Shopping Campaign Priority and Inventory Filters in 2026

Almost every retail account runs more than one campaign against the same feed: Performance Max on the proven sellers, Standard Shopping on everything else, often a third campaign for a promotion. Which raises a question the interface has no button for: when the same product sits in two campaigns, which one actually serves? The answer comes down to Shopping campaign priority and two further rules the interface never spells out.

The answer decides who gets your expensive branded traffic, where your bid is real and where automation quietly overrides it, and why half the budget ends up in the campaign you treated as secondary. This is the mechanics: Shopping campaign priority, inventory filters, the conflict-resolution rules, and three structures that hold up in 2026.

Shopping campaign priority: how Google picks when a product is in two

There are three rules, and conflating them is the root of most structural mess.

Situation	What decides	What it means in practice
Two Standard Shopping campaigns, same account	The higher campaign priority (low / medium / high). On a tie, the higher bid	Priority is a queue switch, not an importance label. High priority wins even with a lower bid
Standard Shopping vs Performance Max	Ad Rank, not campaign type	Since October 2024, Performance Max no longer takes automatic precedence. Before that it won on type alone
Campaigns in different accounts or merchants	The ordinary auction	Priority does not reach across accounts — those are two independent bidders

Row two is the most consequential change of the last couple of years and the reason Standard Shopping came back into experienced accounts. Running PMax and Standard Shopping on one feed used to mean handing the traffic to PMax by default. Now it is a genuine contest decided by bid, expected CTR and relevance — which means you can deliberately keep part of the catalogue under manual control.

Priority only applies between Standard Shopping campaigns in the same account. It has no effect on Performance Max at all. That misconception produces a lot of structures that do not do what their owners expect.

Campaign priority: why “low” is often the right answer

Priority has three values and defaults to low for every campaign. The instinct is to set your most important campaign to high — which is exactly the wrong move until you understand the cascade.

The mechanism: the highest-priority campaign is considered first. If its daily budget is exhausted, the product falls through to the next level down. That cascade is not for ranking importance; it is a sieve that separates traffic by query intent.

The priority ladder

Goal: pay little for broad queries and pay properly for queries with clear purchase intent.

Campaign	Priority	Negative keywords	Bid	Captures
Generic	High	Brand and model terms excluded	Low	“running shoes” — cheap broad traffic
Brand	Medium	Model and SKU terms excluded	Medium	“brandname running shoes” — warm traffic
Exact / model	Low	None	High	“brandname airmax 42 buy” — hot traffic

A query hits the high-priority campaign first. If it contains a brand term, that campaign’s negatives block it and the product drops to Brand. Model terms are blocked there too, so those queries reach Exact with its high bid. Result: a generic query costs you a low bid and “buy this exact model” costs a high one, which is the economically correct allocation.

The ladder only works on one condition: the negative keyword sets must be consistent. One missing negative and your hottest query is stuck permanently in the cheap campaign, where it simply loses the auction. Our guide to negative keywords and search terms analysis covers how to build and maintain those lists.

Inventory filters: restricting a campaign to part of the feed

An inventory filter on a Standard Shopping campaign defines which subset of your Merchant Center products that campaign is allowed to show at all. It is a campaign-level setting, not an ad group one, and it is absolute: a product that fails the filter will never serve in that campaign, regardless of what your product groups say.

Attributes available for filtering:

• brand;
• product category (Google taxonomy);
• product type (your own hierarchy from the feed);
• condition (new / used / refurbished);
• item ID;
• custom labels custom_label_0 through custom_label_4.

Conditions inside one filter combine with AND; separate filters combine with OR. So “brand = X AND product type = footwear” narrows the set, while two separate brand conditions widen it.

Custom labels are the real instrument here. They are how you get information into the feed that product data does not carry: margin band, season, stock level, sell-through rate, clearance status. Designing and maintaining them is covered in custom labels in Merchant Center, and populating them automatically is best handled with feed rules and supplemental feeds.

The Performance Max equivalent

PMax has no campaign-level inventory filter. Catalogue splitting happens through listing groups inside the asset group: you include the branches you want and exclude the rest. The distinction matters — an inventory filter defines what a campaign may show, while listing groups distribute an already-available set. And if you exclude nothing in PMax, the campaign claims the entire feed.

Three structures that work in 2026

1. The priority ladder (narrow catalogue, strong brand)

Three Standard Shopping campaigns per the table above, with Performance Max absent or reserved for remarketing. Fits catalogues where the assortment is tight, margin is known per item, and you want full control of traffic cost. The cost is manual negative-keyword upkeep.

2. The hybrid: PMax on heroes, Standard on the tail

The most common configuration in retail accounts:

• Performance Max takes products with proven conversion and enough volume for the model to learn from. Automation performs where it has data.
• Standard Shopping takes new arrivals, seasonal items, low-frequency SKUs and anything without history. You set bids by hand instead of waiting out a learning period.

Split it with labels: custom_label_0 = hero and custom_label_0 = tail. PMax listing groups include only hero; the Standard campaign’s inventory filter keeps only tail. Remove the overlap properly — otherwise two campaigns bid for the same product and Ad Rank resolves it in favour of the higher expected performance, which is not always the same as your better margin.

3. Splitting by margin

For stores where margin ranges from 5% to 60%, splitting by bestsellers is meaningless: a product can convert beautifully and make no money. Build labels around margin bands (margin_high, margin_mid, margin_low) and give each its own campaign and ROAS target. Optimising for profit instead of revenue is covered in product value optimisation in PMax and Shopping.

Five mistakes that turn structure into mush

1. High priority on a small budget. The campaign takes all traffic first, exhausts its daily budget by lunchtime, and the rest of the day runs on entirely different logic. A ladder requires the top campaign to be funded for the full volume of cheap traffic.
2. Out-of-sync negatives. The signature ladder failure: a term missing from the top campaign stays trapped in the cheap layer forever. Audit with the search terms report, not from memory.
3. Accidental PMax/Standard overlap. A product in both campaigns is not a safety net, it is loss of control — you no longer know which bid is in play. Either split explicitly or keep the overlap deliberately and measure it with an experiment.
4. Filtering on a label that is not in the feed. A campaign filtered to custom_label_2 = winter with that attribute missing shows nothing and sits silently at zero impressions. One minute in Merchant Center diagnostics catches it.
5. Treating priority as importance. Giving the “main” campaign high priority and a high bid makes it collect all the cheap generic traffic at a premium price — the precise opposite of the intent.

Diagnosing who is taking whose impressions

• The product report segmented by campaign. Shows which campaign actually served a given item ID. If a product appears somewhere you did not plan, the structure is not doing its job.
• Impression share per campaign. One campaign’s share falling while another’s rises is traffic leaking across the boundary.
• The search terms report for each ladder level. Branded queries appearing in Generic means a hole in the negatives.

If the suspicion is that PMax is eating your Search campaigns instead, that is a different problem with its own diagnostic path — see cannibalisation between PMax and Search campaigns.

Step by step: building the ladder

1. Confirm the feed passes diagnostics with no critical disapprovals — no structure helps a rejected product.
2. Create three Standard Shopping campaigns on the same feed. Set priority in each: Generic high, Brand medium, Exact low.
3. Set budgets in descending order of volume: the top campaign gets the most, the bottom the least but with the highest bid.
4. In Generic, add your brand names as negatives — every spelling, transliteration and common misspelling.
5. In Brand, add model markers as negatives: SKUs, model numbers, “buy”, anything that signals a hot query.
6. After 7–10 days, open the search terms report per campaign and move misrouted queries into the right negative lists.
7. After a month, compare average CPC by level. It should rise from top to bottom. If it does not, the sieve is not working — return to step 4.

One thing worth holding in mind alongside structure: in Shopping, your CSS arrangement affects click cost, and at volume that difference outweighs fine structural tuning. The mechanics are in CSS in Google Shopping and its CPC impact.

Promotions and seasonality without rebuilding every month

The usual reason a tidy structure degenerates into chaos is promotions. A sale arrives, a fourth campaign gets spun up for it, it overlaps two existing ones, nobody switches it off afterwards, and six months later no one remembers why it exists. Build seasonality into the structure up front rather than bolting it on.

Two approaches that hold:

1. Promotions via a label, not a new campaign. Keep a custom_label_3 = promo in the feed, set by a feed rule whenever a product carries a sale_price. One permanent campaign filters on that label with its own priority and bid. It runs all year and sits empty most of the time, which is fine — the moment discounts hit the feed, the campaign populates itself.
2. Seasonality via scheduling, not creation and deletion. Seasonal campaigns should be paused and resumed, not rebuilt each year. That preserves history, and with it the bidding model’s learning, which otherwise restarts from zero every season.

A budget note for peaks: with a priority ladder, the top campaign is the first to run out of budget in high season, and the cascade then misbehaves — cheap traffic falls through into expensive levels. So before a peak, raise the top level’s budget in proportion to expected volume rather than lifting every campaign equally.

Changing structure without destroying your data

Rebuilding a Shopping structure resets campaign-level history, and bidding strategies serve a fresh learning period. Which is an argument for changing less often and more deliberately than instinct suggests. A sequence that lowers the cost of being wrong:

1. Measure the baseline first. Record four weeks of spend, conversions, revenue, average CPC and impression share per existing campaign. Without that, “it got better” is a matter of taste.
2. Change one layer at a time. Feed labels first, since they do not disturb campaigns; then inventory filters; then priorities and bids. Three simultaneous changes make the result unreadable.
3. Use experiments where they are available. Splitting traffic inside the experiments tool compares both variants over the same period, instead of a before-and-after read with seasonality baked in.
4. Wait two to three weeks. Product-level data accumulates far more slowly than keyword data: half your catalogue will have two clicks in a week, and you cannot conclude anything from that.
5. Do not leave campaigns running “just in case.” A paused campaign is harmless; an active one with unintended overlap keeps quietly taking impressions. If the structure is replaced, the old one goes off.

A practical benchmark: if you are restructuring Shopping more often than once a quarter, you are probably reacting to noise rather than evaluating results. Product-level statistics build slowly, and decision speed has to respect that.

Pre-launch checklist

• Feed passes diagnostics; no critical disapprovals.
• Custom labels are populated and refreshed automatically, not edited by hand once a quarter.
• Every Standard Shopping campaign has a deliberate priority rather than the default low across the board.
• Inventory filters do not overlap where overlap was not intended.
• PMax listing groups constrain the catalogue instead of defaulting to the whole feed.
• The top ladder campaign’s budget covers the daily volume of cheap traffic.
• Negative lists are reconciled across levels and verified against the search terms report.
• Someone checks weekly which campaign served which products.

Structure is half the job; the other half is an account stable enough to build it in. If scaling is the constraint, Google Ads agency accounts from PPC Rebels cover that side. And the measurement without which any structure is judged blind is covered in parallel tracking and URL options and the GA4 BigQuery export.

FAQ

Does campaign priority apply to Performance Max?

No. Priority is a Standard Shopping setting and only governs competition between Standard Shopping campaigns in the same account. PMax versus Standard Shopping is settled by Ad Rank.

Is it true Performance Max always outranks Standard Shopping?

Not any more. Since October 2024 Performance Max no longer receives automatic precedence over other campaign types when products overlap — Ad Rank decides.

What priority should my most important campaign have?

If “most important” means the hot, high-value traffic, it usually wants low priority and a high bid. High priority is the first queue, which is the cheapest and broadest slice of demand.

What happens if every campaign keeps the default priority?

The bid decides: whichever campaign bids higher on that product serves. You get no cascade and no separation of traffic by intent.

Can I filter inventory by price or stock level?

Not directly. The available attributes are brand, category, product type, condition, item ID and custom labels. Price and stock get written into custom labels via feed rules, and you filter on the label.

How do I split the catalogue between PMax and Standard Shopping?

With a custom label: an inventory filter on one value in Standard Shopping, listing groups including only the other value in PMax. Remove the overlap explicitly, or the two campaigns will compete with each other.

How many Shopping campaigns per feed make sense?

Two or three for most accounts. A three-level ladder is about the limit of what negative keyword maintenance can realistically support; more levels add routine faster than control.

Why does my filtered campaign show no products?

Usually the filtered attribute is missing from the feed or carries a different value — casing, whitespace, a translated term. Merchant Center diagnostics and a single product preview confirm it.

Should I duplicate products across campaigns as a safety net?

No. Duplication does not expand reach, but it does cost you bid control and muddies analysis. Overlap is only justified as a deliberate, measured experiment.

How do I quickly check the ladder is working?

Compare average CPC across levels. It should rise from high priority to low. Identical CPCs mean the negatives are not dividing traffic and the sieve effectively does not exist.

Should campaigns be split by country?

If feed content or pricing differs by market, yes — otherwise you mix different economics into one auction. With identical assortment and currency, in-campaign geo targeting is enough.

How often should the structure be revisited?

Labels and filters at every meaningful catalogue or season change. Priorities and negatives monthly, driven by the search terms report. A full structural review quarterly, or whenever the business model shifts.