Ecommerce PPC agency for retailers doing £3m+ online.

Google Shopping

Shopping is fundamentally a product allocation system. Google needs to understand what the product is, how relevant it is, and how likely it is to convert. We add the commercial question: what is this product worth pushing? That means considering margin, stock, customer value, lifecycle, demand, and commercial role.

Performance Max

We use PMax. Quite happily. We're neither evangelists nor doomsday preppers. Performance Max is a powerful execution mechanism when the feed is strong, measurement is reliable, the product set makes sense, the commercial objective is clear, and Google has appropriate boundaries.

PMax is a buying tool. Not a strategy.

Product feed optimisation

The product feed is one of the most important inputs available to ecommerce advertisers. It tells Google what you're selling, what attributes products have, which queries might be relevant, and how products relate. We optimise titles, product types, GTINs, attributes, images, custom labels, Merchant Center data, and commercial classifications.

As Google automates more of the execution, the quality of your inputs matters more.

Search

Search remains one of the clearest expressions of customer intent available. Someone tells Google what they want. The challenge is deciding how valuable that demand is. We manage brand, generic, category, product, competitor, long-tail, and high-intent Search.

Query intent is only half the equation. Commercial value determines what that intent is worth.

Merchant Center

Merchant Center isn't somebody else's administrative problem. For ecommerce advertisers it sits at the centre of Shopping, Performance Max, product eligibility, pricing, promotions, availability, and feed health. We manage it as part of the advertising system.

If 30% of the catalogue disappears due to feed issues, your bid strategy has bigger problems than its learning status.

Specialist contexts

• Shopify: Shopify knows what happened after the click. We use that information to make the next click more valuable.
• High-SKU ecommerce: We don't optimise every SKU manually. We optimise the system deciding how SKUs are treated.
• International ecommerce: Scale in Britain. Profit in Germany. Gateway in America. The market has to earn the investment.
• Subscription ecommerce: Cheap acquisition and valuable acquisition aren't necessarily the same thing.

We care about marginal return too.

Suppose, hypothetically, you're spending £100,000 per month at a historic 5.0 ROAS. That doesn't tell us whether the next £10,000 should go into Google. The next £10,000 might produce excellent incremental return, acceptable return, or destroy value.

Capital allocation happens at the margin. That's why historical averages alone aren't enough.

Incrementality matters.

Attribution asks what Google Ads claimed credit for. Incrementality asks what happened because we advertised. Different question. This becomes particularly important around brand, remarketing, existing customers, Performance Max, and mature demand. Where appropriate, we use experiments, brand suppression, holdouts, geo testing, and causal analysis.

Attributed profit isn't automatically incremental profit.

We don't optimise to activity.

Agencies can manufacture activity indefinitely. Bid changed. Budget moved. Campaign renamed. Keyword added. Recommendation dismissed. Meeting held. Deck produced. Everybody survives another Tuesday.

We prefer: what changed commercially? Why? What decision did we make? What happened afterwards? What should happen next?

We show our working.

If we recommend increasing spend, reducing spend, changing a product's job, or testing something uncertain, you'll understand why, and what we're trying to learn. If we don't know, we'll tell you.

We don't believe expertise requires pretending uncertainty doesn't exist.

We challenge clients too.

Specialist advice is fairly pointless if the agency simply agrees with everything. If we think the ROAS target is constraining profitable growth, a promotion will destroy contribution, a market isn't ready, a product shouldn't receive more spend, the tracking is unreliable, or the commercial assumption is wrong, we'll explain why.

You don't need another person nodding on Zoom.

We use automation. And AI.

Google can process auction signals, customer behaviour, and enormous volumes of historical data. Humans cannot compete with that execution manually, nor should they try. We use machines for execution. Humans remain accountable for the objective. We also use AI for analysis, research, data processing, feed work, automation, and anomaly detection, because refusing to use AI in 2026 would be an unusual commitment to inefficiency.

But AI output still needs judgement. The tool doesn't become accountable because it generated the spreadsheet faster.