Your branded search campaign is returning 18x.
Prospecting sits at 2.8x.
Every quarter, in some meeting, someone looks at those two numbers and says the obvious thing: put more money in the 18x.
I understand the instinct. I also think it is one of the most expensive mistakes in ecommerce paid media, and it is worth understanding exactly why.
Start with who is doing the searching
A branded campaign bids on your own name. Your brand, your product names, the misspellings.
Think about who types that into Google.
Not someone discovering you. Someone who already knows you exist and is trying to get to you. They saw the Instagram ad three weeks ago. A friend sent them a link. They read a review, or they visited last month and did not buy, and now they are back.
By the time your name is in the search bar, the interesting part of the decision has already happened somewhere else.
Why the number looks so good
Ad platforms hand conversion credit to the last interaction before the purchase. And for someone who has already decided to buy from you, that last interaction is very often a branded search.
So the branded campaign collects credit for customers that other things created.
It did not generate 18x of new demand. It stood at the door of a store people were already walking into and charged a small toll on the way in.
That is the whole trick. Branded ROAS measures how cheaply you can collect demand that already exists. Prospecting ROAS measures something far harder: how expensively you can manufacture demand that did not exist at all.
Two different jobs. Two numbers that look identical in a spreadsheet column.
Line them up as though they are comparable and the money flows toward the easy one. The dashboard improves. New customer acquisition quietly starves. Six months later nobody can work out why growth stalled while ROAS went up.
The question that reframes it
If you paused brand tomorrow, how many of those sales would you actually lose?
Sit with that for a second, because it is uncomfortable.
Some would go. But a real share of people who search your name will scroll past the empty ad slot, click the organic result sitting right underneath it, and buy anyway. You were paying to be one line higher on a page where you already appear.
Those sales are not incremental. They were attributed to advertising. They were not caused by it.
How big that share is depends entirely on your brand and your results page. Which is precisely why guessing is useless and testing is not.
Where brand bidding earns its place
I am not arguing you should switch brand campaigns off. I have seen that go badly.
There are real reasons to bid on your own name:
Competitors are bidding on it. If you vacate the slot, someone else takes it, and some of your demand walks to them. Defense costs money and sometimes it is worth every cent.
You control what the ad says. Your current promotion, your sitelinks, your message. The organic listing shows whatever Google feels like showing that day.
Brand clicks are cheap. Usually a fraction of non-brand. Even at low incrementality, insurance at that price can make sense.
Your organic listing might be weak. If marketplaces, resellers, or review sites outrank you for your own name, that ad is doing genuine work.
Those conversions feed the rest of the account. A purchase is a purchase as far as the platform is concerned. Brand conversions land in your pixel and your conversion data alongside everything else, so they inform Meta's delivery and lookalike seeds, and they feed Google's Smart Bidding beyond the campaign that produced them. Cheap conversions arriving steadily do help the machine learn.
With one warning attached, because this argument can be pushed too far. Brand buyers are not like cold prospects. They arrived already knowing who you are. Let them dominate the seed audience and you can end up teaching your prospecting campaigns to go find more people who already know you, which is precisely the job prospecting is not supposed to do. If you lean on this benefit, build your lookalike and value-based seeds from segments that reflect the customers you want more of, not simply from whoever converted most cheaply last month.
None of that is in dispute. My argument is narrower: the 18x is not evidence for any of it. Those are strategic reasons, and they should be argued on their own terms rather than hidden behind a number that was always going to look good.
How to actually evaluate it
Split brand and non-brand properly. If they share campaigns, every number you have is blended and none of it can be compared. Separate them, use brand exclusions in Performance Max where you can, and report them apart from each other every single time.
Compare share of spend against share of return. If brand takes 5 percent of budget and claims 30 percent of revenue, that is not a growth engine. That is an attribution magnet.
Look at new versus returning customers. Brand skews heavily toward people who already bought from you or were already deep in the funnel. Filter to new customers only and brand campaigns tend to look very different.
Test it, if the spend justifies the effort. Hold brand ads back in some regions and not others, or alternate weeks, and watch what happens to total sales rather than attributed sales. It takes discipline and enough volume to read a real result. For anyone spending seriously on brand terms, doing this once a year is cheap insurance against a very expensive assumption.
Know how much of your blended number is brand. An account at 4.5x where half the revenue comes from brand at 18x is a completely different business from an account at 4.5x with no brand spend. Same headline. Nothing else in common.
Not sure how much of your return is brand? That split is one of the first things our free health check reports back. You send two exports, one from Google Ads and one from Meta, and we show you where your spend and your return actually line up across both platforms. No account access, no signup, no card.
See what the health check coversWhich number answers which question
| What you want to know | Where to look |
|---|---|
| How cheaply do we collect demand that exists? | Branded search ROAS |
| How well do we create demand that does not? | Non-brand ROAS, new-customer ROAS |
| Is the advertising efficient overall? | Blended ROAS across all spend |
| Would these sales happen without us? | An incrementality test, not ROAS at all |
The bigger pattern
Branded search is just the cleanest example of something that runs through every ad account I have ever opened.
Every number a platform reports is an answer. The platform almost never tells you what the question was.
An 18x that means "we tax our own existing demand efficiently" sits in the same column, same font, same dashboard, as a 2.8x that means "we are acquiring new customers at a workable cost." Read as the same metric, they will point your budget in exactly the wrong direction, and they will do it while every chart is green.
It is the same problem I wrote about in why Meta and Google both claim the same sale, arriving from a different direction. Platforms answer their own questions. Nobody reads across them.
That is the class of problem we built Coretas for. It puts Google and Meta on one consistent basis, keeps the demand you capture separate from the demand you create, and turns the result into recommendations you review before anything changes. Not a prettier dashboard. Budget decisions made on numbers that answer the question you actually asked.
See the split in your own accounts. Send us two exports and we will show you how your spend and return divide across Google and Meta, what is capturing existing demand, and which tracking issues are distorting the picture. No account access, no signup, no card. Fill this in and we will email you the exact export steps today.
Request received.
Frequently asked questions
What is branded search?
Searches for your company or product by name. Branded search campaigns bid on those terms so your ad sits above or alongside the organic listing you already have.
Why is my branded search ROAS so high?
Because last-click attribution gives credit to the final interaction before a purchase, and for people who have already decided to buy, that final interaction is usually a search for your brand. The campaign is collecting credit for demand that something else created.
Should I bid on my own brand name?
It depends on the situation, and the ROAS figure will not tell you. It is usually justified when competitors bid on your name, when your organic listing is buried under marketplaces or resellers, or when you need control over the message. It is much harder to justify when you already dominate your own results page and nobody is bidding against you.
Does branded search cannibalize organic traffic?
Some of it, almost always. A portion of brand ad clicks would have been free organic clicks. How large a portion varies enormously by brand and by how crowded the results page is, which is the argument for testing rather than assuming.
What is incrementality testing?
Withholding advertising from a control group, usually regions or time periods, then comparing total sales against a group that kept running. It answers whether the ads caused the sales. Attribution only distributes credit among interactions it happened to observe, which is a different question.
How do I separate brand and non-brand performance?
Keep them in separate campaigns, add brand terms as negatives in your non-brand campaigns so they cannot bleed across, use brand exclusions in Performance Max, and report the two groups separately in every review.