i
Brand separation
Brand traffic isolated with account-level negatives so Performance Max cannot absorb it and report it as incremental. This single change routinely reveals that measured efficiency was 30–50% worse than believed.
Services
Six services. One margin to answer to.
Any of these can be bought alone. Taken together they share one product data source, one definition of margin and one profit report — which is where most of the improvement comes from.
I — Shopping & Feeds
The feed is the campaign
Retailers spend months debating bidding strategy while their product titles are still whatever the manufacturer supplied, half their images have white borders that fail policy, and forty per cent of the catalogue has no GTIN. No amount of structure or bidding rescues that.
We treat product data as the primary lever. Titles rewritten to match how people actually search, attributes completed, images cleaned, and custom labels applied for margin tier, stock position, seasonality and return rate so campaigns can be built around commercial reality rather than website taxonomy.
- Title rewriting at scaleRules-based rewriting across the full catalogue, tested against search term data rather than guessed at.
- Custom labels that mean somethingMargin tier, stock cover, return rate and seasonality — the four labels that let a campaign behave commercially.
- Priority structures where they earn itCampaign priority and negative sculpting used deliberately, not as a default template applied to every client.
Typical findings in a first feed audit
Products with no GTIN18–44%
Titles unchanged from supplier60%+
Live disapprovals200–3,000
Products advertised below margin15–40%
Out-of-stock lines still servingCommon
II — Performance Max for Retail
Automation is a colleague, not an oracle
Performance Max works well when it is given clean product data, a sensible structure and a target that reflects margin. Left alone it will optimise toward the easiest conversions available, which in retail usually means people who were going to buy anyway.
ii
Asset groups by product family
Structured by margin tier and product family rather than one group per campaign, so budget can be steered toward what is worth selling and creative can actually be relevant to it.
iii
Complete asset sets
Every ratio populated and real video included, rather than five images and whatever Google auto-generates. Starved asset groups are the most common cause of poor Performance Max results.
iv
Experiments before budget moves
Campaign experiments and geo splits run before material target or budget changes. We would rather spend three weeks proving a change than a quarter unwinding one.
How we report Search
BrandSeparate line
Category & genericSeparate line
Competitor termsBudgeted
Long-tail productSeparate line
Search term reviewWeekly
III — Search & Brand Protection
Brand search is not growth
Defending your own name in the auction is usually worth doing. Reporting it in the same line as category search, and calling the blended figure a return on ad spend, is how retailers end up believing Google Ads is three times more efficient than it is.
We report brand, category, competitor and long-tail product search as four separate lines, with a stated budget for each. It makes the headline number look worse and every decision that follows considerably better.
- Brand reported separatelyAlways, on every report, without exception or aggregation.
- Competitor conquesting budgetedA deliberate decision with a cap, not something that happens because a broad match term drifted.
IV — YouTube & Demand Gen
Creating next season's demand
Shopping and Search harvest demand that already exists. For a retailer trying to grow rather than simply defend, something has to create it — and inside Google Ads that means YouTube and Demand Gen, measured honestly.
i
Product-led video
Short video that shows the product doing its job, cut for silent autoplay with captions. Produced from your existing photography and product footage wherever possible.
ii
Feed-linked Demand Gen
Demand Gen campaigns connected to the same corrected product feed as Shopping, so nothing is promoted at the wrong price or after it has sold out.
iii
New customer acquisition
Optimised toward first-time buyers using new customer acquisition goals and value rules, because retention and acquisition should not share a target.
iv
Measured on incrementality
Geo holdout tests twice a year rather than platform-attributed conversions, because attributed and incremental are not the same number and everyone in retail knows it.
V — Merchant Center Health
Products cannot sell if they are not in the auction
Merchant Center is where retail Google Ads quietly fails. A price mismatch of a few pence, a shipping configuration error, a missing GTIN or an image that trips policy will remove a product from the auction entirely — and nothing in the Google Ads interface will tell you why performance dropped.
We monitor the whole catalogue daily, clear disapprovals at the cause rather than the symptom, and keep promotions, free listings and shipping configuration current through every trading peak.
- Daily monitoringDisapprovals, price mismatches and availability errors caught within a day rather than at the monthly report.
- Promotions and free listingsConfigured and maintained, so you appear in the unpaid surfaces you are entitled to.
- Shipping and tax accuracyChecked per market, because a wrong delivery estimate suppresses conversion before it suppresses impressions.
Merchant Center monitoring
Disapproval checkDaily
Price mismatch checkDaily
Feed refreshMultiple daily
Promotions reviewWeekly
Full catalogue auditQuarterly
VI — Profit Analytics
One number, reconciled to the management accounts
This is the service that makes the other five work. Without it we would be optimising to revenue like everybody else, and we would be no more useful than the agency you are currently thinking of replacing.
i
Contribution margin by SKU
Revenue less cost of goods, delivery, payment fees and modelled returns, at product level, joined to Google Ads cost by campaign. Built in BigQuery, documented, and yours.
ii
Return rate modelling
Return rates vary enormously by product family and customer cohort. Modelling them properly changes which products deserve budget, often dramatically.
iii
Stock-aware pacing
Daily stock feeds drive budget pacing and custom label updates, so spend comes off lines about to sell through and nothing is advertised into an out-of-stock page.
iv
Reporting that reconciles
A monthly profit report that ties to your management accounts. Where it does not tie, we fix the report rather than explain the difference away.
Engagement
Trading retainer, project, or advisory.
Most clients take a monthly trading retainer covering the whole Google Ads account plus profit analytics. We also run fixed-scope projects — a feed rebuild, a profit reporting build, an audit of an incumbent agency — and an advisory retainer for retailers with capable in-house teams who want senior review rather than delivery.
Discuss an engagement