How a product set works
A catalogue holds everything you sell. A product set is a saved rule that picks out part of it — winter jackets, items above a certain price, one brand, everything in a particular category — and an ad set then advertises only what that rule returns. The set is not a fixed list of items but a filter that re-runs, so a new product matching the rule joins the set on its own and a discontinued one drops out.
The rules read the fields in your catalogue, which is why they are only as good as the data underneath. If nothing fills the brand field, no rule can filter by brand. If categories are inconsistent between products, a category rule will quietly miss half the stock it should have caught. Most disappointing sets are a data problem wearing a targeting costume.
Why product sets matter
They are how you stop treating a shop as one undifferentiated pile. A gift range and a bulk business line have different buyers, different margins and different sensible bids, and separating them into sets lets each campaign be judged on its own terms. It also lets you push what you want pushed — new arrivals, overstock, a seasonal range — instead of leaving it entirely to the system to decide.
Margin is the argument people miss. Reported return on ad spend treats every rupee of revenue the same, but your accounts do not. Splitting high-margin lines into their own set means you can bid harder where the profit actually is, which matters more than the headline revenue figure on the dashboard.
Common mistakes with product sets
The first is a set so narrow that the campaign starves. A handful of items gives the delivery system very little to work with, and performance becomes erratic. The second is the opposite: one set containing the entire catalogue, which hands every decision to the algorithm and leaves you nothing to steer with.
The third is set-and-forget. Rules built around a season, a price band or a promotion keep running long after they made sense, so a summer set spends through the winter. And a set built on a field nobody maintains slowly empties as new products arrive without it, which shows up as declining reach rather than an obvious error.
How to act on it
Build sets around decisions you would genuinely make differently: margin tiers, bestsellers against the long tail, categories with different buying cycles. If two sets would get the same budget and the same bid, they should probably be one.
Prefer rules over hand-picked lists, so new stock is included automatically, and make sure the fields those rules depend on are actually populated in the product catalogue. Review the item count in each set every few weeks — a set shrinking towards empty is the earliest warning that your feed has drifted. Then check performance per set rather than per campaign, because that is where the useful differences show up in ecommerce advertising.