additional image fields exist and most feeds leave them empty.
5. Availability and price accuracy
Feeds that lag stock or price generate disapprovals and, worse, spend on items that cannot be bought. This is an integration problem rather than a marketing one, and it needs whoever owns the platform in the room.
6. Custom labels
The most underused fields in the feed. Custom labels let you tag products by margin band, by season, by stock depth, by bestseller status, by clearance. Once tagged, you can structure campaigns and budgets around commercial reality rather than around whatever the platform inferred. If you do one advanced thing with your feed, do this.
What we check before touching bids
Pull the search terms report and read it as a diagnosis of the feed. Queries you are matching that you should not be usually mean titles are too vague. Queries you are clearly not matching that competitors are winning usually mean an attribute is missing.
Then compare impression share against your own margin data by product type. The pattern we find most often is that the platform is confidently spending on the products it understands best, which correlates with which products happened to have good data rather than with which ones make money.
Then check disapprovals and warnings, including the ones nobody escalated because they only affect a fraction of the catalogue. A fraction of the catalogue is often a disproportionate fraction of the margin.
Only after that does bid strategy become worth arguing about, because until then you are optimising the allocation of a budget across a distorted picture of your own range.
Where this connects to the rest of the programme
The work overlaps almost entirely with ecommerce SEO. Better titles and attributes improve organic product visibility at the same time, and the underlying discipline, owning your product data rather than inheriting it from suppliers, is the same one described in our ecommerce SEO audit framework. Our piece on Google Shopping strategy covers the campaign side, and paying for clicks you already rank for covers the overlap question directly.
The commercial case for doing it properly is straightforward. One B2B client saw a 610 per cent increase in acquisition through better budget allocation, forecasting and performance optimisation, and another reduced cost per lead by 32 per cent against marketplace benchmarks while cutting reliance on those marketplaces by £120,000 a year. Neither of those came from a bid strategy change.
Questions we get asked
Should I use Performance Max or standard Shopping?
Performance Max is the default now and it makes feed quality more important rather than less, because you have given up most of the other levers. If you run it, insist on the reporting that shows performance by product and by asset group, and use custom labels so those groups mean something commercially.
How often should the feed update?
As often as your prices and stock change, which for most retailers means daily at minimum and continuously if you run frequent promotions. Scheduled daily fetches are fine for stable catalogues and inadequate for anything with fast-moving stock.
Do I need a feed management tool?
Below a few hundred products, usually not, if someone owns the data. Above that, a tool pays for itself quickly, mostly through rule-based title construction and the ability to fix an attribute across the catalogue without a developer. The tool is not the strategy though. Plenty of accounts run expensive feed tools over data nobody has thought about.
How much does ecommerce PPC management cost?
The same drivers as any paid search engagement: media spend, channel scope, catalogue complexity and who does the work. Our view on the market is in what a PPC agency costs in the UK. What is specific to ecommerce is that feed work is real work, so a scope that does not mention product data is not a complete scope.
Will fixing the feed help my organic listings too?
Frequently yes. Better titles and attributes tend to improve both, and free listings draw on the same feed. One beauty retailer we work with saw product title optimisation lift organic listing click-through to 3.4 per cent, alongside a 400 per cent increase in online revenue in the first year.
What is the fastest thing I can do this week?
Export the feed, sort by spend, and read the titles of the top fifty products aloud. You will find the problem in the first ten.
The short version
Feed first, structure second, bids third. Most ecommerce accounts do that in reverse and then conclude the channel does not work for them.
Our search advertising service sets out how we run paid search for retailers, and paid media as a commercial growth channel covers the wider programme. If you would like a senior read on your feed before you touch anything else, get in touch.