Negatives for job-seeker queries: block 'jobs' and 'salary' or leave them?

Say your B2B software account has 'developer jobs', 'accounting salary' and 'call center careers' eating a few percent of spend. The obvious fix is one shared list of job-seeker negatives across the whole account, but some of those words still reach buyers. 'Salary' shows up in comparison searches like 'payroll software salary calculator', and 'careers' can appear when someone is researching a company before a demo. Block them everywhere and you can quietly cut traffic you actually wanted.

I'd put the unambiguous ones ('jobs', 'hiring', 'resume') in an account-level list and keep 'salary' and 'career' at campaign level, with a weekly look at what they're catching. Is anyone running a single shared list for all of them, and has it cost you anything?

Depends on volume, mostly. If a term like "salary" only fires a handful of times a month, a shared list costs you almost nothing, and a weekly review is just noise. Once the account gets real traffic, splitting the list starts to pay off, because that's when a blocked comparison query actually shows up in the numbers.

Volume is the wrong cut here. A low-volume term can still be expensive if it's the one that leads to a demo, and one blocked "payroll software salary" search is a lost lead. I'd split by what the term means in your category, not by how often it fires.

Meaning should drive the split, and I'd take it a step further: block the phrase a job seeker types, not the bare word. "Salary range", "job openings" and "apply online" almost never come from a buyer. "Salary calculator" can, so a phrase match on the job-intent wording keeps that query alive.

My rule of thumb is simple. If you can't picture a buyer typing the term, it goes on the account-level list. If you can, keep it at campaign level and review it. It's a rough line, but it's easier to apply than judging every word on its own.

The picture test is fine, but I'd check it against data before trusting it. Most of us are good at imagining a job seeker typing "salary range" and bad at imagining a buyer typing "careers" while they vet a vendor. Pull 30 days of search terms containing the candidate word first. If none of the hits look like a buyer, the account-level list is safe. If one or two do, move that word to campaign level. The report settles the argument faster than the thought experiment.