How to Launch New Products in Google Ads: Show, Test, Decide

Get every new product shown and fairly tested, using a budget slice your bestsellers can't take, with data from our study of 1.4 million products.

To launch new products in Google Ads, first make sure Google can show each one. It needs a stable ID, Merchant Center approval, a custom label marking it as new and a listing group that includes it. Then give them a budget your bestsellers can’t take: a campaign or tier of their own with a lower target ROAS. Test each product on at least 15–20 clicks, the minimum in our study of 1.4 million products. Once a month, decide by a rule set in advance: keep, retest, demote or drop.

Skip these steps and a new product joins the part of the catalogue that never gets an impression, so it never gets a test at all. This guide takes the steps in turn, with what our data shows and what to check in your own account. Launching is one part of a wider job, assortment planning: deciding what to carry, what to add and what to keep.

Why do new products get no impressions in Performance Max?

Performance Max decides on its own where your budget goes. Google describes the campaign type as one that “automatically funnels your budget toward the highest performing inventory and placements” (About Performance Max campaigns). A product added yesterday has nothing to show for itself: no clicks, no conversions, no conversion value. Products with a record of sales get the budget, and the new product waits.

The first gate is the impression. Until Google shows a product, nobody can click it, buy it or judge it.

You also can’t buy those first impressions with a higher bid on one product. According to Google Ads Help (Manage a Performance Max campaign with listing groups), Google AI sets bids in Performance Max based on the objective you chose for the campaign. What you control is where the product sits, which budget it competes for and which target that budget chases.

74.1% of products with no ad history were never shown

In our study of 1.4 million products, 1,196,357 products across 130+ stores started the 13 months with zero ad spend. This is what happened to them:

What happened over 13 months	Products	Share
Never received a single impression	886,245	74.1%
Received budget, but no sales	279,106	23.3%
Reached at least one conversion	31,006	2.59%

The products that did reach a sale took a median of 4 months to get there. More than half of them first went through a stage of getting clicks without selling.

The same study shows where the money went instead. In the median store, the top 1% of products took 27.7% of ad spend and the top 10% took 68.1%. A new product placed in the main campaign competes with exactly those products.

This is an observation, not an experiment. The cohort includes older products that were never shown, not only new ones. A new product starts in the same position on its first day. The article on why products get no impressions explains why so much of a typical catalogue never gets shown and which of those products deserve a way back.

Stores that doubled earned 47.3% of revenue from new products

Launches are worth the effort: in growing stores, new products carry a large part of revenue. In our analysis of 100 stores from June 2025 to June 2026, we compared the first three months of the window with the last three. Then we measured what share of revenue came from products that were not there at the start.

Store trajectory over the window	Stores	Share of revenue from new products
Doubled	25	47.3%
Grew 1.2–2×	16	51.4%
Stayed flat	27	31.6%
Declined	32	26.1%

Three checks make the link worth acting on, though it is still an observation:

It is not the budget. The link held among stores that cut spend, kept it flat and raised it.
New products added revenue rather than replacing it. In the stores that doubled, revenue from old products grew too, to 126% of the starting level. Whether launches take sales from your bestsellers is the question of new product cannibalisation.
Volume alone is not the answer. The number of new products a store tested was not what set growing stores apart. Testing more, on its own, did not go together with growth.

One caveat about the word “new”. In this data, a new product is one that was not in the ads data in the first three months of the window. A product whose ID changed in the feed looks new too. To follow this number for your own store, track the share of revenue from new products quarter by quarter.

What must be ready before a new product can get its first impression?

Four conditions decide whether Google can show a new product at all. You can check each one without spending on ads.

Condition	Why it matters	What to check
Approved in Merchant Center	Google’s retail advice asks you to make sure listed products are approved	The product status before the launch date
A stable, unique ID	Google identifies the product by its ID, and the ID can’t be edited	The new product has its own ID, not a reused old one
Included in a listing group	Listing groups decide which products an asset group can show	The product falls into an included group
Marked as new	A label lets campaigns treat new products as one group	A custom label with the launch month

The ID rule is stricter than it looks. The Merchant API reference (REST Resource: accounts.productInputs) marks the product ID (offerId) as required and immutable: it is your unique identifier for the product. So if you send a product under a fresh ID, Google sees a different product, and its ad history starts from zero. For the same reason, give each new product an ID of its own, never an old product’s: Google would treat it as the old product.

Listing groups need the same attention. Google Ads Help on managing Performance Max with listing groups describes them as part of each asset group: they decide which products it shows. When a campaign isn’t getting impressions, Google names listing groups or the feed as possible causes.

The Google Ads API documentation (Listing Groups for Retail) adds that every subdivision must contain an “Other” node. So a new product that matches none of your named groups lands in “Other”. If an asset group excludes “Other”, it never shows the product.

The full pre-launch list, with feed fields and campaign settings, is in the new product launch checklist.

Main campaign, asset group or a separate campaign?

In Performance Max you have three places to put new products. For a launch, they differ in one thing above all: whether new products compete with bestsellers for the same budget.

Google’s own guidance points both ways. Its retail advice (Retailer best practices for AI-powered Performance Max campaigns) asks you to consolidate campaigns where you can, because Google AI works best with a unified budget. The same page names different budgets or ROAS targets for new products as a case where a separate campaign makes sense.

The budget belongs to the campaign: a Performance Max campaign can’t share its budget with other campaigns (Create a Performance Max Campaign Budget). An asset group inside the main campaign can give new products their own images and texts, but not their own money.

The target matters as much as the budget. According to Google Ads Help (About Target ROAS bidding), a target ROAS set too high may limit the traffic your ads get. Lowering it lets the strategy enter more auctions. For products with no history, that is the lever that buys first impressions.

A separate campaign has a cost: its bid strategy starts with a learning period. According to Google Ads Help (Duration of the learning period for campaigns and what affects it), calibration can take a few conversion cycles, typically 1–2. It can be faster with more conversion data, including data from earlier campaigns.

The same page lists a new strategy and a changed setting, such as a new target, as reasons for the “Learning” status. Judge the products only once that status is gone.

	Main campaign, as is	Own asset group in the main campaign	Own campaign or tier
Budget	Shared with bestsellers	Shared with bestsellers	Its own
Target ROAS	The campaign’s	The campaign’s	Its own, usually lower
Bid strategy	Unchanged	Unchanged	New, needs time to calibrate
Control over first impressions	None	Own creative and signals, no budget of its own	A set share of the budget
When it fits	A small catalogue with rare launches	New products need their own assets or theme	Regular launches you want tested

This is how the portal resolves the choice. In its recommended campaign structure, products with no history get a tier of their own, the newcomers tier. The portal sets that tier’s target at 0.9 × the account’s ROAS, the lowest of all tiers. The tiers for proven sellers get 0.95–1.0 ×, and all other tiers 1.2–1.3 ×.

In the full structure, the newcomers tier gets a fixed 10% of the budget. If the account has too few conversions for several campaigns, the portal recommends one campaign, two at most, and leaves the newcomers tier for later.

The guide to a separate campaign for new products explains when it pays off and how to move products out of it.

Example store, not client data.

A tableware shop with 3,000 products spends 40,000 a month at a ROAS of 4.5, or 450%, and pays 5 per click. Only 1,200 of its products got an impression last month. Under the portal’s rule, its newcomers tier gets 10% of the budget, 4,000 a month, at a target of 0.9 × 450%, roughly 400%. At 5 per click, 4,000 buys 800 clicks: a full 20-click test for 40 new products a month.

How many clicks and how much money does a fair test need?

Count a test in clicks, not in days. In our study of 1.4 million products, the minimum honest test was 15–20 clicks, roughly $3–4 at the sample’s average cost per click of $0.20. Below that, the result is noise. The threshold depends on the category:

Category	Minimum test, clicks	At $0.20 per click
Bags and luggage	15	$3
Apparel and accessories	20	$4
Home and garden	20	$4
Electronics	20	$4
Health and beauty	30	$6
Furniture	30	$6
Car parts	30	$6
Hobbies and crafts	50	$10

Your cost per click will differ, so plan the test budget as a formula: products × clicks per test × your average cost per click. Take the example tableware store, which spends 40,000 a month at 5 per click. Then 60 new products × 20 clicks × 5 = 6,000, or 15% of the monthly budget. With 4,000 a month in its newcomers tier, the shop either tests 40 products this month and 20 the next, or gives the tier more for the launch month. The full method is in testing products with Google Ads.

While the clicks add up, watch two things:

Learning comes first. If you created a new campaign or changed its target, the first conversion cycles show the strategy calibrating, not the products.
The June 2026 reporting change. From 15 June 2026, product-level reports for Performance Max include data from all of its networks. Google warns of a possible one-time increase in reported metrics (Retail Campaign Performance). Read a test that spans that date with care.

Keep, retest, demote or drop: how to read the result

A test ends when the product has had its clicks, not when the month is over. Then sort each product into one of five outcomes:

What the test showed	Decision	What to do
Fewer clicks than the category threshold	Not tested yet	Check inclusion and the feed; there is no verdict yet
Sales at or above the target	Keep	Move it to the main campaign by changing its custom label
Sales below the target	Keep on watch	Leave it in the test; check its price and listing
Threshold reached, no sale	Retest or demote	Lower its share of the budget; don’t switch it off yet
50+ clicks and no conversion in 360 days	Drop	Exclude it from campaigns (the portal’s rule)

Why not drop a product after its first failed test? In our study of 1.4 million products, a hard switch-off rule did not hold. The accuracy of “no sales after this much spend, so switch it off” fell from 88.1% at $5+ of spend to 66.2% at $500+.

In the same study, 35.4% of products in its “loser” status converted later, after a median of 2 months. These products had already spent the price of a conversion without getting one. That is why the portal judges on a long window: it excludes a product only after 50 or more clicks with no conversion in 360 days.

Move products between campaigns by changing their custom label, never their ID. The label moves the product into another listing group; the unchanged ID keeps it the same product for Google.

Once a new product has a history, the portal moves it to one of its eight product labels based on how it behaves in ads. A new product that sells and holds the target usually lands in Strong: profitable, but without a long history yet. If it sells only a little at target, it goes to Long tail. One that spends without converting lands in Dormant; after a noticeable spend with no sales, it moves to To remove.

The guide to handling a failed product test shows, step by step, what to do with a product that finished its test without a sale.

A product that passes the test enters a new stage, and its results keep changing after that. The guide to product performance in Google Ads covers how long a bestseller lasts and what to do with the long tail.

How many new products sell, and how many to add each month

Expect most launches to fail. In GetProfit data from June 2025 to June 2026, a new product counted as a hit if it sold at least once in its first 3 months:

Where the store added the new product	New products	Sold in the first 3 months
A category where the store already had sales	696,921	4.0%
A category with no earlier sales	63,921	3.5%

Proven categories did better in 51 of 83 stores, so in the other 39% it was the other way round. Read the gap as a guide, not a law. This share is the new product hit rate, and it depends on the window: don’t compare a 3-month rate with a yearly one.

Because the hit rate is low, set the number of launches from your budget, not from the supplier’s price list. Products you can test a month = test budget ÷ (clicks per test × cost per click). For the example tableware store, with a 4,000 monthly test budget, 20 clicks per test and 5 per click, that is 40. Adding more than the budget can test only lengthens the queue of products that wait for their first impression.

The article on how many new products sell shows what a 4% hit rate means for your test budget. To size the monthly number for your store, see how many products to add monthly.

The launch cycle, step by step

A launch is a cycle that repeats every month: find, launch, test, select, remove.

Find candidates. Start with categories where you already sell; the hit rate there was slightly higher in our data. Check: the category has had sales in the last 12 months.
Prepare the feed. Give each new product its own stable ID and complete product data. Check: the product is approved in Merchant Center before the campaign needs it.
Mark it as new. Put the launch month in a custom label, for example new-2026-10. You can create up to five custom labels in your product data. Google recommends grouping products with them and then targeting those labels in listing groups (Google Ads Help, Manage a Performance Max campaign with listing groups). Check: every product in the batch carries the label.
Include it. Point the campaign or asset group for new products at that label, and decide whether the main campaign excludes it. Check: the product sits in an included listing group, not in an excluded “Other”.
Fund the test in clicks. Set the budget from products × clicks × cost per click, with a target below the account’s ROAS. Check: the budget lasts until every product in the batch has its clicks.
Wait out the learning period. Don’t judge a new campaign or a new target in its first conversion cycles. Check: the bid strategy status no longer shows “Learning”.
Select once a month. Sort the tested products into five outcomes: not tested yet, keep, keep on watch, retest or demote, drop. Check: every verdict rests on enough clicks for the category.
Move, don’t rebuild. Move products that pass to the main campaign by changing their label. Lower the budget share of products with no sales, and exclude a product only after a long run without conversions. Check: no product changed its ID on the way.

If you manage the account for someone else, agree every change with the owner before you apply it.

How the portal treats new products

The portal’s Products section gives every product a label based on how it behaves in ads. Products with no ad history, including those not yet advertised, get the Newcomers label. The advice next to them matches this guide: give them first impressions, add them to a test campaign and watch the response.

In the recommended structure, these products get their own tier with a lower target ROAS and a fixed share of the budget. Their first impressions then come from money the bestsellers can’t take.

Which products bring money in and which spend it. Every product gets its own label based on how it behaves in ads — from bestsellers to dormant ones that spend budget without a single sale. The portal changes nothing without your consent.

Check your catalogue →

Frequently asked questions

Does a product become new again if I change its ID?

For Google, yes, and that is the problem. The ID is how Google identifies the product, and you can’t edit it. So Google treats a product under a new ID as a new product with no ad history. To move a product between campaigns, change its custom label instead.

How long should a new product test run?

Until it has its clicks: 15–20 in four of the eight categories in our study, 30 in three and 50 in hobbies and crafts. The calendar time depends on your budget and cost per click. Patience pays within limits: products that started the study with no spend and later sold needed a median of 4 months to the first sale.

Should each new product get a campaign of its own?

No. Group new products under one custom label and give the group a campaign or a tier. According to Google Ads Help, an asset group can have up to 1,000 listing groups. It also warns that a large number can hurt performance and recommends grouping products with custom labels.

Does the same approach work in Standard Shopping?

In outline, yes. According to the Merchant API reference (ProductAttributes), custom labels group products in Shopping campaigns too, so a “new” label can feed a separate Shopping campaign in the same way.

Sources

About Performance Max campaigns — Google Ads Help: Performance Max moves budget towards the best-performing inventory and placements. Checked 2 October 2026.
Manage a Performance Max campaign with listing groups — Google Ads Help: listing groups decide which products an asset group serves; bids are set by Google AI from the campaign objective; listing groups or the feed as a cause of missing impressions; up to 1,000 listing groups per asset group; up to five custom labels, recommended for grouping items. Checked 2 October 2026.
Listing Groups for Retail — Google Ads API: every subdivision must contain an “Other” node. Checked 2 October 2026.
REST Resource: accounts.productInputs — Merchant API: the product ID (offerId) is required, immutable and your unique identifier for the product. Checked 2 October 2026.
ProductAttributes — Merchant API: custom labels 0–4 are for custom grouping of items in a Shopping campaign. Checked 2 October 2026.
Retailer best practices for AI-powered Performance Max campaigns — Google Ads Help: make sure listed products are approved; consolidate campaigns where you can; separate campaigns for different budgets or ROAS targets for new products. Checked 2 October 2026.
Create a Performance Max Campaign Budget — Google Ads API: the budget of a Performance Max campaign can’t be shared. Checked 2 October 2026.
About Target ROAS bidding — Google Ads Help: a target set too high may limit traffic; lowering it lets the strategy enter more auctions. Checked 1 October 2026.
Duration of the learning period for campaigns and what affects it — Google Ads Help: calibration can take a few conversion cycles, typically 1–2; earlier conversion data can speed it up; reasons for the “Learning” status. Checked 2 October 2026.
Retail Campaign Performance — Google Ads API: from 15 June 2026 product-level reports include all Performance Max networks, with a possible one-time increase in metrics. Checked 2 October 2026.
GetProfit study of 1,404,808 products, 130+ stores, 13 months — products that started with zero spend; concentration of spend; minimum test by category; accuracy of switching products off; later conversions.
GetProfit data: 100 stores, June 2025 – June 2026 — share of revenue from new products by store trajectory; revenue from old products in stores that doubled; testing volume and growth.
GetProfit data: June 2025 – June 2026 — share of new products that sold in their first 3 months, in categories with and without earlier sales; comparison across 83 stores.
GetProfit portal methodology — the Newcomers label; the newcomers tier with a target of 0.9 × account ROAS and 10% of the budget; exclusion after 50+ clicks with no conversion in 360 days.