Here's the fact that should reorder your priorities, and that a surprising number of people managing Shopping accounts have never quite internalised.

There are no keywords in Google Shopping. You don't tell Google which searches you want to appear for. Google reads your product data and decides which searches you deserve. Your feed is your targeting.

Once that lands, a lot of conventional advice starts looking odd. Enormous attention goes on bidding strategy and campaign structure, and comparatively none on the product data underneath, even though the product data is the thing determining what you're matched against in the first place.

Why titles are where the money is

Product titles are the single most important field in the feed, because they carry most of the weight in matching your product to a query.

The problem is that titles are almost always written for a human browsing a category page. "Classic Fit Shirt, Blue" reads fine on your website. To Google it's nearly information-free: no brand, no material, no size, no gender, no product type.

A title built for search looks more like Brand + Product Type + Key Attribute + Size or Variant, in roughly the order people search. Less elegant, considerably more matchable. The right order varies by category, which is why this is worth doing per category rather than as one global rule.

The account that has been "fully optimised" three times usually still has titles nobody has touched since launch.

The second thing: products that can't serve

The other common finding is a Merchant Center account quietly disapproving part of the catalogue while everyone watches the campaigns.

Disapproved products can't serve. Nothing in the Google Ads interface makes this obvious, because Ads shows you performance for the products that are running, not an absence where the others should be. You have to open Merchant Center and look.

Most disapprovals are fixable data problems rather than genuine policy breaches: missing GTINs, a price mismatch between the feed and the live page, availability out of date, an image that fails a requirement. Boring, and worth more than a fortnight of bid adjustments.

Custom labels, or how to stop averaging

The third piece is segmentation. A single efficiency target across your whole catalogue guarantees you're overspending on some products and underspending on others, because it's an average of things that behave nothing like each other.

Custom labels let you tag products by whatever actually matters to your business, most usefully:

  • Margin band. So high-margin lines can carry a lower ROAS target and still be more profitable.
  • Bestseller vs long tail. Your top few percent of products deserve different treatment from the rest.
  • Stock level. There's no point buying traffic for something about to sell out.
  • Seasonality. So the structure can react without somebody rebuilding it twice a year.

Once the labels exist, the campaign structure can act on them. Without them you're stuck bidding on an average.

The margin trap

Worth stating on its own because it catches people out. Bidding on revenue, which is the default, pushes budget towards whatever is easiest to sell. Easiest to sell and most profitable are frequently different products.

If your margins vary meaningfully across the catalogue, a revenue-based target will quietly favour your worst lines. Feeding profit rather than revenue as the conversion value, where your systems can produce it, changes the answer completely.

Do you need a feed management tool?

Usually not, at least not at first, and this is where a lot of money gets spent solving a problem that a better title would have fixed.

Shopify, WooCommerce and most platforms produce a serviceable feed. Merchant Center's own feed rules can do a surprising amount for free. A dedicated tool earns its licence when you're running several countries and languages, or when the product data genuinely can't be fixed at source.

Fix at source where you can, because your site benefits too and there's one version of the truth. Fix in the feed where you can't, which is often, because the title that converts on a product page isn't always the title that matches a query. Most accounts end up with some of both, and that's fine as long as somebody knows which rules are applied where.

Where bidding does still matter

To be fair to bidding: it matters a great deal. Targets that reflect your actual margin, sensible budget allocation, and knowing when a target is constraining volume rather than protecting efficiency are all real work with real returns.

The argument is about sequence, not importance. Bidding decides how hard you compete for the auctions you're in. The feed decides which auctions those are. Getting the second one right first makes the first one worth doing.

Written by Michael Veal, a freelance PPC consultant based in Edinburgh.
Something in here apply to your account? Get in touch, or read more in the blog.