Ecommerce PPC Research
What actually happens when you look beyond ROAS?
Google Ads produces an extraordinary amount of data. Clicks, conversions, revenue, ROAS, CPC, CPA, impression share, auction insights, search terms. All useful.
But ecommerce businesses contain another dataset entirely. Margin. Contribution. COGS. Returns. Stock. Customer type. Lifetime value. Inventory age. Cash.
And when you connect the two, some uncomfortable patterns start appearing.
- The products generating the most revenue aren't always creating the most value.
- The highest-ROAS campaigns aren't always where the next £1 should go.
- Some apparently inefficient advertising is doing exactly what the business needs it to do.
The question we're trying to answer
Most PPC benchmarking asks: what's a good ROAS? We think that's too broad to be useful. A 3.0 ROAS can be exceptional. A 5.0 ROAS can be terrible. A 10.0 ROAS can mean you're dramatically underspending. It depends on the economics underneath it.
So our research starts with a different question: what makes Google Ads investment commercially valuable?
The problem with ecommerce benchmarks
Retailer A
£100,000 monthly spend. £500,000 attributed revenue. 5.0 ROAS. High-margin own-brand products.
Retailer B
£100,000 monthly spend. £500,000 attributed revenue. 5.0 ROAS. Heavily discounted third-party products, expensive fulfilment, high returns.
Benchmark them conventionally and they look identical. A ROAS benchmark without economic context can tell you how you compare while completely missing whether you're making money.
Introducing POAS
ROAS measures revenue ÷ advertising spend. POAS measures profit ÷ advertising spend. For ecommerce businesses, that may require accounting for product cost, discounting, fulfilment, payment fees, shipping contribution, returns and other variable costs.
We're trying to measure the economic value created by advertising, not simply the revenue attributed to it.
There is no universal "good POAS".
Anyone searching for "what is a good POAS?" will probably find somebody willing to give them a number. We won't. A company prioritising short-term contribution, a subscription business with strong lifetime value, a fashion retailer recovering cash from ageing inventory, a high-growth business investing in acquisition and a mature retailer protecting profitability can all rationally accept very different first-order economics.
A useful POAS target comes from your economics and commercial objective. Not an industry-average number downloaded from the internet.
So what should you benchmark?
Not just the headline return. Useful ecommerce PPC benchmarking needs several layers.
01
Contribution
What did advertising actually create?
Revenue is the starting point. Contribution gets closer to what the business receives from the transaction, and lets us compare products, campaigns, categories, markets, customer types and periods using a more commercially meaningful denominator.
02
POAS
How efficiently did advertising create profit?
Two products can generate identical ROAS. The one producing materially more contribution from the same advertising investment deserves to be understood differently.
03
New customer acquisition cost
What does it actually cost to acquire somebody new?
Blended CPA hides this. Existing customers convert more easily. Brand demand converts more easily. If acquisition is the objective, acquisition has to be isolated.
04
Customer value
What happens after the first transaction?
Subscription, repeat purchase, cross-sell, replenishment and customer lifetime can all materially change what a business can rationally afford to pay for a customer.
05
Product economics
Which SKUs actually create the result?
Where contribution is concentrated. Where spend is concentrated. Whether those things align. Which products subsidise others, and where profitable products are constrained.
06
Inventory
What happened to the stock?
Advertising can protect margin, increase inventory velocity, release working capital, prevent markdown or make a stock problem worse. Sometimes the best result appears on the balance sheet before it appears in ROAS.
07
Marginal return
What happened when you spent more?
Average ROAS tells you what existing investment produced. It doesn't tell you whether scaling was sensible. The useful question is what return the next £1 produced.
The JudeLuxe Ecommerce PPC Benchmark
Our aim is to build a commercially useful benchmark for ecommerce paid search. Not "average ecommerce ROAS is 4.37", which would be wonderfully shareable and almost completely useless without context.
Instead, we want to understand the relationships between advertising investment, revenue, contribution, margin, customer acquisition, catalogue size, product concentration, inventory and incremental return.
The questions we want the data to answer
How much Google Ads spend is typically concentrated in a small percentage of products?
Does the 80/20 rule actually hold, or is concentration even greater?
How often do high-revenue SKUs underperform on contribution?
In other words: how frequently does ROAS lead to a different allocation decision than POAS?
How much advertising spend sits against products that never convert?
And importantly, how much of that is actually unjustified? Not every non-converting click is waste.
How much profitable demand is constrained?
Most audits obsess over overspending. We're equally interested in where additional capital could have generated acceptable incremental contribution.
How different are new-customer economics from blended account performance?
How much does repeat and brand demand flatter the apparent cost of acquisition?
How does catalogue size affect spend concentration?
Does automation increasingly favour a narrow group of established products as catalogues grow?
What happens to marginal return as budgets scale?
At what point does additional spend begin creating materially less commercial value, and how does that differ by category, market and business model?
What we won't do: manufacture a headline.
If the research doesn't support a conclusion, we won't publish it.
If the sample is too small, we'll say so.
If a finding only applies to a particular subset, we'll say so.
If correlation doesn't establish causation, we'll say so.
If our original hypothesis turns out to be wrong, even better. That's useful information.
The objective isn't proving JudeLuxe's framework correct. It's understanding what the evidence actually says.
Methodology matters.
Sample
How many accounts, businesses, SKUs and how much advertising spend. Which sectors and which markets.
Period
What date range the analysis covers, and how seasonal effects have been treated.
Definitions
What counts as revenue, contribution, profit, a new customer, advertising spend and a conversion.
Data quality
Where the information came from (Google Ads, Shopify, Merchant Center, client finance data) and how discrepancies were handled.
Limitations
What's missing, what the research can't establish, and where readers should be cautious.
If those things aren't visible, the benchmark isn't particularly useful.
Current research
Ecommerce PPC Benchmarks 2026
Our current view of ecommerce paid search performance, with the sample, definitions and limitations stated up front.
Read the ecommerce ppc benchmarks 2026 researchPOAS Benchmark 2026
How profit on ad spend behaves when product margin, returns and fulfilment are brought into the measurement.
Read the poas benchmark 2026 researchResearch we're working on
These studies are in progress. They'll be published when the underlying analysis genuinely exists, with sample, definitions and limitations attached. A research hub gains authority by publishing evidence, not by giving hypothetical spreadsheets impressive names.
In progress
Ecommerce Google Ads Waste Study
Where advertising spend disappears across large product catalogues. Spend concentration, non-converting products and the difference between apparent waste and commercially unjustified investment.
In progress
ROAS vs POAS
When revenue efficiency and profit efficiency disagree. How product margin changes the apparent winners and losers inside ecommerce Google Ads accounts.
In progress
The SKU Concentration Study
How much of your catalogue actually receives meaningful Google Ads investment, and how Shopping and Performance Max allocate spend across large product ranges.
In progress
The New Customer Gap
How blended ROAS can disguise the real economics of acquisition, and the difference between account-level efficiency and genuinely incremental customer acquisition.
In progress
The Marginal ROAS Study
What happens to commercial return as Google Ads budgets increase, and the point at which additional spend stops creating sufficient incremental value.
What we've already learned from client work
Published benchmark data and individual client evidence aren't the same thing, so we keep them separate. But our client work has repeatedly surfaced patterns worth investigating at greater scale.
Blended ROAS can hide product-level economics.
We've seen accounts where apparently strong aggregate performance contained products producing radically different contribution.
Inventory can completely change the correct advertising decision.
For UKSoccerShop, Google Ads became part of the mechanism for recovering more than £520k from ageing stock.
Read the UKSoccerShop case studyRevenue growth and profit growth can diverge.
For Thermos, rebuilding measurement and allocation around commercial outcomes contributed to a 94% increase in contribution.
Read the Thermos case studyCustomer economics can justify different acquisition decisions.
For Wilsons Pet Food, looking beyond the first transaction supported 112% contribution growth alongside an 83% increase in customer lifetime value.
Read the Wilsons case studyThese are case studies, not universal benchmarks. A 94% contribution increase for one business doesn't mean another should expect 94%. A £520k inventory recovery doesn't establish an industry average. Research asks whether the pattern survives across a larger dataset.
Research should change what we do.
The useful test is whether the evidence changes the decision.
- If spend concentration increases sharply beyond a certain catalogue size, that should influence architecture.
- If POAS and ROAS consistently disagree in certain categories, that should influence measurement.
- If acquisition efficiency is systematically overstated by repeat demand, that should influence bidding.
- If marginal returns deteriorate predictably at particular scaling rates, that should influence budgets.
Research becomes valuable when it changes allocation.
Our hypotheses are allowed to be wrong.
Every SKU has a job. Profit matters beyond revenue. Commercial context should guide automation. Advertising budget should be treated as capital. Those are strong views, and they should still be tested. If evidence challenges something we believe, we change the belief, not the evidence.
Want to contribute data?
For some research projects we may invite ecommerce businesses to contribute anonymised account data. Where we do, we'll explain exactly what information is required, how it will be used, what will be anonymised, what participants receive and what will be published. Client confidentiality comes first: no identifiable client information appears in aggregated research without explicit permission.
The question underneath the research
Where should the next £1 go?
Google Ads gives ecommerce businesses more automation than ever. That makes execution easier. It doesn't make the commercial decision easier. Someone still has to decide which products deserve investment, what a customer is worth, how much efficiency to trade for growth and when additional spend stops making sense.