Our Position

Automation is excellent at execution.
It's dangerous at strategy.

PPC is becoming a commodity. Button-pushing will be fully automated within two years. What won't be automated is the judgment layer - deciding what to optimise for, where to allocate capital, and when to stop.

These are the six decisions we refuse to hand to an algorithm. Not because we're opposed to automation, but because these decisions require commercial context that no platform has access to.

Book a Strategy Review Commercial Reality

The Thesis

The agencies that survive won't be the ones that push buttons better.

Google's algorithms are getting better every quarter. Within two years, bid management, audience targeting, and creative rotation will be fully automated. The agencies built on those capabilities will be commoditised out of existence.

What survives is the strategic layer. The ability to map platform metrics to business outcomes. To make intentional trade-offs between revenue and profit, between acquisition and retention, between short-term performance and long-term commercial health.

The what. The why. The how. Not which buttons to press, but which questions to ask before pressing them.

The Six Decisions

What we refuse to delegate to algorithms

Each of these decisions requires commercial context - your margins, your cash flow, your strategic priorities - that no platform has access to. Automating them means optimising for the wrong objective.

Decision 1

Which SKUs deserve ad spend

What automation does

Smart Bidding allocates budget to whatever converts cheapest. It doesn't know - or care - whether that conversion makes you money.

What we decide instead

We map contribution margin per SKU and assign commercial roles (Scale, Protect, Recover, Pause). A product that converts easily at 5% margin shouldn't get the same budget as one that converts harder at 45% margin. This is a commercial judgment, not an algorithmic one.

The risk: Without this: your budget gravitates toward high-volume, low-margin products. Revenue looks great. Profit shrinks.

Decision 2

When to slow down, not speed up

What automation does

Algorithms optimise for volume within a target. They don't have a concept of 'enough' - only 'more, within constraints.' Hit your tROAS? Spend more.

What we decide instead

We monitor marginal CPA at every spend increment. When the next £1,000 generates less profit than the last £1,000, we stop - even if ROAS is still above target. Diminishing returns are invisible to platforms but obvious in your P&L.

The risk: Without this: you scale past the profit ceiling and enter the zone where every additional pound of spend reduces total profit.

Decision 3

How to value a first-time customer vs a repeat buyer

What automation does

Google values every conversion at the same reported value. A £50 first order and a £50 reorder are identical to the algorithm.

What we decide instead

A first-time customer buying a gateway product at break-even can be worth £800+ in lifetime value. A repeat buyer needs no acquisition spend at all. We adjust conversion values to reflect commercial reality, not transactional parity.

The risk: Without this: you overspend on customers you'd get anyway and underspend on acquisition that builds long-term value.

Decision 4

Which Google recommendations to reject

What automation does

Google's optimisation score rewards you for accepting recommendations. More broad match. More automation. More budget. All optimised for Google's revenue, not yours.

What we decide instead

We evaluate every recommendation against your margin structure and commercial objectives. We reject ~70% of them. 'Add broad match to your exact match campaigns' is advice that benefits Google's auction revenue, not your contribution margin.

The risk: Without this: your account gradually drifts toward Google's preferred structure - maximum spend, minimum control, opaque performance.

Decision 5

Where to draw the line between brand and non-brand

What automation does

PMax and Smart Shopping blend brand and non-brand traffic into a single campaign. Your 12x ROAS on brand searches inflates the apparent performance of your 2x non-brand activity.

What we decide instead

We isolate brand traffic so you can see true acquisition performance. Some brands discover their 'amazing' 6x blended ROAS is actually 15x brand + 1.5x non-brand - meaning they're losing money on every new customer acquired.

The risk: Without this: you celebrate ROAS numbers that are 60-80% subsidised by customers who would have bought anyway.

Decision 6

When a product should stop being advertised

What automation does

If it converts, bid on it. The algorithm has no concept of 'this product's margin doesn't justify advertising.' It only sees conversion probability.

What we decide instead

Some products should exist on your site but never in your ad account. A £12 item at 8% margin with a 25% return rate costs you money on every sale Google drives. It should be discovered organically or as a basket addition - not as a paid acquisition target.

The risk: Without this: you pay to drive sales that lose money, and the algorithm keeps finding more of them because they convert easily.

For Balance

What we happily automate

We're not anti-automation. We automate aggressively - where automation is genuinely better than human execution. The distinction is between execution tasks and judgment calls.

Real-time bid adjustments

Algorithms process millions of auction signals per second. No human can match this speed or granularity. We let Smart Bidding execute - within the strategic constraints we set.

Feed attribute updates

Price changes, stock status, promotion overlays. Automated feed management ensures your product data is always current without manual intervention.

Budget pacing

Daily spend distribution, dayparting, and seasonal pacing rules. Once the strategy is set, execution should be automatic and consistent.

Negative keyword harvesting

Automated scripts identify and exclude irrelevant search terms. The rules are set by us; the execution runs continuously without manual review.

Performance alerts

Automated monitoring flags anomalies - CPAs spiking, conversion rates dropping, spend pacing issues. We set the thresholds; automation does the watching.

Reporting aggregation

Pulling data from Google Ads, Analytics, Shopify, and your P&L into a single view. Automation handles the plumbing; we handle the interpretation.

The Framework

Strategy sets the constraints. Automation operates within them.

This is the operating model that separates a strategic partner from a button-pusher. The value isn't in the execution - it's in defining what to execute.

The What

Which products get budget. Which metrics define success. Which customers are worth acquiring at a loss. These are commercial decisions that require P&L visibility, not just ad account data.

SKU role assignment
Margin-banded targets
Acquisition vs retention split

The Why

The commercial rationale behind every allocation. Why this SKU gets 3x the spend of that one. Why we're capping budget on your best-converting product. Why we're spending at break-even on a product that's never profitable.

Gateway SKU investment thesis
Diminishing returns analysis
Working capital constraints

The How

The implementation architecture that translates strategy into platform configuration. Campaign structure, feed customisation, conversion value rules, and bidding constraints that encode your commercial priorities.

Campaign isolation architecture
Custom label taxonomy
Post-sale value adjustments

Frequently Asked Questions

Automation, strategy, and the future of PPC

Smart Bidding optimises within the constraints you set, but it can't set the constraints themselves. It doesn't know your COGS, your return rates, your cash flow cycle, or your strategic priorities. It can find the cheapest conversion - but deciding whether that conversion is commercially valuable is a judgment call that requires understanding your business, not just your ad account.

Automation is excellent at execution - adjusting bids across millions of auctions in real time. But execution without strategy is just efficient waste. The future isn't less human involvement, it's human involvement at a higher level: setting the commercial framework that automation operates within. The agencies that survive will be the ones who provide the strategic layer, not the ones who push buttons.

Every decision follows our documented frameworks: the SKU Job Framework for product roles, the Commercial Risk Index for spend allocation, and margin-banded custom labels for bidding architecture. These aren't opinions - they're systematic processes with clear inputs (your COGS, margins, return rates) and clear outputs (spend allocation, bid targets, product exclusions). The judgment is in the framework design, not in ad-hoc guesswork.

We test every strategic decision before committing fully. SKU role assignments start with a 14-day validation period. Budget reallocation happens in controlled increments with rollback thresholds. We document every decision, the rationale behind it, and the success criteria - so if something doesn't perform as expected, we know exactly what to adjust. Transparency means owning the downside too.

Most agencies set a tROAS target across your entire account and let Google's algorithms do the rest. They'll call this 'strategy.' We call it abdication. The difference is specificity: we make product-level, margin-aware decisions that require your commercial data, not just your ad account data. If an agency can manage your account without ever seeing your P&L, they're not making strategic decisions.