The bid mistake ACOS can’t see

A while back I audited an account where the seller was doing most things right. Bulk files, regular bid reviews, a target ACOS, rules applied consistently. And still, when I totaled it up, roughly 12% of ad spend was leaking to search terms that had never produced a sale — and some of those terms were getting bid increases.

Not because the seller was careless. Because the logic most people use to manage bids is built on ACOS — and ACOS has a blind spot exactly where the money leaks.

ACOS can't see a row with no sales

ACOS is ad spend divided by ad sales. A search term with 15 clicks, $11 of spend, and zero orders has no ACOS at all — the math divides by zero, the cell sits empty, and any rule built on “ACOS above target → cut, below target → raise” simply skips the row. The worst performers in the account are invisible to the one metric everyone manages by.

Worse: a common habit is finishing a bid review with a default — “+1% on everything else.” That quietly raises bids on every zero-sale term the ACOS rules just skipped. The account leaks, and the leak gets a raise.

The rules I run instead

• The zero-sale rule. Meaningful clicks (my threshold: 8+) and zero orders → cut the bid hard or pause. Never, under any logic, raise a bid on a term with spend and no sales.
• No blanket increases. A bid only goes up when the term is below target ACOS with real order data behind it. “Everything else” holds.
• Low-data terms hold flat. One to three clicks and no orders isn't evidence of failure — it's an unfinished experiment.
• Proportional cuts on high ACOS. A term slightly over target gets a trim. A term at triple your target gets cut to the floor. The cut scales with how far over the term is.

Verdict first, rows second. Before touching a single bid, compute the account headline: spend, sales, ACOS, ROAS, CPC, conversion rate — broken out by match type. If you don't know whether the account is working, you're not ready to edit it.

What this looks like in practice

In that audit, the fix wasn't clever. We cut or paused the zero-sale terms, removed the blanket increase, and left the proven winners alone. No new tools, no restructure — just closing the gap between what the rules could see and where the money was going.

If you manage bids by ACOS — and almost everyone does — pull your last bulk file and filter for clicks with zero orders. That column of empty ACOS cells is where I'd look first.

← All insights