Why Google Ads ROAS Can Be Misleading for Ecommerce Brands
Google Ads ROAS becomes misleading when it tells you how much revenue Google claimed, but not how much revenue Google actually created.
That distinction matters more than most ecommerce brands realise.
I’ve spent around seven years inside Google Ads. Across that time, I’ve analysed more than 1,000 accounts and we’ve generated over $100 million for clients. One of the most consistent problems I see is businesses treating ROAS as the final score for an account.
A 6x ROAS looks better than a 3x ROAS. That part is easy.
But what if most of the 6x came from people already searching your brand name?
What if Meta created the customer, Google captured the branded search at the end, and both platforms took credit?
What if two purchase events are accidentally recording the same order?
What if the 3x campaign is acquiring completely new customers while the 6x campaign is capturing people who already know you?
Those are not edge cases. I see variations of them constantly.
ROAS is useful. We use it every day.
I just don’t trust it until I understand where the revenue came from, how the conversion was tracked and what role the campaign actually played in creating the sale.
TL;DR
ROAS tells you attributed revenue divided by ad spend. It does not automatically tell you profitability, incrementality or new customer growth.
The three biggest reasons I see ecommerce ROAS become misleading are:
Branded and non-branded traffic are mixed together.
Conversion tracking is inflating or missing revenue.
Different campaign types are being judged against the same ROAS expectation.
A branded Search campaign and a cold YouTube campaign do completely different jobs. Comparing them purely by ROAS is the wrong way to read an account.
Our goal is not to manufacture the highest ROAS possible.
Our goal is to use Google Ads to generate profitable new customers and incremental revenue, then scale as far as we can while staying inside the business’s real commercial targets.
Why can a high Google Ads ROAS still represent weak growth?
A high ROAS can represent weak growth when the account is heavily weighted toward customers who already know the brand.
This is the biggest attribution problem I see in ecommerce.
Branded traffic is someone searching directly for your business. If your company is called Peter’s Pans and somebody searches “Peter’s Pans”, that person is very different from somebody searching “best frying pan” or “10-inch aluminium frying pan”.
The branded customer already knows who you are.
They have probably seen the product before. They may have come through Meta, email, organic content, word of mouth or a previous website visit. Their purchase intent is extremely high.
In the accounts I’ve analysed, branded searches can convert at dramatically higher rates than cold traffic. I’ve seen branded conversion rates sit 10 to 20 times higher than new-customer traffic.
That naturally produces an excellent ROAS.
The problem starts when the account treats that ROAS as proof that Google created the customer.
Imagine you spend $100 on Meta to introduce someone to your brand.
They watch the ad, visit the site and decide they like the product, but they don’t purchase immediately.
Two days later they search your brand on Google.
Your branded Search ad appears. They click it and spend $200.
Google now has a conversion.
Inside Google Ads, that sale looks fantastic. The click was cheap, the customer had high intent and the resulting ROAS can be enormous.
But Google did not necessarily create that $200 of demand.
Meta did a large part of the work. Google captured the customer at the end of the journey.
I’ve explained this to prospects as effectively paying twice. You have already paid to acquire the customer through Meta or another channel, then you pay again when they search the brand, and Google takes credit for the final conversion.
That does not mean I turn branded campaigns off.
I disagree with that advice as well.
If someone searches your brand and a competitor is bidding on your name, they can appear above you. I want our clients protecting that traffic. Branded campaigns are also cheap and provide useful conversion data back to Google.
The mistake is confusing branded protection with new customer acquisition.
In most of our client accounts, we keep branded activity as a small, clearly separated part of the account. Normally we typically aim for branded traffic to account for around 10–20% of total Google Ads spend.
Once brand starts dominating spend, I become very sceptical of the headline ROAS.
At that point, we are not necessarily scaling.
We may just be capturing demand the business already created.
Performance Max makes this even easier to hide
PMax is particularly good at finding easy conversions.
That is one of its strengths and one of its biggest weaknesses.
If we give Performance Max access to branded traffic without strong controls, Google has an obvious incentive: find the people most likely to convert and hit the campaign target.
Someone searching your brand name is an easy win.
A previous visitor is an easy win.
An existing customer is an easy win.
The dashboard can suddenly show an 8x ROAS and everyone is happy.
But if most of that 8x comes from people already deep in the buying journey, the number tells us very little about how well PMax is finding new customers.
This is why we separate branded traffic and apply brand exclusions to cold campaigns.
I want to know what PMax, Shopping and Search can do without being allowed to hide behind the easiest customers in the account.
One of the clearest examples we have is an ecommerce brand we scaled from $0 to $369,000 in Google Ads revenue in under 90 days.
When I broke the account down, 87% of the revenue came from non-branded terms and only 13% came from branded campaigns.
That is a far more meaningful result to me than simply saying the account had a good ROAS.
The brand started virtually from scratch on Google. There was very little existing branded search demand to exploit.
The account had to find new customers.
At the 60-day mark it had generated $266,000, with more than $100,000 per month being added through Google. By the time we passed $369,000, the overwhelming majority of that revenue was still non-branded.
That is what I want to see when somebody tells me Google is scaling an ecommerce business.
Not just a high number in the ROAS column.
Who bought? Were they new? What were they searching for? Would that revenue have existed without the campaign?
Those questions matter more.
How can conversion tracking make ROAS completely wrong?
ROAS is meaningless if the conversion data underneath it is wrong.
This sounds obvious, but broken tracking is one of the most common problems I find when auditing Google Ads accounts.
I have seen:
purchase events missing entirely, add-to-cart events set as primary conversions, begin-checkout events treated like sales, static ecommerce conversion values, duplicate purchase events and accounts where different tracking systems report completely different revenue.
The most extreme version is also one of the simplest.
If an account has two purchase conversion actions and both are set to primary, Google can count the same order twice.
Your apparent 5x ROAS can literally be a 2.5x ROAS.
Nothing about the campaigns changed. The business did not suddenly become less profitable.
The dashboard was wrong.
That is why I refuse to judge an account by ROAS before we understand the tracking setup.
The problem is bigger than reporting.
Bad tracking also changes how Google optimises.
If we tell Google that an add to cart is just as important as a completed purchase, Google will look for people who are good at adding products to their cart.
That is not what I want.
I want purchasers.
For ecommerce, I want the primary conversion action to represent the revenue-producing event and I want the actual transaction value passed back with it.
A $50 purchase and a $500 purchase should not look identical to the algorithm.
The customer who spends $500 is commercially more valuable. Google needs that information if we want value-based bidding to make intelligent decisions.
Tracking errors can therefore make ROAS misleading in both directions.
They can inflate results through duplicate conversions.
They can understate results by missing purchases.
And they can quietly reduce future performance because Google is learning from the wrong customers.
I saw another account where two systems were both supposed to record purchases. One had captured around $12,000 in conversion value and 847 purchases. The other had captured roughly $7,800 and 582 purchases.
Those are not close enough for me to ignore.
One setup was losing data.
If we had trusted the weaker event alone, we would have underestimated revenue and fed fewer purchasing customers back into Google’s algorithm.
This is why we like having a primary event plus backup or contingency tracking. The secondary data gives us something to compare against rather than blindly trusting one number.
A real ROAS improvement should change the economics, not just the dashboard
ROAS itself is not the enemy.
When it improves for the right reasons, the commercial impact can be massive.
We worked on one account that was sitting at roughly 0.93 ROAS.
The business was spending about $54,000 to generate roughly $50,000.
After restructuring targeting, Merchant Center, campaign separation and budget allocation, spend came down to roughly $15,000 while revenue increased to $63,900.
ROAS moved to 4.26, a 332% increase.
I care about that increase because the underlying economics changed.
We were not simply manipulating attribution to make the dashboard look nice.
The business spent substantially less and generated more revenue.
That is a real improvement.
If ROAS rises because we cut wasted traffic, improve targeting, prioritise the right products and generate more revenue from less spend, great.
If ROAS rises because PMax found more branded customers or because tracking started counting conversions twice, I do not care how attractive the number looks.
Why shouldn’t every campaign be judged by the same ROAS?
Every campaign should not be judged by the same ROAS because different campaigns sit at different stages of the customer journey.
This is one of the biggest mistakes I see once ecommerce brands move beyond basic Search and Shopping.
A branded Search campaign is targeting somebody who already knows the business.
A non-branded Shopping campaign is targeting somebody looking for a product but who has not necessarily chosen a brand.
A YouTube campaign can reach somebody who was not even thinking about buying the product until they saw the ad.
Those customers are not equally difficult to acquire.
So I do not expect the same immediate ROAS from each campaign.
This is particularly important with YouTube and Demand Gen.
For many ecommerce brands, my goal with YouTube is not to achieve the highest last-click return in the account.
I want to generate profitable exposure.
Someone sees the video, learns about the product and does not buy.
Later, they search the product category.
Or they search the brand.
Then Search or Shopping receives the conversion.
If we judge the YouTube campaign entirely by the revenue Google directly attributes back to that first impression or click, we can understate the role it played in creating demand.
This is the opposite problem to branded Search.
Branded Search can receive too much credit because it sits close to the final purchase.
Top-of-funnel activity can receive too little credit because it sits further away.
That is why channel context matters.
In one of our ecommerce accounts, we built Search, Shopping and PMax first. Once those campaigns started reaching diminishing returns, we introduced Demand Gen.
The direct ROAS on Demand Gen started lower.
We expected that, as it was reaching colder customers.
Within two to three weeks, the campaigns became profitable and continued improving as Google gathered more data. The purpose was not to beat branded Search on ROAS. It was to bring more people into the ecosystem so the rest of the account had additional demand to capture.
If I had killed those campaigns because their initial ROAS was lower than branded Search, I would have been comparing two completely different jobs.
There is no universal “good ROAS”
I also disagree with using one ROAS benchmark across every business.
I have clients where a 10x makes sense.
I have clients where a 2x makes sense.
The number depends on the economics.
In our sales conversations, I commonly explain that ecommerce businesses can operate around a 2–4x ROAS depending on their margins, AOV and customer economics, while service businesses can sit materially higher because lifetime value and gross margins are completely different.
Even among two ecommerce brands, the same ROAS can mean completely different things.
A brand with strong gross margins, high repeat purchase behaviour and a high lifetime value can afford to acquire a customer at a lower first-purchase ROAS.
A low-margin business with very little repeat purchasing cannot.
That is why one of the first conversations I have with a brand is about the economics.
What are your margins?
What is your AOV?
What can you afford to pay for a customer?
Where is break-even?
I need those answers before I can tell you whether a 3x ROAS is good.
Without them, ROAS is just a number.
What do we look at instead of relying on ROAS alone?
We still look at ROAS. We just put it inside a wider commercial picture.
My priority is understanding whether Google is producing new sales, new customers and incremental revenue.
That is the philosophy we keep coming back to at Kraken.
I would rather have an account at a sustainable 4x that is consistently finding new customers than manufacture an 8x by pouring the budget into branded Search and remarketing.
The second account looks better in a screenshot.
The first one builds the business.
When I assess an ecommerce account, I want to understand four things together:
How much revenue is genuinely non-branded?
If most of the growth comes from people who were not already searching the company, Google is doing a real acquisition job.
What does a new customer cost?
Blended CPA can hide a lot. I want to understand what we are actually paying to bring somebody new into the brand.
What are the economics behind that customer?
AOV, margins and the brand’s acceptable acquisition cost determine whether we can keep scaling.
Can the account maintain those economics as spend increases?
A campaign that achieves 8x ROAS on $100 per day but collapses at $500 per day is not more valuable than a campaign that can sustainably operate at 4x while absorbing thousands per day.
Scale changes the discussion.
This is why we see a lot of brands arrive around the $10,000-per-month Google Ads mark.
That level is a fairly predictable plateau in the accounts I see.
The easy traffic has been captured. The account now has to move further into non-branded acquisition, broader demand and more sophisticated campaign structures.
Some businesses have a lower natural ceiling. Others can go dramatically further.
Our job is to find the real ceiling, not mistake a poorly structured account for the limit of the market.
That requires ROAS to be read alongside the account structure.
If the ROAS starts falling as spend increases, I want to know why.
Are we entering genuinely new audiences?
Is branded contribution falling because more budget is moving toward acquisition?
Is new customer revenue rising?
Are margins still acceptable?
Or are we simply wasting money?
Those scenarios can produce similar headline ROAS changes while telling completely different stories.
When do I trust Google Ads ROAS?
I trust ROAS after three things are true.
First, the conversion tracking is accurate enough that I believe the revenue data.
Second, branded and non-branded traffic are separated enough that I know whether Google is capturing demand or creating new customer growth.
Third, I understand the economics of the business well enough to know what ROAS we actually need.
Once those foundations are in place, ROAS becomes extremely useful.
We can use it as a scaling KPI.
If we are comfortably above the target and the recent data remains strong, we increase spend.
If performance falls below the point the business can tolerate, we hold or pull back.
Then we continue pushing until we find the real point of diminishing returns.
That is a very different approach from trying to maximise ROAS at all costs.
Maximising ROAS is easy if I am allowed to shrink the account down to the highest-intent branded customers.
I could make plenty of accounts look incredible that way.
But the business would stop growing.
Our recommendation
Do not ask your agency, freelancer or internal team only:
“What is our ROAS?”
Ask:
“Where is that ROAS coming from?”
If the answer is heavily branded, separate the traffic.
If the purchase data looks suspicious, audit the conversion tracking before making campaign decisions.
If a top-of-funnel campaign has a lower ROAS, look at the role it plays in creating future Search, Shopping and branded demand before killing it.
And before deciding that a 2x, 4x or 8x ROAS is good, work backwards from your actual business economics.
I want to know the margin, AOV, acceptable acquisition cost and new customer performance.
Then ROAS becomes useful.
Without that context, it is one of the easiest numbers in Google Ads to misunderstand.
Final takeaway
A high Google Ads ROAS does not automatically mean Google Ads is growing your ecommerce brand.
It can mean the campaigns are excellent.
It can also mean Google is taking credit for customers generated elsewhere, PMax is leaning into branded demand or the conversion tracking is simply wrong.
After years of working inside these accounts, my view is straightforward:
I care less about getting the highest ROAS possible and more about understanding whether we are creating profitable incremental growth.
Protect branded demand, but separate it.
Track purchases accurately.
Judge campaigns according to the job they perform.
Know the economics of your customer before setting a target.
Then use ROAS for what it is good at: helping you decide how far you can scale.
Not telling you the entire story of the business.
Frequently asked questions
Is a high Google Ads ROAS always good?
No. A high ROAS is only useful when the underlying conversion data is accurate and you understand where the revenue came from. Branded traffic, returning customers and duplicate conversion actions can all make reported ROAS look stronger than the incremental growth Google actually produced.
What is a good ROAS for an ecommerce brand?
There is no universal target. In our experience, ecommerce accounts can operate around 2–4x depending on margins, AOV and customer economics. I set the target from the business’s break-even point rather than using an arbitrary industry benchmark.
Why do we separate branded and non-branded Google Ads traffic?
Because they represent different levels of customer intent. Branded customers already know the business and convert much more easily. Separating them lets us see what we are paying to protect existing demand versus what we are paying to acquire genuinely new customers.
Written by Max Crakanthorp, Founder of Kraken Digital