Product Feed Optimization · focused work
Custom labels for Google Shopping
Custom labels are five free-text fields in a product feed that mean whatever the account decides they mean. They carry no weight in matching and they do nothing by themselves. What they do is let campaign structure follow the business, so budget and targets can differ by margin, season, stock position or the role a product plays rather than only by category.
This is the right work if
- Every product is bid the same way because nothing but category separates them
- High margin and low margin lines share one budget and one return target
- Seasonal ranges keep spending in the months when they cannot sell
- Best sellers and slow movers are treated identically in every campaign
- Clearance stock needs different bidding and there is no attribute that selects it
What it is
What the five fields are for
The work is choosing what each of the five slots holds, sourcing the data that fills them, keeping them current, and building the campaign structure that uses them. The usual assignments are a margin band, a price band, a seasonality marker, a sales velocity or stock cover tier, a lifecycle stage such as new, core or clearance, and a flag for products the business does not want promoted at all. Only five exist, so the choice is a real one, and because campaigns reference the values by name, changing what a label means later means rebuilding every group that uses it.
The data almost never lives in the store. Margin sits in accounting or an ERP, stock cover in the inventory system, velocity in the store's own sales reports. So labels are normally computed outside the platform, keyed on the product identifier and applied through a supplemental source on a schedule that matches how fast the underlying figure moves. A stock cover label refreshed once a quarter is misleading by the time a campaign acts on it. Once the labels exist and are reliable, the payoff is structural: listing groups and product groups split by band, budgets and targets set per band, and reporting that finally shows performance in the terms the business already uses internally.
How it is done
The work, in order
What changes
- Bidding follows margin, season and stock position instead of category alone
- Seasonal ranges are throttled outside their selling window without manual work
- Clearance stock and best sellers are managed as separate groups with separate goals
- Reporting shows performance in the terms the business already uses internally
Define the five before building anything
Write the meaning, the permitted values and the owner of each slot, and get agreement from whoever owns margin and inventory. This is the hardest step and the only one that is genuinely expensive to redo once campaigns reference the values.
Use bands rather than raw figures
Labels are text used in group names, so a margin band or a velocity tier stays readable and stable. An exact margin figure would change with every cost update and would produce a structure with as many groups as products, which is unusable.
Compute where the data lives
Join sales, margin and inventory data in whichever system holds it, produce a file keyed on the product identifier that campaigns will use, and generate the label values there rather than maintaining them by hand in a spreadsheet.
Refresh on a matching cadence
Deliver the values through a supplemental source at a frequency that tracks the underlying data, and monitor that each run succeeded. A failed refresh does not raise an error in the campaign; it simply leaves last month's labels in place.
Build the structure on the labels
Split listing groups or product groups by label so budgets, bids and return targets can differ by band, and keep the structure documented alongside the definitions so the two do not drift apart as campaigns are edited.
Report by label and act on it
Read spend, revenue and margin per label each period and respond by moving products between bands or changing targets, not only by adjusting bids. The labels are worth the effort only if something in the account changes because of them.
Platform notes
Shopify
Metafields can hold a label value that a merchandiser maintains, provided the feed app is configured to map them, which is worth checking early because not every app exposes every metafield to the feed.
WooCommerce
Custom fields on the product or variation can carry the values, and where the data comes from an accounting system the practical route is a scheduled import that writes those fields before the feed is generated.
Do all five labels need to be used?
No, and filling them for the sake of it creates maintenance without benefit. Start with one that would change a bidding decision today, prove it works in the campaign structure, then add another. Two well-maintained labels beat five that nobody refreshes.
Can the meaning of a label be changed later?
Technically yes, and it is disruptive. Campaign groups reference the values, so a redefinition means rebuilding those groups and losing the history attached to them. Treat the definitions as a schema decision and pick values general enough to survive a few years.
Do custom labels affect whether a product is shown?
Not directly. They are not used to match a product to a query, and adding one does not change eligibility. Their effect is entirely through the structure and bidding built on top of them, which is why an unused label is genuinely doing nothing.
Which label is the most useful place to start?
For most stores, a margin band, because it separates products the business can afford to bid on from those it cannot, and that distinction is invisible to any bidding system otherwise. Seasonality is a close second for catalogs with a strong selling window.
Related work and answers
Find the leak.
A free Growth Analysis ranks what your store should fix first, by revenue at stake.