A jacket comes in four colors and eight sizes. The shopper chooses one combination. Merchant Center may receive 32 items. The buying team may still manage the jacket as one product.
For Google Shopping, each submitted variant needs its own identity and accurate offer details. A commercial priority belongs at product level when the retailer is making one decision across the variants. Expanly uses that model: one priority per product per scored day, informed by variant data and inherited by eligible offers. Choices that need different economics or strategic treatment call for a deliberate review of the product mapping.
Getting these levels straight prevents a familiar reporting problem: 32 advertised items being discussed as if they were 32 separate commercial decisions.
What Is the Difference Between a Product, a Variant and an Offer?
A product is the commercial item being managed. A variant is a specific choice, such as size and color. An offer is the purchasable item represented in the channel with its ID, price, availability and destination page.
The terms differ between systems. Merchant Center calls submitted items products, including items a retailer would call variants. Treat the mapping as an explicit piece of onboarding rather than assuming every system's “product ID” means the same thing.
| Example level | What it identifies | What the team uses it for |
|---|---|---|
| Jacket model | The commercial product | Launch support, product priority and product reporting |
| Blue, medium | A variant | Availability, size selection and stock diagnosis |
| Merchant Center item ID | The submitted offer | Matching the channel item and checking delivery |
| Merchant Center item_group_id | A group of related variants | Representing the variant relationship to Google |
Keep identifiers stable. A mutable title or current price is a poor identity key because either can change during the period being analyzed.
How Google Shopping Groups Variants
Google's item group ID guidance calls for a unique item ID for each submitted variant and a shared item_group_id for related variants. The grouping and variant details should agree with the landing-page experience.
That allows the channel item to retain the attributes the shopper needs. The advertised size, color, price and availability should match the selected offer on the site. A sold-out medium does not make every other size sold out.
Follow Google's product data specification for the actual submission. The commercial priority model is an additional decision; it is not permission to collapse the primary feed into parent rows or change valid Merchant Center grouping to suit a campaign tactic.
When Should Priority Belong to the Parent Product?
Parent-level priority works when the business is making one coherent decision across the choices.
A footwear launch may deserve a fair test as a model. A seasonal jacket may be approaching its selling deadline. A supplier commitment may concern a particular product range. Splitting those decisions by size can create conflicts that nobody intended.
One size could become High because it sold recently, another Low because stock is thin, and a third Standard because it has little history. The retailer then has to explain several advertising strategies for what was supposed to be one launch.
Product-level priority keeps the decision together. It does not erase the variants. Their stock information can still disqualify the parent from High when the available assortment becomes too narrow.
The test is whether the commercial owner can defend one priority for the whole group. If different choices regularly need different prices, economics or strategic treatment, review whether they are really one product in the priority model.
Google Grouping and Expanly Product Identity Are Separate Decisions
Google permits variant groups whose offers have different prices. Expanly's current product model expects sibling variants to share their non-missing price, sale price and shipping price values.
That distinction matters for a laptop with different processors, a refurbished phone with different storage capacities, or a furniture option with a different commercial price. A group that is valid in Merchant Center may need a more granular internal product mapping in Expanly.
Review the source mapping with the data owner. Use a stable identifier for a separately priced commercial choice; do not create a new identity every time its price changes. Preserve Google's required grouping and the shopper's landing-page experience in the primary feed.
If inconsistent sibling prices enter one Expanly product, the platform averages the available values. The data owner should check the grouping before relying on that figure in a rule. An average is not a substitute for the intended commercial model. It can make a margin or price condition mean something different from what the retailer expected.
Agree the mapping before writing rules that depend on those fields.
Variant Availability Is a Signal, Not a Demand Forecast
A jacket can have plenty of units left in two uncommon sizes while most shoppers cannot find a suitable option.
Expanly's Variant Availability measure can help identify a thinning assortment. The current calculation divides the number of in-stock variants by the number of variants with a known stock signal. It applies to products with at least two variants. A variant with no stock information is left out of the calculation; it is not counted as sold out. If no variant has a known stock signal, the measure has no value rather than 0%, and a rule such as “below 50%” does not match the product.
That means a high share is only as useful as the coverage behind it. If four variants have known stock and four have unknown stock, the result describes the known four. Check missing information before treating the percentage as a complete picture.
The share also does not weight sizes by demand. Two available sizes out of eight is 25% when all eight have known stock, whether those two sizes are the bestsellers or the least popular. Core-size importance needs appropriate retailer-supplied data or a separate commercial check.
Start with representative products. Compare what the metric says with what a shopper can actually buy. Then choose the guardrail.
Put Protective Rules Before the Launch Rule
For a jacket launch, an ordered model might be:
- Low if Variant Availability falls below the agreed minimum.
- High if the product belongs to the active launch and meets the remaining conditions.
- Standard if no rule matches.
Expanly uses the first matching rule. A launch placed above the availability rule can win first, so order is part of the commercial decision.
Take the jacket from the start. All 32 variants have known stock, but only eight remain available. The product still counts as in stock, while its Variant Availability is 25%. If the retailer has set a Low rule for anything below 50% and placed it above the launch rule, the jacket receives Low. Its eligible variant offers inherit that priority when exported.
Missing data needs its own treatment. A product whose availability is unknown should not quietly be treated as passing the guardrail. The team may hold it for review or use another agreed rule, depending on the risk and the available fields.
Preview the product groups before publishing. Count commercial products and offers separately so a launch with many sizes does not appear larger than it is.
One Priority Does Not Mean Equal Advertising Delivery
When a product receives High priority, its eligible offers inherit that decision in the Google destination. Google can still deliver much more advertising to one variant than another.
Here, eligible means the offer is present in Expanly's feed data on the scored date and has an ID mapped to the corresponding Merchant Center offer. Google's own approval and advertising eligibility are separate.
The shared priority gives the specialist a coherent group to use in campaign structure. It does not reserve spend for every size or instruct the auction to divide impressions evenly.
If the retailer needs a specific choice to receive separate treatment, the agency must assess the campaign tactic and the data owner must check the product model. Simply attaching a High label will not guarantee the desired exposure.
Report the Decision at Product Level and Diagnose the Detail Below It
The commercial review should show the products the team meant to support, their recorded priorities and the advertising opportunity they received. Variant detail can explain a weak result: poor size availability, a disapproved offer or a mismatch between the ad and landing page.
Keep metric sources explicit. Google Ads attributed conversions, GA4 item purchases and the store's order records are not interchangeable sales counts. Aggregate additive measures carefully and calculate ratios from their underlying totals.
For allocation over time, use each product's priority on the relevant scored date. Do not apply today's label to last month's spend. For a launch test, preserve the original product cohort as well, so products leaving High do not disappear from the evaluation.
Before activating a new range, take one representative product through the whole chain: source ID, variants, commercial grouping, matched rule, exported offer IDs and report. If the team can reconcile that example, it has a much stronger foundation for product prioritization across the rest of the catalog.
