When Standard Shopping is the right container
The Shopping and Performance Max hub sets out what Standard Shopping controls: product groups within ad groups, bids per group, a campaign priority of low, medium or high, negative keywords, device and location, with manual CPC, Maximise clicks or Smart Bidding as the bidding options and a full search terms report. Its ads appear on the Shopping tab, in Search results and on search partners. Performance Max reaches far more inventory but hands most of those decisions to the system.
Standard Shopping is therefore the container when the business needs to see or choose the queries it pays for: products that attract irrelevant searches, a brand that must not pay for its own name, a range where generic and specific searches are worth very different amounts, or a feed and tracking set-up that is not yet ready for automation. The comparison page's decision table covers those cases. It is also the cleaner read when the question is how product ads perform on Shopping alone. A common mature pattern runs both: Performance Max for most of the catalogue and Standard Shopping for the product sets where query control pays, which the section on coexistence below covers.
Product groups and how to subdivide them
A Standard Shopping campaign starts with one product group containing every product in the feed that the
campaign's settings allow. Structure is built by subdividing it on feed attributes, and the
product data specification defines the attributes that
subdivisions use: product_type for the store's own category path, brand, the five custom labelsFree-text product feed attributes a store fills in to group its products for campaign structure, bidding and reporting, such as margin tier or season. from
custom_label_0 to custom_label_4, the item's id, and Google's own category. The
product feeds guide explains why product_type and the custom labels
are the better structural fields: they belong to the store, carry no policy weight and change only when the store
changes them. Titles decide matching rather than structure, and the
shopping feed title test is a design for retitling a random half of one
category while both halves keep the same product groups and bids.
| Subdivide by | When it fits | Watch out for |
|---|---|---|
| Product type | The catalogue's own categories map to different economics or demand | Category paths that the platform rewrites when the site changes |
| Brand | Brands sold carry different margins or deserve different bids | Brand values that vary in spelling across suppliers |
| Custom label | Margin band, price tier, best-seller status, season or stock depth | A label whose meaning nobody wrote down drifts and breaks the structure |
| Item ID | Individual best-sellers or problem products need their own bid | Item-level groups everywhere multiply bids and thin the data |
| Everything else | Always, at the bottom of each subdivision | Leaving it out means new products land in no group at all |
Subdivide only where bids should differ. A product group exists to carry a bid, so two sets of products that would receive the same bid gain nothing from being separated except more rows to manage. The usual shape is a first split on the attribute that best describes the economics, often a margin-band custom label, and a second split on item ID for the handful of products that sell enough to deserve their own bid. Keep the everything-else group: the Shopping products not showing diagnosis traces many missing products to subdivisions that left new product types or labels in no group.
Ad groups sit above product groups and are mainly a reporting and routing unit. One ad group per broad category is enough for most stores, with more only where a set of products needs its own negative keywords.
Bids per product group
On manual bidding, each product group carries a maximum cost per click. The bid should come from what a click on those products is worth, which comes from margin. The break-even ROAS entry defines Break-even ROASMinimum ROAS to avoid a loss. as one divided by gross margin: the return at which the margin on the sales just covers the advertising. Working backwards gives the maximum a click can cost before it loses money: the average order value for the group, multiplied by its conversion rate, divided by the break-even return. A target return adds the profit the business requires on top, and the ROAS calculator makes the arithmetic explicit.
Three practical rules keep manual bids honest. Use each group's own order value and conversion rate where it has enough history, and the category's figures where it does not, so a new product is not bid on the store average. Review bids on a fixed rhythm, moving them in modest steps, rather than reacting to a few days of data. And read ROASRevenue generated per unit of ad spend. per product group against that group's break-even, not for the account, because a blended figure hides which groups pay. A group that returns below its break-even after a fair trial gets a lower bid or its own investigation; a group well above it with lost impression share is the place to bid more.
Campaign priority and routing queries
The hub lists campaign priority, with the values low, medium and high, as the setting that governs precedence between Standard Shopping campaigns that include the same products. That single documented setting, combined with negative keywords, supports a structure many retailers use and that is industry practice rather than something Google prescribes: routing different kinds of search for the same products to campaigns with different bids.
The reasoning is that a generic search ("road bike") and a specific one naming a brand and model are worth different amounts even when they show the same product. The specific searcher is further along and converts more often. Without routing, both land in one product group with one bid. The routing practice builds two campaigns over the same products:
| Campaign | Priority | Negatives | Bids | Intended traffic |
|---|---|---|---|---|
| Generic | Higher | Brand and model terms, and other specific words, added as negatives | Lower | Broad, early searches |
| Specific | Lower | Only the account's usual negatives | Higher | Searches naming a brand or model |
The higher-priority campaign takes the products first, and its negative keywords remove the specific searches from it, so those searches are left for the lower-priority campaign with the higher bids. Some advertisers add a third campaign for brand searches about the store itself. This guide makes no further claim about how priority interacts with bids and budgets; the current help pages are the reference for that detail. What can be checked directly is whether the routing works: read each campaign's search terms report and confirm the generic campaign holds the generic queries and the specific campaign holds the named ones. If named queries appear in the generic campaign, the negatives are incomplete.
Negative keywords in Shopping
A product campaign has no keywords, so negative keywords are the only query-level control it has. The negative keywords guide, which follows Google's documentation, applies in full: negatives can live in shared lists applied across campaigns, at campaign level and at ad group level; negative match types are stricter than positive ones; and negatives do not match close variants, so plurals and misspellings must be listed. In a routing structure, the campaign-level negatives are what route the traffic, so they are maintained with more care than ordinary exclusions.
The search terms report is where the list grows. Review it on a fixed rhythm, weekly for the campaigns that carry most of the spend, and sort queries into three groups: relevant and converting, relevant but in the wrong campaign, and irrelevant. The second group feeds the routing negatives; the third goes into a shared list applied to every Shopping campaign. A product that keeps appearing for searches it cannot satisfy usually needs a better feed title, not only a negative, because the title is what the matching reads.
Running beside Performance Max
The hub states the rule that shapes every mixed account: when a Performance Max campaign and a Standard Shopping campaign in the same account are eligible for the same product, the Performance Max campaign is served. Campaign priority orders Standard Shopping campaigns among themselves; it does not override that rule.
The structure follows. Decide which products Standard Shopping owns, mark them with a custom label in the feed, and exclude that label from every Performance Max listing group, as the Performance Max structure guide describes. Then include the same label in the Standard Shopping campaigns. Each product sits in exactly one campaign type. Where overlap is deliberate, accept that Standard Shopping will mostly serve what Performance Max does not want, and do not read its results as a fair test of anything. The shopping and catalogue ads guide places this split within a wider retail acquisition plan. The academy capstone on a Google Ads operating plan for a hot tub retailer practises deciding which products Standard Shopping owns beside Performance Max as one part of a year's plan.
Moving to automated bidding
Manual bids are a starting point, not a destination. Google documents Smart Bidding as auction-time bidding that uses signals such as device, location, time, the query and audience membership to predict the likelihood and value of a conversion, signals no manual bid can use. The Smart Bidding guide sets out what it needs: complete conversion tracking, the right conversion actions in the campaign's goals, and, for value strategies, values that reflect what the business earns.
The move is sensible when purchases are tracked with order values and the campaign records conversions steadily. Choose maximise conversion value with a target return, start the target near what the campaign has returned recently, and move it towards the margin-based target in steps. Google documents a learning period after a strategy is created or significantly changed, so change one thing at a time and judge the move at account level over weeks.
Automated bidding changes what routing means. The routing practice above depends on the two campaigns having different manual bids. Under value-based bidding each campaign bids to its own target, so the split becomes two targets: a stricter return for the generic campaign, a looser one where specific searches convert well. Some advertisers keep the routing for its reporting value and let the targets do the work; others merge the campaigns once the model has data, because one campaign with more conversions often predicts better than two with fewer.
A priority split for generic and model searches (illustrative)
Sprocketlark Cycles routes generic and model searches
Sprocketlark Cycles, an invented bike shop, sells road, gravel and hybrid bikes from four brands, with an average order value of £1,150 and a gross margin of 28% on bikes, so its break-even return is about 357%. One Standard Shopping campaign on manual bids shows every bike in one product group per category. The search terms report shows that searches naming a brand and model convert at 3.1%, generic searches at 0.9%, and both pay the same bid.
| Product group | Generic campaign (high priority) | Model campaign (low priority) |
|---|---|---|
| Road bikes | £0.55 | £1.30 |
| Gravel bikes | £0.50 | £1.20 |
| Hybrid bikes | £0.35 | £0.80 |
| Two best-selling road bikes (item ID) | £0.65 | £1.60 |
| Everything else | £0.30 | £0.60 |
The generic campaign carries every brand name and model name as negative phrase match, plus "size", "review" and "spare". After six weeks:
| Query type | Clicks | Conversion rate | Spend | Revenue | Return |
|---|---|---|---|---|---|
| Generic, before the split | 4,800 | 0.9% | £4,100 | £49,700 | 1,212% |
| Generic, after | 4,200 | 1.0% | £2,050 | £48,300 | 2,356% |
| Model-name, before the split | 1,100 | 3.1% | £940 | £39,200 | 4,170% |
| Model-name, after | 1,650 | 3.0% | £2,010 | £56,900 | 2,831% |
Both query types now return well above break-even, and total revenue rose from £88,900 to £105,200 for £4,060 of spend against £5,040 before. The search terms report shows a few model searches still reaching the generic campaign through spellings the negatives missed; the shop adds them weekly.
Common mistakes
- No everything-else group. New products and relabelled items fall into no group and never serve.
- Item-level groups across the whole catalogue. Hundreds of bids, none with enough data to set well.
- Bids set from habit rather than margin. A product group's bid should trace back to its break-even return.
- Routing by priority without the negatives. Priority alone does not separate generic and specific searches; the negatives in the higher-priority campaign do.
- Never checking where queries landed. The search terms report of each campaign is the only proof the routing works.
- Overlapping Performance Max by accident. Performance Max is served for the same products, so Standard Shopping quietly loses the products it was meant to control.
- Fixing every bad query with a negative. A product that keeps matching the wrong searches usually needs a better feed title.
- Switching to Smart Bidding and restructuring in the same week. Two changes, one unreadable result.
