The five priorities, in order
For most ecommerce accounts, the order is: trustworthy measurement, commercially important product coverage, accurate and useful product data, deliberate budget allocation, then product-page conversion. The order matters. A team cannot optimise profit with false revenue values or scale a range that Merchant Center has not approved.
This guide is the executive view. Each priority has a pass condition, an owner and a metric. It is intended to help decide what receives the next week of work, not to replace the field-level implementation guides.
1. Trust the purchase and profit data
Run a test purchase and verify one transaction, correct value, currency and transaction ID. Confirm bidding uses the intended primary conversion. Reconcile Google Ads revenue with backend orders over a fixed period and explain differences caused by attribution, VAT, shipping, refunds or consent.
Then add product economics. A revenue ROAS target should be derived from contribution before advertising, not copied from another retailer. If a £100 order leaves £30 after VAT, product cost, fulfilment and expected returns, 333% is the break-even revenue ROAS before any required profit. Different margin groups may need different targets.
Pass condition: the team can explain how an Ads purchase becomes net contribution and which conversion actions drive bidding.
2. Maximise accurate commercial coverage
Use Merchant Center’s Needs attention view and export the affected items. Google separates warnings, product disapprovals and account-level issues in its Merchant Center issues guidance. Prioritise account risk and high-value products, not the largest issue count.
Track revenue-weighted eligibility. If 95% of SKUs are approved but the missing 5% represents a quarter of sales, the account is not 95% healthy. Verify price, availability and shipping through checkout, and monitor source processing after every major catalogue release.
Pass condition: nearly all commercially important, in-stock products are eligible in the intended markets, with no unresolved account-level risk.
Unsure which Shopping priority comes first?
Upscale combines measurement, Merchant Center, product economics and conversion evidence into one ordered plan.
Book My Free Ad Audit3. Give Google accurate product information
Requirements come first: stable IDs, accurate title, description, link, image, price and availability, plus conditional fields for the product and market. Use Google’s current product data specification.
Then improve understanding. Titles should identify the product and variant within the 1–150-character limit, with decisive attributes early. Product type should reflect a clean retailer hierarchy; Google product category should use the appropriate Google taxonomy. Submit manufacturer-assigned identifiers accurately and align variant title, image, price, availability and landing-page selection.
Pass condition: a reviewer can identify the exact item from the processed feed, and high-impression products do not rely on vague collection names.
4. Allocate spend by commercial role
Use the product report to identify concentration, waste and suppressed opportunity. Products should be separated only when they need a different budget, target, campaign type, market or promotional window. Custom labels can encode margin, stock and lifecycle, but a label changes nothing until it supports a real decision.
Set a decision threshold for non-converting spend based on allowable acquisition cost and conversion lag. An item with £10 spend and no order may need more data; an item with four times the allowable acquisition cost and no basket activity needs diagnosis or exclusion.
Pass condition: every campaign boundary has a stated commercial reason, and the account can report contribution and spend by that boundary.
5. Fix the product-page stage that leaks
Shopping clicks usually land on a product page. Analyse product view → add to basket → checkout → purchase by item and device. Low basket rate points toward relevance, offer or page persuasion; strong basket rate with poor purchase rate points toward delivery, checkout or payment friction.
The page must also satisfy Merchant Center landing-page requirements: the exact product must be purchasable, and price and availability must agree. Optimise beyond compliance with clear images, variant controls, delivery date and cost, returns, reviews and mobile performance.
Pass condition: the team can name the leaking funnel stage for high-spend products and has a measured test attached to it.
A practical priority matrix
| Finding | Priority | Reason |
|---|---|---|
| Purchase value duplicated | Immediate | Invalidates bidding and reporting |
| Flagship range not approved | Immediate | Direct loss of eligible demand |
| High-impression titles omit product type | Next | Material matching opportunity |
| Ten low-volume descriptions are short | Later | Low expected commercial impact |
| Checkout fails for a mobile payment method | Immediate | Affects all paid traffic, not only Shopping |
The weekly scorecard
Keep it small: conversion-value reconciliation, revenue-weighted eligible coverage, item-level spend beyond the no-sale threshold, net contribution by product group, products receiving impressions, add-to-basket and purchase rate by device, and unresolved high-priority Merchant Center issues.
Review trends and absolute numbers. A rising ROAS with collapsing order volume may be the result of a restrictive target. More eligible products without more impressions may mean demand or campaign inclusion is the next constraint. The scorecard should reveal the next question, not reward a single number.
A 30-day sequence
- Week 1: verify purchase measurement and Merchant Center account health.
- Week 2: recover high-value product coverage and map product economics.
- Week 3: fix the highest-impact feed or allocation problem in one cohort.
- Week 4: diagnose the product-page funnel and review early operational results.
Do not change titles, campaign structure, target and product page for the same cohort in the same week. That may create movement, but it destroys learning.
Worked priority decision
A hypothetical retailer has three concerns: ROAS has fallen from 500% to 420%, 12% of flagship products are not approved, and mobile checkout completion is 35% lower than desktop. The team should not begin by raising the ROAS target.
First, reconcile Ads and backend sales to confirm the decline is real. Second, restore accurate, eligible coverage for flagship items and verify their campaign inclusion. Third, inspect the mobile checkout failure with error and payment-method data. Only then should it decide whether bidding needs adjustment. This order addresses corrupted inputs and lost conversion opportunity before suppressing demand.
Decision thresholds for leadership
Agree thresholds in business terms. Examples: immediate escalation for an account-level warning; same-day response when revenue-weighted coverage drops more than five points; product review after spend exceeds three allowable acquisition costs without a purchase; and CRO investigation when a high-volume device or template sits materially below the baseline for two complete trading cycles.
These thresholds are examples, not Google requirements. Set them from conversion lag, order value, risk and team capacity. Their value is consistency: the same evidence should trigger the same level of attention.
What not to do
Do not run a catalogue-wide title rewrite, campaign restructure and product-page redesign at once. Do not treat warnings by count without weighting the affected products. Do not call a higher ROAS a win if order volume and contribution fall. Do not increase spend because the account is “learning” when purchase measurement is unreliable.
When performance is already weak, use the Shopping diagnostic guide; when the leak is post-click, use the product-page CRO procedure.
Leadership should ask for the evidence behind every priority and the metric that will close it. “Feed work” is not a workstream until the affected products, failure mode, owner and acceptance condition are named.