Search
Why 40% of your ROAS is just brand traffic
A high ROAS is often not proof of good work but of a strong brand. How to separate the two cleanly, how much of it is truly incremental, and which test answers the question.
Published on · 8 min read
- 1Searches for your own name convert brilliantly because the purchase decision has already been made. They make the ROAS pretty and the account impossible to steer.
- 2As long as brand and generic run in one campaign or inside Performance Max, the reported ROAS is an average of two worlds that no longer allows any decision.
- 3How much brand revenue you would have got without the ad is something no report tells you. Only a geo test over four to six weeks does.
- 4The answer is rarely zero and rarely a hundred. In most accounts it sits somewhere in between, and that is exactly where the budget you can reallocate is.
In almost every account we open for the first time, a substantial part of paid revenue comes from searches for the brand name itself. These people already know you. They saw your Instagram ad, got a recommendation, saw the parcel at the neighbour. They type your name into Google to buy. And at the very top, above the organic result that is you anyway, sits your ad. They click it because it is the first result. Not because they had to be convinced.
That is not a mistake in the account. It is the default setting of Google Ads, and it is very comfortable for Google: the conversion rate of these clicks is high, the ROAS shines, and the dashboard looks like excellent work. It just does not answer one question: how many of those orders would have come without the ad?
Why reporting hides it
As long as brand searches have their own campaign, at least you can see the number. In most accounts they do not. Brand and generic terms run in the same campaign, or both disappear into Performance Max, where Google decides which query gets which ad. Then a ROAS of 12 blends with a ROAS of 2 into a pleasant average, and the average says nothing anymore.
| Campaign | Media spend | Revenue | ROAS | What the number really says |
|---|---|---|---|---|
| Brand searches | €10,000 | €120,000 | 12.0 | The decision was made, the ad collected it |
| Generic searches | €40,000 | €84,000 | 2.1 | This is where new customers are actually fought for |
| Both mixed | €50,000 | €204,000 | 4.1 | Looks healthy, allows not a single decision |
Invented worked example. The proportions match what we regularly see in mixed accounts.
You can live well with the blended 4.1, and that is exactly the problem. Nobody asks at a ROAS of 4.1 whether the generic campaigns at 2.1 are even above margin. Nobody asks whether the 10,000 euros for brand clicks were necessary. The average is a sedative, and it works on management and agency alike.
12.0
ROAS of the brand campaign in the example
2.1
ROAS of the generic campaign that carries growth
4.1
ROAS the dashboard shows when both run together
How much of it is truly incremental
The decisive measure is incrementality: the share of brand orders that would not have come without the ad. It is almost never zero, because some users click the first result whoever it belongs to, and competitors bid on your name. But it is almost never high either. From our tests and from what stores show us from their own, rough ranges emerge:
- Strong organic position, no competitors on the nameapprox. 15%
- Competitors bid on the name occasionallyapprox. 25%
- Competitors bid permanently, Shopping ads above the organic resultapprox. 40%
- Young brand, organic position not yet establishedapprox. 55%
Rules of thumb, not measurements for your account. The only reliable number comes from your own test.
In the example above, at 25 percent incrementality, that means: of the 120,000 euros of brand revenue, 30,000 are genuinely thanks to the ad. The other 90,000 would have come organically. The true ROAS of the brand campaign is therefore 3, not 12. Still decent, but a completely different conversation than before.
The test that answers the question
There is no formula for it, only an experiment. It is unspectacular, takes a few weeks and answers the question for your account, not for the average of all accounts.
- 1
Split regions
Pick two groups of regions that resemble each other in revenue, season and competition. States or postcode areas work, cities are usually too small. Brand ads keep running in one group and are paused in the other.
- 2
Run for four to six weeks
Shorter is not reliable, because weekdays, promotions and attribution windows distort. Whoever stops after ten days because revenue wobbles in the test group has not run the test, only paid for the unrest.
- 3
Measure total revenue, not ads revenue
The question is not how much the brand campaign loses. The question is how much the store loses in the test region overall, across all channels, including organic and direct. Only that number shows what really fell away.
- 4
Treat the result as a budget decision
If the test region loses 25 percent of brand revenue, that is your incrementality. From it follows a true ROAS for brand, a target the campaign can be run on, and an amount that can move to generic campaigns.
What we build instead
What many accounts look like
- Brand and generic terms in one campaign
- Brand searches also inside Performance Max, without exclusion
- One target ROAS for everything
- Reporting on total ROAS
What it should look like
- Separate campaigns for brand, generic and competitors
- Brand exclusions active in Performance Max
- Own target per campaign, for brand derived from the test
- Reporting on contribution margin without brand
On Performance Max
Since brand exclusions exist for Performance Max, there is no reason left to let brand searches run inside it. Without the exclusion the algorithm takes the cheapest conversions first, and those are your brand searchers. The PMax ROAS then looks great, and the generic reach you actually built the campaign for gets less budget than you think.
The question is not whether brand ads convert. The question is how many of those conversions you would have got without them.
If you do not know your own share: pull everything containing your brand name out of the search terms report for the last 30 days and calculate the ROAS without those rows. That number is the one you are actually working with. If it sits well below margin, it is not the brand campaign carrying the account. It is the brand campaign hiding the problem.