We Monitored 26 DTC Google Ads AccountsHere's What Actually Breaks

The headline ROAS number is the least interesting thing in a Google Ads account. Across six months of daily monitoring on 26 DTC and ecommerce accounts, the problems that actually cost money sat one level down: individual products burning budget, and changes nobody on the managing team made.

This is first-party data from the monitoring system we run on the accounts we manage, not a survey or a scraped roundup. Every figure is a rate or a percentage, never a raw spend or revenue figure, and no account is identifiable.

Methodology

The data comes from the monitoring system that runs daily on every Google Ads account Shopiator manages. It stores account-level daily performance, a daily 30-day product performance snapshot for Shopping and Performance Max, performance alerts, and the account's change history pulled from the Google Ads API.

The dataset covers 26 DTC and ecommerce accounts from March 12 to September 26, 2026. Accounts that are no longer managed by us and non-ecommerce accounts were excluded. Each finding states how many accounts it draws on, because not every account runs every campaign type. ROAS is measured by conversion time (conversion value attributed to the day of the click), not by reporting date. No account names, domains, IDs or raw spend and revenue figures appear anywhere in this piece.

Product-level problems outnumber account-level ones roughly 10 to 1

Of every performance alert the system raised across 26 accounts, 88% were product-level: a single SKU with a CPC far above the account norm, a ROAS well under target, or spend with no conversions at all. Account-level alerts, meaning ROAS drops, cost spikes and sustained declines across the whole account, made up about 8%. The remaining 4% were positive signals, campaigns beating target with room to scale.

Product-level alerts fired in 21 of the 26 accounts, nearly every account running Shopping or Performance Max. That's the practical point: if you only watch the account ROAS, you are watching the one number that almost never breaks.

18% of product spend was doing little or nothing

In the latest 30-day product snapshot across 22 ROAS-target accounts, 13% of all product-level spend went to products with zero conversions in the window. Another 5% went to products that did convert, but at under half of the account's target ROAS.

Pooled across accounts that's roughly one dollar in six, and the typical account was worse: the median account had a quarter of its product spend on zero-conversion products. Some of that is deliberate, because feeder campaigns exist to spend on discovery, but most of it is exactly what a weekly SKU review should catch. The account-level ROAS was on target in most of these accounts at the same time, because a handful of strong products carried the average. Nothing in the headline number tells you the leak is there. For the structural fix, see our Shopping feeder and converter architecture and the product-level checks in our wasted spend checklist.

Most accounts hit target ROAS. The margin is the story.

Of 15 ROAS-target accounts with meaningful spend in the last 30 days, 11 (73%) beat their target by conversion time. Two of the four misses were within 25% of target. The other two were small accounts spending under $10K a month.

That's a healthier baseline than the industry narrative allows for, and it's exactly why the product-level finding matters. Hitting target is not the same as spending well: an account can beat target ROAS while one dollar in six goes to products that shouldn't be getting it.

Two in three accounts were changed by someone outside the managing team

Over the six months, 16 of 25 accounts with change history (64%) had at least one edit made by someone other than the managing team, and those outside edits were 8.6% of all logged changes. In practice that's the brand's own team, a past freelancer who still has access, or a connected app.

None of that is wrong by default. The brand owns the account and should have access. The risk is that nobody reconciles it: a budget raised on a Friday or a campaign paused from a phone becomes an unexplained swing in next week's numbers. A weekly change-history review, and removing stale user access, closes that gap.

So what

The recurring theme is that the costly problems live below the headline. Product-level leaks and unreconciled changes don't show up in an account-level ROAS, and they don't fix themselves with more budget or a new campaign type. They get fixed when someone checks SKU-level performance and the change log every week.

If your account hasn't had a structural review in the last 60 to 90 days (bid strategy by campaign role, SKU-level Shopping and PMax performance, change history), there's a good chance at least one of these patterns is sitting in it now.