Why one target cannot serve every product
A target applies per campaign, and a campaign with one target bids for every product in it against the same expectation of return. The Shopping and Performance Max hub lists the result among its common problems: best-sellers starve the rest of the catalogue when there is one campaign, one target and uneven product performance. The products with the strongest history attract the spend, because the system can predict them, and everything else waits.
Lifecycle makes the unevenness predictable. A new product has no conversion history, so it is hard for any model to favour. A best-seller has plenty and will take spend whether or not it needs help. A seasonal range is worth a lot for a few weeks and little the rest of the year. Clearance stock has a different margin from the same product a month earlier. And a product that is out of stockA period when a product is out of stock and cannot be sold, which caps sales and should stop advertising for that product until stock returns. or discontinued should not be advertised at all. The Shopping products not showing diagnosis adds a related case: low-price items rarely earn back a click at the same ROASRevenue generated per unit of ad spend. as expensive ones and need their own target.
Google does not document how Performance Max allocates spend between the products in a campaign, and this guide makes no claim about it. What can be read is the outcome: the structure guide describes listing group and product reports that show performance by product, including which products received no impressions. The approach here is to make the structure match the economics and then read those reports.
Labelling products by lifecycle stage
The feed guide describes custom labels as five free-text attributes that Google does not use for matching or display, provided so the advertiser can tag products with facts the campaign needs; launch date, best-seller status, seasonality, stock depth and clearance are among its examples. One label slot can carry the lifecycle stage. The guide's rules apply in full: write down what the label means, populate it from data rather than by hand, and keep the values few and stable. The product life cycle framework sets out what each stage asks of marketing beyond the campaign, from explaining a new product to recovering value at the end.
| Stage | Rule that assigns it | Data source | Campaign treatment |
|---|---|---|---|
| New | First listed within the business's launch window, or no sales history yet | Catalogue launch date | Own listing group, or own campaign where the launch needs protected budget |
| Core | Sells steadily at or above an agreed rate | Order system | Main campaign at the margin-based target |
| Long tail | Listed beyond the launch window, sells slowly | Order system | Its own listing group, so its results are visible; target suited to its economics |
| Seasonal | Belongs to a range sold in a defined period | Merchandising calendar | Switched on and off with the season |
| Clearance | Marked for clearance or end of line | Stock and merchandising decisions | Own group with a target from the clearance margin |
The thresholds behind each rule, such as how long a product counts as new or what rate makes it core, come from the
business's own sales history, not from a general rule. Once set, refresh the label on a schedule. The
Google Ads setup for online sales warns that a margin label set once and never refreshed
drifts as costs and promotions change, and a stage label drifts faster, because products sell, age and go out of
season every week. A daily or weekly job that rewrites the label from the source data, delivered through the primary
feed, a feed rule or a supplemental feed keyed on id, keeps the campaigns following the catalogue.
New products without history
A new product starts with nothing for the system to learn from: no clicks, no conversions, often a thin product page. Placed in a campaign alongside proven products under one target, it competes with items the model can predict, and it may receive few impressions while the season it was launched for passes.
Give new products a listing group of their own within the main campaign, so their impressions and results can be seen apart. Where a launch needs budget that the rest of the catalogue cannot absorb, the structure guide's table of reasons to split names a feed segment with its own budget rule, launches among them, as a reason for a separate campaign. Weigh that against the same guide's warning that a segment with too few conversions to learn on its own is better kept in the larger campaign; a short launch campaign that is merged back once the products have history is a common compromise.
Three things help a new product more than structure does. Complete feed data from the first day: a title written
for search, identifiers, the product type and a good image, as the feed guide sets out. Search themes on the asset
group for a new line the landing page undersells, which the structure guide recommends for new lines without
history. And for products not yet shipping, the preorder availability value with an availability_date, which
the specification requires for preorder itemsAn order and payment taken before the stock is ready, with terms that state what the buyer commits to, when dispatch happens and what if the date slips., so that the ad does not promise immediate delivery. Then set a
graduation rule, so a launch moves into the core or long-tail stage on a date or a sales level the business chose,
rather than staying "new" indefinitely. The store's side of the same launch, from the product page and pre-order
terms to the order in which the waitlist, list and ads hear about it, is set out in the
tutorial on launching a new product in an online store.
The channel plan for paid search during a product launch places this Shopping
treatment beside the product-name campaigns and category queries that run in the same window.
Best-sellers and the rest of the catalogue
Best-sellers look after themselves. They have the history, the reviews and the conversion rate, and a campaign with one target will spend most on them. That is acceptable while they still have room to grow, and a waste when the spend simply buys sales they would have made anyway, while the rest of the catalogue never appears.
Separating them is a structure decision. A listing group, or a campaign where the economics differ enough, for the core products, and another for the long tail, each with a target that fits its margin and history. The product report then shows whether the long tail is receiving impressions at all; the structure guide's list of common mistakes ends with a catalogue where a large share of products never receive an impression and nobody has asked why. Two cautions apply. A long tail split into many small groups starves each of data, so keep it as one or two groups. And low-priced items within it may need a target of their own, as the products not showing diagnosis explains, because their order value cannot carry the same return. If a best-seller's paid impressions fall after a change to its label, group or target, the diagnosis for a best-selling product's sales falling separates moved paid support from a limited listing, lost rankings and the store's other causes.
Seasonal ranges
A seasonal range is a lifecycle compressed into weeks: launch, peak and clearance in one season, then nothing until next year. The feed guide's example label scheme includes a seasonality slot with values by season, used for seasonal campaigns that are switched on and off. The work is in the timing.
Change the label before demand begins, not on the first day of the season, so the products are approved, grouped
and serving when people start searching. Raise the range's budget as demand rises and lower it as the season fades,
reading the range's results against the same weeks a year earlier, because SeasonalityPredictable, repeating changes in demand tied to the calendar, such as months, holidays, weekdays or hours. makes the
month-on-month comparison misleading. At the end of the season, move the unsold remainder into the clearance stage
rather than leaving it in the seasonal group with a peak-season target. For a short promotional period, the online
sales setup notes that the help centre documents seasonality adjustments for bidding, which are safer than raising
targets by hand and forgetting to lower them. Price changes for a sale go through the sale_price attribute with
sale_price_effective_date, not by overwriting price, and the
sale promotion readiness checklist checks that the sale price agrees across
site, feed and ads before the day.
Clearance and end-of-line stock
Clearance changes the margin, so it changes the target. The break-even ROAS entry defines Break-even ROASMinimum ROAS to avoid a loss. as one divided by gross margin, and the important decision for clearance is what counts as the cost. If the stock was bought at a price the clearance price no longer covers, the margin on original cost is thin or negative and the break-even return is very high. If the honest alternative is selling the stock to a liquidator or writing it off, many businesses treat that recovery value, plus fulfilment, as the relevant cost, which raises the clearance margin and lowers the break-even. Either way, write the assumption down next to the target, as the break-even entry advises for any judgement of that kind, and use the break-even calculator and the ROAS calculator to make the arithmetic explicit.
Put clearance in its own listing group or campaign so its target cannot drag down, or be dragged up by, the core range; the feed guide describes a clearance group with a looser target as the textbook use of labels. Add stock depth to the picture: an item with a handful of units left is not worth a fresh push, and the feed guide's example scheme uses a stock-status label to avoid over-promoting thin stock. When an item sells out, its availability changes, and the next section takes over. The decisions around the ads, from which route the stock should take to keeping reference prices honest, are covered in the guide to clearing excess stock.
Out of stock and discontinued items
The feed guide is explicit that availability must match the landing page and that a mismatch is treated like a
price mismatch. Keeping availability accurate is what stops out-of-stock items serving, and it means updating the
feed as often as stock changes. The
Merchant Center products not updating diagnosis
traces stale availability to failed fetches, slow schedules, quiet integrations and overriding supplemental feeds.
For items that will return, backorder with an availability_date is the specification's value for products that
can be ordered now for later delivery. Shoppers who asked to be told when such an item returns are reached by
email rather than through the feed, and the
tutorial on back-in-stock and price-drop alerts
keeps that alert queue in step with the same stock status.
Discontinued products should leave the feed. That diagnosis also explains that items not refreshed within the window the help centre documents expire and stop serving, which is a reason not to rely on expiry as a way of removing them: an expired item left in a partial feed looks like a fault, while a removed one is a decision. When a discontinued product is removed, check its listing group and any item-level product group, so the structure does not keep a branch for something that no longer exists. The Shopping and Performance Max playbook agrees who decides each product's stage and who writes it into the feed, so labels keep pace with the merchandiser's calendar.
A kitchenware shop's label scheme (illustrative)
Ladlecrest Kitchenware labels its catalogue by stage
Ladlecrest Kitchenware, an invented online kitchenware shop with 1,400 products, runs one Performance Max campaign on a target return of 450%. The product report shows 610 products with no impressions in the last 90 days, most of them launched in the last two months, while 120 best-sellers take most of the spend. Clearance items sit in the same campaign at the same target and barely move.
The shop sets custom_label_2 to new, core, long tail, seasonal or clearance, rewritten nightly from the catalogue
launch date, 90-day sales and the merchandising calendar. New means listed within the last 60 days; core means at
least eight orders in 90 days. Clearance margin is calculated on the price a liquidator would pay plus fulfilment.
| Stage | Listing group | Return target | Share of spend before | Share of spend after |
|---|---|---|---|---|
| Core | Main campaign, label = core | 450% | 78% | 58% |
| Long tail | Main campaign, label = long tail | 450% (shared) | 9% | 14% |
| New | Launch campaign, label = new | 300% | 6% | 13% |
| Seasonal (barbecue range) | Seasonal campaign, on from April to August | 380% | 4% | 9% |
| Clearance | Clearance campaign, label = clearance | 220% | 3% | 6% |
After two months, products with no impressions fall from 610 to 240, new products record their first conversions, and new products leave the launch campaign once they reach 60 days. Total spend is unchanged at £18,000 a month; revenue rises from £79,000 to £84,500, and clearance stock falls by 40% in units, recovering more than the liquidator offer would have paid.
For the clearance group's own target, the worked break-even method for clearance stock moves the margin to recovery value plus fulfilment and reads the lower break-even that results.
Common mistakes
- One target for the whole catalogue. Best-sellers take the spend, launches and the long tail wait.
- Stage labels set by hand once. Within weeks the label describes a catalogue that no longer exists.
- New products in a separate campaign forever. A launch group with no graduation rule becomes a second, starved catalogue.
- Seasonal labels changed on the first day of the season. The range is still being approved when demand peaks.
- Clearance at the normal target. The stock cannot reach a return set for full-price margin, so it never moves.
- Clearance at a target nobody can explain. A looser return is fine when the cost assumption behind it is written down; a guess is not.
- Sale prices written into
price. The strike-through is lost and the price is wrong when the sale ends. - Letting discontinued items expire instead of removing them. The feed looks broken, and the listing groups keep branches for products that are gone.
