Google Shopping Agency UK: Scope, Selection and First 90 Days

Last updated: 11 July 2026

What you are actually hiring

A Google Shopping agency is useful when it improves the system that decides which products are eligible, which searches they can match, what value Google receives and where budget is allowed to move. Campaign settings are only one part of that job.

The agency does not need to own every component, but every component needs an owner. Feed engineering may stay with ecommerce development, claims with compliance and promotions with trading. The management scope should define the handoffs rather than pretending one PPC login contains the whole problem.

The responsibility map

Diagnose by symptom

When impressions fall, check product status, feed freshness, target changes, budget, demand and competition before rewriting titles. When clicks rise but orders do not, inspect query fit, price competitiveness, page consistency, delivery promise, mobile experience and product availability. When ROAS rises while total revenue is flat, investigate brand capture, returning customers, lost category volume and campaign overlap.

This sequence matters because the same symptom can have several causes. A team that reaches immediately for its favourite tactic will create activity without diagnosis.

Feed work that changes outcomes

Product titles should communicate the attributes people use to distinguish and search for the item. Product type creates a retailer-controlled taxonomy for analysis and campaign grouping. Google product category, brand, GTIN or MPN, variant fields, colour, size and material help Google understand the offer where applicable. Custom labels can carry commercial groupings such as margin band, stock status, season or strategic priority.

Do not bulk-optimise titles without a test design. Define the product set, original pattern, proposed pattern, expected query effect and review period. Keep rules versioned and consider how changes affect free listings, SEO and other channels using the feed.

Merchant Center is an operating queue

Warnings, limited status and disapprovals should become prioritised work, not a monthly screenshot. Product-level issues and account-level issues have different blast radii. Price or availability mismatches often require website or structured-data fixes, while policy enforcement may require product, claim, business-information or destination changes.

Google’s review guidance warns that repeated unsuccessful review attempts can create cooldown periods. Fix the underlying issue and gather evidence before appealing. An agency promising a quick workaround is increasing risk.

PMax and Standard Shopping roles

PMax provides broad inventory and automation; Standard Shopping provides a more controlled product campaign and can support a formal comparison. The right mix depends on the objective, data quality and need for control. Avoid campaign duplication that makes both interpretation and budget allocation harder.

For PMax, define primary conversions, values, customer-acquisition settings, brand controls, URL behaviour, product selection, asset groups and creative cadence. For Standard Shopping, define product groups, query-review process, negatives and bidding approach. The answer is often a portfolio, not a permanent winner.

Reporting for retail decisions

Account ROAS is not enough. Reporting should identify products or categories to scale, repair, isolate or stop. Add stock, margin and return context where reliable. Separate brand demand, new customers and returning buyers when those distinctions affect the decision.

A useful monthly meeting ends with actions owned by media, feed, merchandising, development or finance. If every recommendation belongs to the agency’s Google Ads operator, the analysis is probably too narrow.

The first 90 days

Weeks one and two should produce access ownership, tracking reconciliation, Merchant Center triage, feed diagnosis and a category-level commercial baseline. Weeks three to six should implement high-confidence fixes and launch controlled tests without rebuilding stable campaigns for appearance.

By the end of month three, the retailer should have early test readouts, a documented feed and campaign architecture, a prioritised category plan and a clear list of client-side blockers. Success is not the number of changes; it is whether the account now makes better decisions.

Selection questions

1. Where will feed transformations be stored and can we export them?
2. Who monitors Merchant Center and what is the response process?
3. How do margin, returns and stock change budget decisions?
4. Which products should stay outside PMax?
5. Who produces assets and landing-page changes?
6. What does the proposed team do each week?
7. Which work is outside the retainer?
8. How is the account handed back at exit?

Choose the scope, not the agency label

Ask the agency to recommend the smallest scope that can resolve the problem. If tracking is clean and the in-house feed team is strong, a full feed-platform migration may add cost without benefit. If Merchant Center outages and source-data errors recur, media-only management may be structurally incapable of fixing performance.

Build a useful procurement evidence pack

Give finalists twelve months of monthly spend, orders and revenue; a product export with status; the current feed method; Merchant Center issue history; campaign and conversion-action lists; margin or contribution bands; return rates; stock constraints; promotion calendar; target markets; and the team ownership map. Use ranges or anonymisation where confidentiality requires it.

Include two decisions the business cannot currently make, such as whether a low-ROAS category should be cut or whether PMax is cannibalising brand demand. This reveals the analysis you are buying. Agencies that respond only with account hygiene have not addressed the commercial question.

Three decision scenarios

High ROAS, falling profit

Check discounts, product mix, returns, fulfilment and branded demand before tightening or loosening targets. The agency should reconcile platform value with finance data.

Strong products receive little spend

Check eligibility, feed attributes, campaign inclusion, target constraints, demand and price competitiveness. More budget does not solve a product Google cannot match or serve.

Merchant Center suspends the account

Stop speculative edits, identify whether the cause is business information, destination, policy or product data, complete fixes and assemble evidence before review. The contract should identify the incident lead.

Measurement caveats

Google Ads uses the conversion actions and values supplied to it; it cannot infer true profit from order revenue. Refunds may arrive after reporting, attribution models do not prove incrementality and returning customers can inflate acquisition performance. Decide which limitations remain in bidding and which are corrected in business reporting.

A Shopping agency should state uncertainty rather than manufacture precision. The output of good measurement is often a range or a decision threshold, not one allegedly perfect ROAS number.

What not to outsource blindly

The retailer should retain final authority over product truth, price, availability, claims, promotions and commercial priorities. An agency can recommend values and automate transformations, but it should not silently redefine a product or publish a promotion that trading has ended. Keep one internal owner capable of reading Merchant Center, finding feed rules and pausing campaigns. Outsourcing execution should buy expertise and capacity, not remove the company’s ability to understand its largest paid-commerce channel.

Sources

• Google product data specification
• Google Merchant Center issue guidance
• Google retail PMax practices