Case studyDTC ecommerce90 days
A 4.0x blended ROAS was hiding a non-brand account barely breaking even
We split brand from non-brand, found the real acquisition number, then doubled it
- Blended ROAS (before)
- 4.0x
- True non-brand ROAS
- 2.4x → 2.5x
- Non-brand spend
- $45K → $92K
- New-customer revenue
- +96%
The lie inside a blended number
A blended ROAS averages your cheapest clicks with your most expensive ones. It tells you almost nothing about whether you're growing.
When we picked up this account, the headline was a 4.0x blended ROAS and a team that felt good about it. The problem with a blended number is that it mixes two completely different jobs into one figure. Branded search, where someone already typed the company name, converts cheap and high. Non-brand, where you're actually buying a stranger, costs more and returns less. Average them and you get a number that hides the only thing that matters: is the part that grows the business working?
We pulled the two apart. Brand search was running at roughly 14x on $7K a month. Non-brand and the catalogue inside Performance Max were running at 2.4x on $45K, and Performance Max was quietly serving on branded queries and claiming the credit. So the real acquisition engine was at 2.4x, the flattering 14x was mostly demand that already existed, and nobody could see it.
What the split exposed
Before: one blurry number
- Brand and non-brand sharing campaigns and budget
- Performance Max serving on branded queries and taking the credit
- A 4.0x blended ROAS nobody could break down
- Non-brand quietly capped because it looked inefficient next to brand
After: two systems you can steer
- Brand isolated with its own budget and an impression-share target
- Performance Max brand-excluded at the account and campaign level
- Non-brand tiered by intent with margin-based targets
- Non-brand budget doubled once the real ROAS was finally visible
The build: five campaigns, no overlap
This is the structure we deploy. Copy it. The whole point is that each campaign has one job and clean signal.
The campaign map we built
| Campaign | Role | What it targets | Bid strategy |
|---|---|---|---|
| Brand Search | Protect and harvest existing demand | Exact + phrase on brand terms only | Target impression share, capped CPC |
| Non-brand Search (high intent) | Acquire ready-to-buy buyers | "[product] best / buy / price", exact + phrase | tROAS set off break-even + margin |
| Non-brand Search (research) | Acquire early-stage demand | "how / which / vs" phrase + broad with audiences | Max conversions, then tROAS once it learns |
| Performance Max (brand excluded) | Scale Shopping + remarketing | Full catalogue, brand term list excluded | tROAS, account-level brand exclusion on |
| Brand defence (optional) | Block competitors bidding on your name | Conquesting protection on brand SERP | Manual CPC, tight budget |
No overlap, no mixed signals, no "everything" campaign. Each one feeds the bidder a clean pattern to learn from.
How we stopped PMax stealing the branded credit
01
Turned on the account-level brand exclusion list
This stops Performance Max from serving on your own brand terms at all. Without it, PMax claims your easiest conversions, reports a flattering ROAS, and the algorithm pours budget into it because it looks like your best campaign. It isn't. It's eating demand you already owned.
02
Added a brand-term negative list to non-brand Search
Keeps research and high-intent non-brand campaigns from drifting onto branded queries. Now each campaign only competes for the demand it's supposed to create, and the reporting tells the truth.
03
Measured new-customer ROAS, not blended ROAS
Switched the scorecard to Google's new-customer acquisition reporting plus GA4 new-versus-returning. That's the number that tells you if you're growing, instead of how cheaply you can re-buy people who'd have come back anyway.
What happened over 90 days
We didn't chase a bigger blended number. We chased real customers.
Before and after
| Metric | Before | After (day 90) |
|---|---|---|
| Non-brand spend / month | $45,000 | $92,000 |
| Non-brand ROAS | 2.4x | 2.5x |
| Brand spend / month | $7,000 | $5,000 |
| Brand ROAS | 14x | 14x |
| New-customer revenue | Baseline | +96% |
| Blended ROAS | 4.0x | 3.1x |
Blended ROAS dropped on purpose. We traded a vanity number for acquisition that actually scales.
“We thought the account was crushing it at 4x. Turns out we were paying to win clicks from people already typing our name. The first real growth came the month after we stopped flattering ourselves.”
Questions we get asked about this
Should I even bid on my own brand name?
Usually yes, but as its own campaign with its own budget and a target impression share, never lumped in with acquisition. The reason is measurement as much as protection. When brand sits inside the same campaign as non-brand, its 14x averages up the whole number and hides whether your real acquisition works. Separate it and you can judge brand spend on its own merits: competitor pressure, organic overlap, and margin.
How do I stop Performance Max from eating my branded search?
Two layers. Turn on the account-level brand exclusion list so PMax stops serving on your brand terms, and add a brand-term negative list to the campaign. Without this, PMax claims branded conversions, reports a flattering ROAS, and starves your non-brand campaigns of budget because it looks more efficient than it really is.
What's a healthy non-brand ROAS?
It depends on your margin, not on a universal number. Work out your break-even ROAS (1 divided by gross margin) and treat anything above it as profitable acquisition. A 2.4x non-brand ROAS on a 55% margin product is a real, scalable business. Chasing a 5x non-brand number usually just means you're leaving most of your growth unbought.
Won't my blended ROAS drop if I scale non-brand?
Yes, and that's fine. Blended ROAS falls because you're buying more new customers and leaning less on cheap branded clicks. The number that matters is whether non-brand stays above break-even as it scales. Ours held at 2.5x while spend doubled.
Frameworks behind this case study
Where these numbers come from
Illustrative example. The campaign structure shown is the exact build we deploy. The figures are representative of what it produces for accounts in this spend band over a 90-day window, not an export from one client's account.
More breakdowns on the daily newsletterWant this structure on your account?
Free 30-minute audit. No deck. No pitch.
We screenshare your Google Ads, GA4 and Shopify, run the same audit playbook this account came out of, and leave you the document whether you become a client or not.
Book the call