How Long Should You Test a Product Before Killing It on Google Ads?

Day seven arrives at the same time for every product. It does not arrive with the same money behind it. Our published test floor for a new store is 50 euros a day, and plenty of people start at 30. Run both for a week and one product has eaten far less than the other. Same calendar. Different test.

So days are the wrong unit. Spend is the right one. Judge the product by the money it actually ate and the day count stops mattering.

Your target cost per order is the most you are willing to pay Google for one sale. Start from what is left on a sale after the product and the shipping have cost you. That leftover is your break-even. Your target sits at or under it.

Work it out per product, before the test starts. We never ask for or hold client cost data, so this number is yours to know. We only tell you what it has to clear.

Three times your target cost per order. Our Google Ads dropshipping mastermind post says it in one line: 3x your target spend with no target cost per order means cut. Write the number down before you launch, not halfway through a spend cycle when you are already in the hole.

Take the example from our test-budget answer. A product that costs you 8 euros and sells for 20 leaves you 12 euros. Three times 12 is 36. Spend 36 euros on that product with no sale at 12 euros or better, and the test is over.

The rule does not prove the product can never work. Sales can land late. A better page or a better price can change the answer. A handful of clicks is a small sample. What the rule does is cap what you pay to find out. That is its whole job.

There is a second published trigger, and it carries the same weight. Once the learning phase is over and the feed is clean, a product still under your break-even ROAS is a kill too. Google's own help on the learning period puts it at up to three weeks, or one to two conversion cycles: support.google.com/google-ads/answer/13020501

You need the product's own numbers, not the campaign's. Performance Max is the Google campaign type that runs your feed across Shopping, Search, YouTube, Display and Gmail from one place: support.google.com/google-ads/answer/10724817. Its headline screens report the campaign as one blended number, so it is easy to read a whole campaign and call it a product test.

Open the products report inside the campaign and pull cost, clicks and sales per item id. The item id is the code that names one product in your feed. That report is where you see whether this product ate 36 euros or 4. Without it you are killing on a guess.

One multiplier, but the number it hands you moves with your price point. The 8-euro product that sells for 20 gets 36 euros of runway. A product with a much wider gap between cost and price gets a much bigger number, because one sale is worth more.

That is why a flat 'give it 7 days' fits nobody. Seven days is the same week for a cheap impulse buy and a high-ticket product. Three times your target cost per order is not.

The kill line only counts once the product got a fair test. Zero impressions, no clicks at all, a disapproved product, a broken feed: none of that is the market saying no. That is the plumbing saying nothing got through.

Check the Products tab in Merchant Center first. Google's product data specification is the standard your feed has to meet: support.google.com/merchants/answer/7052112. A product that never served never got tested, and cutting it there throws it away before you really tried.

Four cases where you wait.

The campaign has no sales yet. Smart Bidding, Google's automatic bidding, is still in its learning period, so the numbers cannot tell you much either way: support.google.com/google-ads/answer/7065882. The product sits in one campaign with the whole catalogue, so its clicks were never its own test. It went out of stock mid-test. Or the season is wrong and you are testing a garden chair in November.

None of those four is the market rejecting the product. Fix the setup, then start the spend count from zero.

Exclude the product or label it out of the campaign so budget stops flowing to it. Keep the budget in the account and start the next test the same day. A fast kill is worth doing because it buys you another test, not because of the money you did not spend.

The product call sits inside the store-level window and never replaces it. Our published floor is 50 euros a day for 30 days, and one dead product does not end that test.