You have an account with thousands of search terms. The dashboard says ACoS is in range. The campaigns look healthy. But somewhere in there, money is leaking.
You run the bulk sheet through BulkSheet Pro. The Search Terms page surfaces a list of negation recommendations — some of which surprise you. Terms you'd glanced at before. Terms that looked fine in one campaign but were bleeding money in another. Terms you didn't even know were running in multiple campaigns simultaneously.
This is the story of how BulkSheet Pro finds them.
Three different questions about the same term
When BulkSheet Pro analyzes search terms, it doesn't just ask "is this term spending too much?" That question is too simple to be useful at scale.
It asks three different questions in three different layers, and each one catches a different kind of waste.
Question 1: Is this term wasting money in this specific ad group?
The most direct check. For every search term in every ad group, BulkSheet Pro asks two things: has it spent past the patience threshold without producing any orders? Or has it produced orders but at an ACoS so high it's clearly broken?
The patience threshold isn't an arbitrary dollar amount. It's calibrated to your account's actual economics. BulkSheet Pro calculates your account-level cost per order (your total ad spend divided by total orders across all search terms), multiplies it by your target ACoS, and then by seven. The result is the spend ceiling for a zero-order term — the point at which BulkSheet Pro stops giving the term the benefit of the doubt and flags it as waste.
If your account CPO is $20 and your target ACoS is 30%, the patience threshold for a zero-order term is $42. Any term over that without producing orders gets flagged. This scales with you. A premium account with higher CPO has a higher patience threshold. A budget account has a lower one. The math respects your economics.
There's one floor that protects new or sparse accounts from over-eager negation. If your account is so new that the computed patience would drop below $5 — for instance, if you've barely run any orders and the math says the per-order patience is $0.80 — BulkSheet Pro clamps it at $5 anyway. That guards against the edge case where a small accountCPO produces an unrealistically low threshold and starts flagging terms that haven't really earned the negate. The floor only matters for the smallest accounts; once your data is meaningful, the floor never binds.
For terms with orders, the rule is simpler: if ACoS is more than four times target, it's broken — flag it regardless of confidence.
Question 2: Is this term wasting money across the account?
A term might look reasonable in one campaign — one or two orders, ACoS only slightly over target — and look reasonable in another, and reasonable in a third. But add all three up and the picture changes. You're paying for the same bad term from three budgets simultaneously, and it's only acceptable in each one because each one is too small to flag in isolation.
BulkSheet Pro aggregates each term across all the campaigns where it appears. If the term shows up in at least two distinct campaigns and the aggregate spend has crossed the patience threshold without orders (or aggregate ACoS is broken with orders), it's flagged for cross-campaign negation. Different surface, same logic.
Question 3: Is this term winning in one ad group and losing in another?
This is the most interesting layer.
When the same search term appears across multiple ad groups, BulkSheet Pro looks at each one independently and asks: who's winning with this term? Winner is defined as the ad group with the most orders, tiebroken by lowest ACoS. The reasoning: that's the ad group where the audience is right — where the SKU, the keyword targeting, and the ad copy are actually connecting with shoppers searching for this term.
Then BulkSheet Pro looks at the other ad groups running the same term. For any one of them to qualify as a loser, it has to fail both of two tests:
- It has to be performing significantly worse than the winner (its ACoS more than 1.5× the winner's ACoS)
- It has to be clearly failing the account's own bar (ACoS more than 2× target)
Both gates have to pass. Just being worse than the winner isn't enough. Just being above target isn't enough. The loser has to be visibly worse than where the term is succeeding and it has to be unacceptable on its own terms.
If both pass: BulkSheet Pro recommends negating the term in the loser's ad group while leaving the winner alone. You don't lose the term. You just stop paying for it where it doesn't work.
For zero-order losers, only the first test applies — but using the patience threshold instead of a comparison to the winner. A zero-order loser that's burned past the patience budget is qualified regardless of how the winner is doing.
What BulkSheet Pro won't do
It won't act on noise. The high-confidence threshold for negation requires at least 10 clicks. Below that, the recommendation may still appear, but flagged LOW, and it won't show up in the default Rec Actions filter. You can widen the filter to see them, but BulkSheet Pro won't pre-select negations on traffic too thin to support a real judgment.
It won't double-flag. If you've already negated a term as exact in an ad group, BulkSheet Pro skips it on the next pass. If you have a phrase negative that covers the term, it skips that too. The same term won't appear in multiple negation recommendations across the layers — BulkSheet Pro deconflicts them through a unified queue check before showing anything to you.
It won't break your winners. The whole point of the winner-vs-loser logic is that the right place to keep a term is in the ad group that's already converting on it. Negating a term broadly would kill the goose. Negating selectively keeps your conversions intact.
And it won't pretend the three questions are the same question. Each layer of detection is a separate evaluation with its own confidence scoring. You see them as three separate panels on the Search Terms page, not one collapsed list. You're the one looking at all three and deciding what to ship.
The export
You hit Export. The bulk sheet now has 117 new negative-exact rows — some pure waste (zero orders, past patience), some bleeding losers in specific ad groups (winners untouched), some aggregate offenders running across multiple campaigns. Each one was qualified by its own evidence.
The month that follows isn't quieter than the last one in terms of activity. It's quieter in terms of spend on the wrong things. Account-level ACoS comes down. Total spend comes down. Orders hold. The campaigns are doing the same work — just on the terms that earn their keep.
The point isn't that BulkSheet Pro found 117 things you missed. It's that BulkSheet Pro asked three different questions at three different layers, and every flagged term had to pass the gates for at least one of them. Nothing was a guess.