How many products should a shopping campaign have?

There is no correct number of products, but there is a correct way to arrive at yours. Take your monthly budget, divide by your average cost per click, and you have monthly clicks. Divide those by the number of products in the campaign, and you have the average attention each product will receive.

If that number is small, the campaign cannot teach you anything about most of its products, and automated bidding has almost nothing to work from at the product level. Spend will concentrate on a handful of items anyway, usually whichever are cheapest and attract the most clicks, and the rest will sit there generating impressions.

So the practical question is how many products can your budget actually support. A modest budget supports tens of products meaningfully. A large budget supports thousands.

This is why the same catalogue needs different structures at different budgets. A store spending a little should advertise its best products properly rather than its whole range badly. As spend grows, the eligible set grows with it.

Start narrow. Adding products later is easy. Recovering from a campaign that spread a small budget across three thousand items takes longer, because you cannot tell what worked.

The instinct is to split by category, because that is how the catalogue is organised. A better split is by the role a product plays commercially, because that is what determines how you want to bid on it.

Bestsellers with healthy margin. These deserve their own campaign and their own target, because they carry the account and you want to see them clearly.

High margin products that are not yet selling much. Worth a separate group so they can get budget without competing against the bestsellers.

Thin margin or heavily discounted lines. If they are in at all, they need a different target, because the ratio that makes a bestseller profitable will lose money on them.

Seasonal or launch items. Separate so they can be turned up and down without disturbing anything else.

The long tail. Everything that sells occasionally, grouped together and expected to perform modestly.

Splitting this way gives you a structure where each campaign has one commercial intention and one target, which is what makes reporting readable and bidding sensible. Splitting by product category gives you campaigns that each contain a mix of good and bad economics, and no target that suits both.

Most catalogues have a large group of products that sell rarely. They are not worthless, but they should not be treated the same as the products that pay the bills.

The first decision is whether they belong in paid campaigns at all. They can stay in your Merchant Center feed for the unpaid listings whether or not you advertise them, which means the tail can earn traffic without spending anything. That is often the right answer.

If you do advertise them, put them in one campaign with a low budget and a target that reflects what they are: an experiment. Do not expect the reporting to be readable per product, because it will not be. Read it as a group.

Run a periodic promotion out of the tail. Every few months, look for items that received clicks and converted, and move them up into a properly funded campaign. That is how products graduate from the tail rather than sitting in it forever.

And exclude the genuinely uneconomic. Items with almost no margin, items you are discontinuing, items with poor product pages and items you cannot restock should be removed from campaigns entirely. Advertising something you cannot fulfil profitably is a cost with no upside.

A campaign is defined by what is in it, and most structures leak because nobody set the exclusions.

Make sure every product sits in exactly one campaign for a given purpose. When the same product is eligible in two campaigns, they compete, the reporting splits and neither campaign shows a clean picture. Use inventory filters or campaign priority deliberately so eligibility is unambiguous.

Exclude out of stock items automatically. Paying for clicks on something a shopper cannot buy is pure waste, and it damages the experience enough to affect whether they return.

Exclude items below a margin threshold, unless you have a specific reason such as an entry product designed to acquire customers. If you do keep such products, run them in their own campaign so the deliberate loss is visible rather than hidden inside an average.

Exclude products with poor pages until the pages are fixed. Spending on traffic that lands on a page with one photograph and two lines of description is buying a bounce.

Review the exclusions monthly. Catalogues change, stock changes and prices change, and an exclusion rule set once and forgotten stops matching reality within a season.

The right number of products is a function of budget, so it moves when the budget moves.

When spend increases, the constraint loosens. More products can gather enough data to be worth managing, so widen the eligible set gradually rather than all at once, and watch whether the new products earn their share.

When spend decreases, do the opposite. Narrow to the products that were carrying the account. Cutting budget while keeping the same broad product set is how a campaign becomes ineffective at every level at once.

Seasonality counts as a budget change. In a peak period the same money buys fewer clicks because competition rises, so the number of products your budget can support falls even if the figure on the page has not changed.

Set a quarterly review. Look at spend and conversions by product, promote the items that earned it, demote the ones that did not, and check that the biggest spenders are the ones you actually want to win.

And write the rule down. Structures drift because nobody remembers why a product was in a particular campaign. A one page note on what each campaign is for keeps the account coherent as people and priorities change.