Performance Max Agency UK
Performance Max agency for established ecommerce brands
We manage Performance Max around your product margins, stock priorities and growth goals. Our work covers campaign structure, product feeds, conversion values, brand controls and reporting, with fixed-fee senior management.
For UK ecommerce brands spending £15k–£500k+/month. From £2k/month fixed fee, never a percentage of media spend.
This page is about choosing an agency: how we work, what it costs, what you get and how to evaluate the answers. The Performance Max management service page covers the implementation itself: campaign structure, targets, exclusions, listing groups and reporting.
Trusted by leading brands
What control inside PMax actually means
Structure by commercial role, not category
Campaigns are split by our BOI® framework, where each SKU's job (scale, profit, protect, recovery, gateway) would justify a different target or budget and there is enough conversion signal for the split to be learnable. Where those conditions are not met, a single campaign with margin-weighted values is the sounder structure. A blended target across products with very different economics can move budget towards whatever converts most easily rather than what contributes most, which is the risk the split is there to address.
Margin in the values
Margin-adjusted conversion values and, where the data supports it, cart-level COGS, so bidding is weighted towards contribution rather than order value alone. This improves the inputs; it does not turn every bid into a net-contribution decision, so we keep checking the outputs against your own numbers.
Brand handled deliberately
Brand lists maintained at account level and the exclusion applied to the campaigns we select, so branded and non-branded demand can be read apart rather than blended.
Feed as targeting
Titles, product types, custom labels and attributes decide what PMax can match against and how listing groups can be organised, so feed work is campaign strategy rather than housekeeping.
Reporting used for what it can show
Performance Max now has native channel performance reporting across Google Search, Display, YouTube, Discover, Maps, Gmail and Search partners, alongside search-term reporting and placement reports. We use them, and we are clear about the limits: disclosure is partial, and there is no native per-channel budget control. Where the data stops, targets get more conservative, not less.
See our Google Shopping page for how we run feeds, and our POAS methodology for how margin reaches the bidding.
Published work where Performance Max was in scope
One published case study documents Performance Max inside the scope of the work. The figures below are account-level outcomes over the period, not an isolated Performance Max effect, and the case study sets out how they were measured.
Thermos
Margin-aware bidding across Search, Shopping and Performance Max
Scope included correcting double-counted conversion tracking, restructuring the account by COGS tier, moving targets onto contribution margin, and improving product feed data so Shopping and Performance Max had something better to work with across the drinkware range.
- +94% contribution margin, account level over the period
- +86% revenue and +86% conversions, measured on the corrected tracking basis
This is a before-and-after account comparison, not a controlled or incrementality test, and Performance Max was one of several channels changed at the same time. No part of the result is attributed to Performance Max on its own.
Read the Thermos case study, including how the figures were measuredWant to see what your PMax reporting can tell you?
Our guide walks through the channel performance and search-term reporting Google provides for Performance Max, what each report can and cannot answer, and what to do where disclosure stops. Google's own documentation is at support.google.com/google-ads/answer/16260130.
How should PMax asset groups and campaigns be structured for a high-SKU catalogue?
By commercial job, not by product category. It helps to be precise about what each level does: asset groups organise creative assets and, through listing groups, which products are eligible to serve. They do not carry their own budget or their own target ROAS. Budget and target ROAS sit at campaign level. So a 60% margin bestseller and a 6% margin clearance line inside one campaign share one target however the asset groups are arranged. We assign every SKU one of five BOI® jobs, map those jobs onto listing groups, and separate them into different campaigns only where the difference in economics justifies the extra structure and each campaign would still have enough conversion signal to bid on.
BOI® job
Scale
Margin, stock and auction headroom. Given the widest budget headroom, which means its own campaign where the target differs from the rest.
BOI® job
Profit
The default job for healthy SKUs. Campaign target set against contribution margin per pound, not revenue.
BOI® job
Protect
Contested, strategic products. Hold visibility without chasing incremental volume at a loss.
BOI® job
Recovery
Aged or overstocked lines with a deliberate objective to convert stock into cash within a contribution floor and a time box. Budget and listing groups follow that objective rather than a blanket suppression of clearance products.
BOI® job
Gateway
Acquisition products. Lower per-unit margin accepted, measured on new customers rather than blended ROAS.
At catalogue scale this runs to dozens of listing groups across a small number of campaigns, reassigned as stock and margin move. Read the framework on the BOI® methodology page and how it applies to high-SKU catalogues.
How do you stop PMax cannibalising branded search?
Performance Max can serve on branded queries, which are usually the cheapest conversions in the account. That tends to flatter reported ROAS and make prospecting performance harder to read. The response is structural, not a bid adjustment.
1. Brand exclusions, applied campaign by campaign
Brand lists can be maintained at account level in Google Ads, but the exclusion is applied to the campaigns you select, so we check which PMax campaigns actually have it on and keep the list current as new variants and misspellings appear. Which inventory a campaign is eligible for, including whether it is opted into Shopping inventory, is set separately and verified against Google's current documentation.
2. A ring-fenced brand campaign
Exact-match branded search with its own budget and its own target, so brand performance is visible instead of blended into PMax.
3. Regular search terms reconciliation
PMax search category data checked against total branded volume, within the limits of what Google discloses for this campaign type.
4. Brand separated in reporting where the data allows
We report non-branded performance separately wherever the available reporting supports the split, and say so where PMax's disclosure limits make a clean split impossible. Excluding brand improves how the numbers read; it does not on its own establish incrementality.
Related: how to fix PMax brand cannibalisation.
When should you consider Standard Shopping alongside Performance Max?
When you need per-product bid control that Performance Max does not offer. Suppressing products is not one of those reasons: PMax listing groups can exclude individual products or product groups from serving, so removing a line from circulation does not require Standard Shopping. These are the signals we look for:
- A small number of SKUs carry most of the contribution margin and need individual bid control
- Low-volume accounts where PMax has little signal to work with
- Catalogue-wide margin variance so wide that any single campaign target is wrong for most products
- No reliable product economics available to measurement or bidding yet, so automated value bidding is working without contribution data
In practice most accounts we take on run both: Standard Shopping holding the margin-critical SKUs where control matters, PMax carrying the long tail. See the full comparison in PMax versus Standard Shopping and our Google Shopping management service.
How do you prove Performance Max is incremental?
Only a designed experiment can support a causal claim. Reported PMax ROAS counts conversions the campaign was credited with, including customers who may have bought anyway, and no report on its own separates the two. One method tests incrementality; the other two are diagnostics that describe the mix of demand.
Controlled experiment or holdout (a test)
A geo or audience holdout run over a full purchase cycle, using Google Ads experiments where they apply, comparing total sales rather than platform conversions. Results depend on design, spend, seasonality and volume, and short or small tests may be inconclusive.
Brand-excluded baseline (a diagnostic)
Rerun performance with brand separated. The gap shows how much reported performance rests on branded demand. It does not measure incrementality.
New customer share (a diagnostic)
Track new-customer acquisition and contribution margin after ad spend rather than total conversions. Useful signal, not causal proof.
Google's own guidance on running campaign experiments is at support.google.com/google-ads/answer/12997711.
Further reading: the PMax reporting guide: what the channel and search-term reports can show and why we bid on POAS rather than ROAS.
Worked example: when reducing PMax spend may be justified
"Show me a case where you reduced or paused PMax despite favourable platform reporting" is a fair question to put to any Performance Max agency. Our fee is fixed rather than a share of spend, so recommending a smaller PMax budget does not reduce what we earn. Here is how the answer sounds when a method sits behind it.
Hypothetical illustration
The following is a hypothetical illustration of the method, not a client result and not an outcome we are claiming. Suppose an account shows PMax at a 7:1 reported ROAS and the incumbent agency calls it the best campaign in the account. Separate brand out and non-branded ROAS falls below 2:1. Load the same spend against contribution margin rather than order value and a share of it sits on clearance lines that lose money per order once returns and shipping are counted. The response would be to reduce the PMax budget, take per-product control of the margin-critical SKUs, and reassign the clearance lines to Recovery jobs. Reported ROAS would fall while contribution margin improves. Whether that trade holds in a given account is a question for measurement, not assertion.
Why your best-reported campaign can be your worst investmentQuestions worth asking any agency
- Is more PMax budget the default answer, or an evidenced one?
- Are asset groups and listing groups built around commercial roles or website categories?
- Which number is reported as the headline, and does it account for margin?
- How is branded demand separated from non-branded in the reporting?
- Is the fee tied to media spend?
JudeLuxe
- PMax budget reviewed against experiment evidence where a test is feasible
- Campaigns and listing groups built by BOI® commercial job
- Contribution margin after ad spend as the headline number
- Brand exclusions applied and checked on the relevant campaigns
- £2k/month fixed fee, independent of media spend
How to evaluate a Performance Max agency
Four questions worth asking any agency, including us. The answers are checkable.
What data do they review before proposing anything?
At minimum: conversion tracking and value configuration, Merchant Center diagnostics and product eligibility, current campaign and listing group structure, brand exclusions and their applied state, and your cost data (COGS, shipping, returns). A proposal built only on platform ROAS has not seen your economics.
What must a proposed structure justify?
Every split should have a reason on both sides: a genuine difference in product economics, and enough conversion volume for each campaign to bid on. Structure that satisfies one and not the other usually costs more in lost signal than it gains in control.
How is brand and profit reporting reconciled?
Ask how branded and non-branded performance are separated, which reports that relies on, and where Performance Max disclosure makes a clean split impossible. Then ask how platform-reported revenue is reconciled with your own order and contribution data, and how unresolved differences are documented rather than smoothed over.
How are results tested?
Diagnostics describe what happened; only a designed experiment, such as a geo or audience holdout run over a full purchase cycle, supports a causal claim. Ask which changes will be tested, which will not, and why.
Performance Max agency FAQs
What does a Performance Max agency actually do?
A Performance Max agency owns the inputs Google still lets you control: asset group structure, listing groups, feed quality, audience signals, exclusions, budget allocation between PMax and the campaigns around it, and the conversion values PMax bids against. The bidding itself is automated. Everything that decides what the automation optimises towards is not.
How should PMax asset groups and campaigns be structured for a high-SKU catalogue?
By commercial job, not by product category. Asset groups organise assets and, through listing groups, which products are eligible to serve; they do not carry their own budget or their own target ROAS. Budgets and target ROAS are set at campaign level, so where products genuinely need different economics we separate them into different PMax campaigns rather than assuming a group can hold its own target. JudeLuxe assigns every SKU one of five BOI® jobs (Scale, Profit, Protect, Recovery, Gateway) and maps those jobs onto listing groups and, where the difference in margin justifies it and each campaign would still have enough conversion signal to bid on, onto separate campaigns.
How do you stop Performance Max cannibalising branded search?
Brand exclusions, a dedicated exact-match brand campaign with its own budget, and search terms reporting reconciled regularly. Brand lists can be maintained at account level in Google Ads, but the exclusion itself is applied to the campaigns you select rather than switched on across every PMax campaign automatically, so the applied settings need checking campaign by campaign against Google's current documentation. Without exclusions, PMax can serve on branded queries, which tends to flatter reported ROAS.
When should you consider Standard Shopping alongside Performance Max?
When you need per-product bid control that PMax does not offer: thin catalogues where a handful of SKUs carry the margin, low-volume accounts where PMax has little signal to work with, and catalogues whose margin spread makes any single campaign target wrong for most products. Product suppression is not a reason on its own, because PMax listing groups can exclude specific products or product groups from serving.
How do you prove Performance Max is incremental?
With a designed experiment, such as a geo or audience holdout run over a full purchase cycle, using Google Ads experiments where they apply (support.google.com/google-ads/answer/12997711). Results depend on test design, spend, seasonality and available volume, and small or short tests may not be conclusive. Brand-query splits and new-customer reporting are useful diagnostics, but they describe the mix of demand rather than prove causation.
Why does our PMax ROAS look great while profit is flat?
Two common causes: branded demand included in the reported figures, and revenue-based conversion values that treat a thin-margin order as equal to a high-margin one. Both are visible if you look, using brand exclusions plus channel and search-term reporting for the first, and your own contribution data for the second. Neither is separated for you by default.
Does Google's new gross-profit bidding fix this?
It helps. Native COGS-aware bidding is real progress, but gross profit as Google calculates it does not account for returns, shipping variance or the commercial job of a SKU. Uploading cart or COGS data does not by itself make every bid contribution-based; it improves one input. We use it where it fits and keep our own correction layer.
How much does Performance Max management cost in the UK?
JudeLuxe charges a fixed monthly fee from £2k/month fixed fee, never a percentage of media spend, with a minimum of £15k/month on Google Ads. Because the fee does not move with spend, a recommendation to reduce a PMax budget costs us nothing.
Find out what your Performance Max campaigns are really buying.
We will separate branded from non-branded performance as far as the available reporting allows, rebuild the last 90 days on contribution margin, and show you which campaigns and product groups are funding the account. Yours to keep either way.
Book a commercial review