Recently, a large retailer brought us a launch problem with nothing visibly wrong. The new collection was ready, the stock had landed and the Performance Max account was healthy. The plan gave the collection a four-week commercial window. What the account did not have was any reason to reserve visibility for the new range.
The established products entered that window with months of evidence behind them. The new collection entered with a feed listing and little or no performance history. The business needed an answer inside four weeks, while the account could keep backing products it already knew.
This is often called the PMax cold start problem. For retailers, it is more useful to treat it as an allocation problem. A new product needs traffic to build evidence, but it competes for that traffic against products that already have evidence. The goal is to give the launch a fair, measurable test without assuming that every new product will succeed.
Why New Products Start at an Allocation Disadvantage
Performance Max uses Smart Bidding to predict conversion probability and value in each auction. History matters, but it is one of many inputs. Google also uses query-level performance and contextual signals such as device, location, time, audience and search context.
What the system does not receive automatically is the merchandising decision behind a launch. It does not know that a range has a short full-price window, that the buying team has made a large inventory commitment or that a category needs meaningful exposure before the next commercial review.
An established bestseller has already produced clicks, conversions and value. A new product has not. When both sit inside the same campaign with the same target and no launch structure, there is no guarantee that the new product will receive enough exposure to show whether demand exists.
This can become an evidence loop. The product needs visibility to collect data, but the account has little data to justify that visibility. The same loop can leave parts of a large catalogue as zombie products.
The risk is not that every new product receives zero impressions. The launch may simply receive too little opportunity during the period when the answer matters.
Prepare Merchant Center Before Launch Day
Priority cannot rescue a product that is not eligible to advertise. Feed readiness comes first.
Google's product launch guidance recommends submitting new product data three to five business days before launch while allowing time for review and processing. It also highlights the attributes that determine whether a product can match relevant demand: correct identifiers, useful titles, a detailed product type, accurate price and availability, and a working landing page.
Before building launch logic, check that:
- the products are approved and eligible in Merchant Center
- price and availability match the landing page
- GTINs and other identifiers are correct where required
- product titles and product types describe the items clearly
- the intended products are included in the PMax listing groups
- the launch cohort can be identified through a structured field
Merchant Center's condition = new is not a launch marker. It means the item is unused rather than used or refurbished. PMax's new customer acquisition goal is also unrelated because it prioritises new customers, not new products.
Custom labels need lead time too. Google notes that a new or changed label can take 24 to 48 hours to become available in Google Ads product groups. Build and verify the structure before the commercial window opens.
Choose the Right PMax Structure
A custom label gives Google Ads a usable product group. The next decision is what the account should do with that group.
| Structure | What it gives the launch | Main trade-off |
|---|---|---|
| Listing group in the existing PMax campaign | Clear inclusion, exclusion and reporting for the launch cohort | Products still share the campaign's budget and ROAS target |
| Dedicated asset group | Launch-specific creative assets tied to the relevant products | No separate asset-group budget or ROAS target |
| Separate PMax campaign | A dedicated budget, target and launch window | More campaign complexity and a separate ramp-up to manage |
Google's retail guidance for Performance Max recommends custom labels for high-priority products. It also says a separate campaign can make sense for new products when they need a different budget or ROAS target. Otherwise, Google recommends consolidating campaigns so its models have more data to work with.
A steady flow of ordinary arrivals may only need a maintained listing group inside the main campaign. A major collection with a fixed investment and a four-week deadline may justify a campaign of its own. An asset group is useful when the products need different creative, but it does not reserve budget.
Build the Launch Rule with Guardrails First
The fragile part of any launch structure is keeping the cohort current. Products arrive, others age out, stock changes and some items lose their sellable size range before the campaign ends. A static SKU list starts going stale as soon as it is created.
Expanly tracks when a product first appears in the connected catalogue. Its is_new signal remains true for the first 30 days from that observation. A Business Rule can combine newness with availability, stock, margin and other product data. Rules run once a day, from top to bottom, and the first matching rule assigns the product's High, Standard or Low segment.
A continuous new-arrivals model could use this order:
| Order | Rule | Segment |
|---|---|---|
| 1 | Unavailable products or critically weak stock | Low |
| 2 | New products that have passed an agreed test threshold without a conversion | Standard or Low |
| 3 | New products that are available and above the stock floor | High |
| 4 | Normal margin, inventory and performance rules | High, Standard or Low |
Guardrails come first because the first match wins. A product should leave the launch segment after it sells out, even while it is still new. Where variant data is available, the rule can also consider how much of the size or colour range remains in stock.
Stock cover sounds like a natural launch guardrail, but it requires a sales pace. A product with no purchases does not yet have a meaningful days-of-stock value. Use current availability or stock quantity at the beginning of the launch, then bring stock cover into the logic after sales begin. The same principle sits behind inventory-aware advertising: demand generation should follow what the retailer can still sell.
Preview the rule before publishing it. If an intended launch rule matches most of the catalogue, the cohort definition is wrong or the catalogue is too new for an automatic newness signal.
First Seen Is an Observation Date
Expanly's newness signal uses the date the platform first observed the product. It is not an imported launch date from the retailer's ERP.
That difference matters during onboarding. When a catalogue is connected for the first time, existing products can initially look new to Expanly too. An always-on is_new rule should therefore not be activated blindly during the first 30 days of a new connection.
For a planned collection launch, use a reliable identifier from the source data, such as a collection tag, custom label or product type. Give the rule an explicit start and end date. The rule enters the evaluation order on launch day and expires after the protected period. This is also the safer pattern when products must be submitted to Merchant Center before their commercial release.
For a wider treatment of collection changes, outgoing stock and overlapping seasons, see our seasonal Google Shopping strategy.
What the Rule Changes
After each daily evaluation, Expanly writes the result to a selected Merchant Center custom label through a supplemental feed. It can export either the three-tier segment, such as high_priority, or the stable ID of the rule that matched the product.
Segment mode works for a general High, Standard and Low campaign structure. Rule ID mode is useful when the launch rule needs its own listing group or campaign without mixing the products with every other High-priority item.
Expanly does not change Google Ads bids or budgets directly. It does not send stock, margin or other raw business data to Google. The label makes the current business decision available to the campaign structure, and that structure gives the decision consequences.
Adding a High label without changing how Google Ads uses it will not create a launch strategy. The label keeps the right products in the right group each day. It cannot reserve spend unless the account has been built to treat that group differently.
Give the Launch a Clear Exit
Launch support should be temporary. For continuous new arrivals, products stop matching is_new after the 30-day observation window. For a planned collection, the scheduled rule can end on the date chosen by the business.
An earlier stop can be part of the rule order. Once a product has accumulated enough spend to make a judgement, a higher rule can move it to Standard or Low if it still has no conversions. The threshold should reflect the retailer's economics and conversion lag. This is a classification rule evaluated on the next daily run, not a real-time product budget cap.
Avoid rebuilding the account every few days in response to early noise. A separate PMax launch campaign introduces its own learning period, and significant budget or target changes can cause the campaign to readjust. A stable campaign structure with product membership maintained by daily labels is easier to interpret than repeated manual restructures.
When the launch window ends, the product returns to the normal rule order. Its next segment depends on the retailer's baseline rules. A successful product is not guaranteed to remain High, and an unsuccessful one is not guaranteed to become Low unless the rules say so.
Measure Visibility Before You Judge Demand
A launch review should answer two questions in order:
- Did the products receive a meaningful test?
- What happened after they received it?
Google Ads' Products reporting provides product-level visibility into impressions, product clicks, cost, conversions and conversion value. Use it to separate an exposure problem from a demand problem.
Track:
- the share of eligible launch products that received impressions
- time from feed entry to the first impression and product click
- the launch cohort's share of impressions and spend relative to its share of the catalogue
- time to the first conversion, allowing for conversion lag
- CPA, ROAS and margin once the window is long enough to judge them
- stock, sell-through and variant availability during the launch
- spend still reaching unavailable or Low-priority products
Compare the cohort with a similar previous launch or a matched group of products where possible. A simple before-and-after chart can be distorted by seasonality, promotions, pricing and changes in demand.
A product can receive a fair test and still fail. That is useful information. A product that never received meaningful exposure has not produced the same evidence.
A Fair Test Is the Point
Every launch is a commercial decision the retailer has already made. The products were bought, priced, stocked and given a place in the calendar. The remaining job is to carry that decision into the ad structure with a defined cohort, a protected but bounded runway and guardrails that follow the product data.
A launch rule cannot make a weak product strong. It can make the result interpretable.
