Brand search terms: the four-bucket method.
Open a brand campaign, pull the search-terms report for the last few weeks, and you rarely see just your brand name. You see brandname login. You see brandname youtube. You see a student who wants your software for free, a typo, someone typing brandname download. One column, all mixed together, and most people do one of two things with it: leave everything running, or switch the whole brand campaign off because “you already win on your own name anyway”. Both lazy.
We sort every search term into one of four buckets. Only once a term is in a bucket do we know what to do with it.
Bucket A — exclude from everything. Terms that will never convert on any campaign type. A student looking for a free licence. youtube. Old OS versions. The whole login cluster: log in, sign in, admin. These aren’t leads, they’re people who want something other than to buy. We don’t set them as a campaign negative, we put them in a shared exclusion list attached to every campaign, including the ones that don’t exist yet. Done once, blocked forever.
Bucket B — exclude from Brand only. Almost always empty. This bucket is for the rare term you do want a future generic campaign to catch, but not your brand campaign. In practice a term that’s bad for your brand is usually bad for the rest too. Reserve it for the genuine edge case, and be suspicious if it fills up.
Bucket C — promote to exact. A phrase match catching a repeating pattern earns its own exact keyword. Better quality score, lower cost per click, and you tell Google which variant you actually want to match. This is where the misspellings of your brand live, the reversed word order, the modifiers (brandname pdf, brandname demo), and the informational query with commercial intent (what is brandname). Our threshold: at least five impressions in the window, a clear pattern, and not a one-off.
Bucket D — its own ad group. Sometimes a group of terms clusters around one clear intent with enough mass behind it. The most common is download-or-trial: people typing your brand plus download or free trial. They don’t belong among your regular brand ads. They get their own ad group, their own landing page (the trial page, not the homepage), and ad copy written for that one moment.
One thing we’re strict about: account and login terms never get their own group, however much volume they carry. Someone searching brandname login is already a customer. Paying for that is buying existing revenue again, not new revenue. That cluster goes straight to bucket A.
The order you push these changes matters too. If you route a download cluster into its own group and at the same time set download as a negative on your existing exact group, that negative can block your own keywords if you’re not careful. Route before you exclude, or you build a wall straight through your own campaign.
This isn’t a yes-or-no on brand bidding. It’s a method for seeing what’s actually coming in on your own name, and treating it separately. We run the same sorting in an audit by default, alongside the other things that usually go wrong on an account: what a free audit finds every time. It takes us half an hour, and it’s free.