Branded search campaigns and non-branded search campaigns at a glance
A branded campaign bids on searches that contain the brand's name, its product names or common misspellings of either. The people behind those searches have already decided they want this brand, or at least want to find it. They are cheap to reach because few advertisers compete for them, and they convert well because they arrived intending to. The account structure guide explains why brand terms belong in their own campaign in almost every account; this page takes that separation as given and looks at what follows from it for budgets, targets and testing.
A non-branded campaign bids on everything else: the category, the problem, the product type, the service in a place. The people behind those searches have a need but no settled supplier, so every competitor in the market is bidding for them, clicks cost more and fewer convert. The keyword match types guide covers how those searches are reached. This is where a search programme finds customers who would not otherwise have come.
The difference that matters most is not cost or conversion rate but IncrementalityConversions caused by marketing.. A non-branded click often brings a customer who would otherwise have bought elsewhere. A branded click often brings a customer who would have clicked the organic result just below the ad. Both show up in reports as conversions; only one of them is reliably new.
Compare table
Branded search campaigns vs Non-branded search campaigns
| Row label | Branded search campaigns | Non-branded search campaigns |
|---|---|---|
| Intent | Searchers already know the brand and are often close to buying | Searchers have a need but no chosen supplier |
| Competition | Low, unless competitors bid on the brand's name | High: every competitor in the category bids for the same searches |
| Typical return | High reported return from cheap clicks and high conversion rates | Lower reported return from expensive clicks and lower conversion rates |
| Incrementality | Uncertain: many buyers would have arrived through the organic result | Usually higher: many buyers would otherwise have gone elsewhere |
| Budget role | Protecting demand the brand has already created | Winning new demand and new customers |
| Targets | Impression share or a cost ceiling, or a return target set from a test | A cost or return target set from the economics of a new customer |
| Reporting | Separately, with attributed and tested figures side by side | Separately, with new-customer share where it can be measured |
| Risk when paused | Competitors or aggregators may take the top of the page; some sales may move to organic | Lost reach in the category; demand goes to competitors |
The incrementality and targets rows drive everything else. A campaign whose return is mostly borrowed from the organic result should not set the target for one that has to earn every sale, and blending the two produces a figure that describes neither. It is a case of Goodhart's law: a blended return set as a target rewards moving budget to the campaign that borrows its sales rather than the one that earns them.
When to choose branded search campaigns
Run a branded campaign when competitors bid on the brand's name. If a rival's ad appears above the organic result whenever someone searches for the brand, an ad of the brand's own protects the click it would otherwise lose. This is the clearest case for brand bidding, and a test usually confirms it. Affiliate partners can be in the same auction too, and the guide to setting brand bidding rules for affiliates covers how a programme restricts them. The other side of the same auction, deciding whether to bid on a rival's name yourself, is covered in the guide to bidding on competitor names.
Run one when the brand needs control of the message. A branded ad can carry the current offer, point to the right page and use sitelinks to route people to support, pricing or a specific product, which the organic listing may not do. That control is worth something even where the click itself is not new. Many brand searchers land on the homepage either way, and the guide to the homepage as a conversion page covers routing them from there to the offer, their account or support.
Run one when the organic result is weak or crowded. If the brand's own listing sits below aggregators, review sites or resellers, the ad may be the only way to hold the top of the page.
Keep its targets separate and modest. Many teams run branded campaigns on Impression ShareShare of eligible impressions you received. rather than on a return target, because the job is to appear when the brand is searched, not to chase volume. Whatever the target, never let the branded campaign's reported ROASRevenue generated per unit of ad spend. pull up an account-wide average that then justifies spend elsewhere.
When to choose non-branded search campaigns
Run non-branded campaigns when the business needs customers who do not already know it. Category and problem searches are where people choose a supplier, and the brand only wins them by competing. This is the part of a search programme that grows the customer base rather than serving the existing one.
Set their targets from the economics of a new customer. The cost or return a non-branded campaign is held to should come from what a new customer is worth, not from a blended account figure that includes cheap branded conversions. The comparison of target CPA and target ROAS covers which kind of target fits. Expect the reported return to look weaker than the branded campaign's, because it is doing the harder job.
Judge them on new demand. Where the business can identify first-time customers, report the share of non-branded conversions that were new against the branded share. That comparison usually shows where the search budget is creating customers and where it is collecting them. The comparison of SEO and PPC covers the longer-run question of which of these searches organic content could eventually win.
Decision table
| Situation | Choose | Why |
|---|---|---|
| Competitors bid on the brand's name | Run a branded campaign | The ad protects clicks a rival would otherwise take |
| Brand's organic result is strong and no one else bids on the name | Test branded bidding before funding it fully | The ad may mostly replace free organic clicks |
| Account reports one blended return for all search | Split branded and non-branded with separate targets | The blend hides both campaigns' real figures |
| The business needs new customers | Fund non-branded search on new-customer economics | Category searches are where new buyers choose a supplier |
| Budget is tight and must be cut | Test the branded campaign's incrementality before cutting non-branded | The branded return is the one most likely to be overstated |
| A launch or promotion needs a specific message on brand searches | Run a branded campaign | The ad controls what brand searchers see |
Testing whether brand bidding is incremental
The question a brand-bidding test answers is narrow: when the brand's ads are switched off, how many sales from brand searchers are lost, after the organic result has picked up what it can? Attribution cannot answer it, because it credits the ad for every sale it was present for. The troubleshooting page on retargeting and brand search taking all the credit shows how to spot the inflation; a test measures it.
Design the comparison so it is fair. The cleanest version pauses branded ads in some regions and keeps them running in comparable ones, then compares total sales from brand searchers, paid and organic together, between the two. A time-based pause can work if nothing else changes in the period, but season and promotions make it weaker. The incrementality testing guide covers how to choose regions, size the test and read the result.
Watch the auction during the test. If competitors start bidding on the name in the paused regions, record it: that is part of what branded bidding protects against, and it belongs in the result. Then report the tested figure beside the platform's figure, set the branded campaign's target from the tested one, and retest when competition on the brand name changes.
A linen retailer tests its brand campaign
Fernlace Linen spends £3,000 a month on its branded campaign, which reports £54,000 in attributed revenue, a return of 18. Its non-branded campaigns spend £15,000 for £48,000, a return of 3.2, and the account reports a blended return of about 5.7. Fernlace pauses its branded ads for four weeks in half of its matched regions. In those regions it saves £1,500 of spend, and total revenue from brand searchers, paid and organic together, falls by £6,500 compared with the regions where ads kept running. The branded campaign's incremental return is therefore about 4.3, not 18. A competitor's ad appeared on the brand name in two of the paused regions, which explains part of the loss. Fernlace keeps branded bidding on an impression share strategy with a cost ceiling, stops quoting the blended return, and sets its non-branded target from new-customer value alone.
Common mistakes
- Running branded and non-branded terms in one campaign, so a shared budget and target hide both.
- Quoting a blended search return that borrows the branded campaign's flattering figure.
- Setting non-branded targets from the account average, which includes cheap brand conversions.
- Cutting non-branded search first because its reported return is lower, when it is the part finding new customers.
- Pausing brand ads without a control and reading a seasonal change as the test result.
- Ignoring competitors in the auction when judging whether brand bidding is worth it.
- Treating one brand-bidding test as permanent, when competition on the name changes over time.
