A few months back I published an audit of a real ecommerce account and found five problems that were quietly costing the owner money. People ask me the same question every time I post one of these: okay, but what actually happened after you fixed it? So here's the honest answer, with the real monthly numbers, including the part where the account got worse before it got better.
01The Five Problems From the Original Audit
Quick recap for anyone who hasn't seen the original audit. The account, an ecommerce store I'm keeping anonymized the same way I keep every account in this series anonymized, had five issues stacked on top of each other: broken and duplicated conversion tracking, live policy violations nobody had gone in and cleared, a single Performance Max asset group running hundreds of products with no brand exclusion list, and a Search campaign set to Maximize Clicks instead of Maximize Conversions, burning budget on broad match terms that had nothing to do with what the store actually sells.
None of these five problems are exotic. I see some version of this combination on a large share of the accounts I audit. What makes this account worth a follow-up isn't the diagnosis, it's what the fix actually looked like on the numbers over the following two months, because that part rarely gets shown.
02Seven Months of ROAS, Month by Month
Here's the account's blended ROAS for the first seven months of 2026, pulled straight from the exported campaign report and cross-checked against the live Google Ads UI. January through March sits in a healthy, boring range. April starts slipping. May is where I recorded the original audit. Then the fixes roll out, and June is the low point of the entire year before July jumps to the best month of the whole window.
| Month | Cost | Conv. Value | ROAS | Cost / Conv |
|---|---|---|---|---|
| Jan 2026 | $3,673.01 | $13,623.30 | 3.71x | $50.33 |
| Feb 2026 | $3,612.31 | $14,266.22 | 3.95x | $54.11 |
| Mar 2026 | $3,680.88 | $14,482.89 | 3.93x | $42.66 |
| Apr 2026 | $3,656.09 | $11,783.60 | 3.22x | $40.19 |
| May 2026 | $4,661.01 | $10,808.64 | 2.32x | $55.38 |
| Jun 2026 | $5,151.42 | $8,742.22 | 1.70x | $66.51 |
| Jul 2026 | $3,382.15 | $18,452.24 | 5.46x | $36.06 |
May is when I recorded the audit video. June is when the fixes were live and the account looked, on paper, like it had gotten worse: ROAS down to 1.70x, cost per conversion up to $66.51, both the worst readings of the entire seven-month stretch. If you stopped watching after June, you'd think the audit made things worse. July is why you don't stop watching after one month.
I review every submission myself. Request a free audit and I'll tell you what I'd fix first and why, the same way I approached this account.
03Why the Dip Happens When You Fix Tracking or Structure
The June dip isn't a sign the fixes were wrong, and I want to be specific about why, because "it needs time to relearn" gets said a lot without much explanation attached to it.
Fix broken conversion tracking and you've changed the actual data the bidding algorithm learns from. It doesn't carry forward what it thought it knew. It has to relearn, from a colder start, which clicks are the good ones, on a cleaner signal it hasn't seen before. That relearning period shows up as a dip almost every time, and the size of the dip roughly tracks how wrong the old signal was.
Restructure a Performance Max asset group and add a brand exclusion list, and you're doing something else at the same time: telling Google's algorithm to stop taking credit for branded searches it used to auto-win with almost no effort. That's an artificial ROAS boost disappearing, not a real one. The campaign looks worse in the report. It's actually being measured honestly for the first time since it launched.
04Where the "Missing" Revenue Actually Went
The clearest proof this was a measurement fix and not a real drop in performance sits inside the Brand Search campaign specifically, the one Performance Max had been quietly cannibalizing before the brand exclusion list went in.
$468.91 in tracked conversion value in May. Nearly invisible in the account, because PMax was auto-winning most of that branded traffic and taking the credit for it.
$3,433.57 in June, then $10,986.65 in July, roughly a 23x increase, on spend that stayed flat in the $100 to $350 a month range the whole time.
That jump isn't new revenue appearing out of nowhere. It's revenue getting correctly credited to the campaign that actually earned it, instead of Performance Max quietly claiming it. For contrast, PMax's own reported ROAS only partially recovered on its own: 2.44x in May, dropping to 1.14x in June, then 2.46x in July. Looking at PMax alone would tell you the wrong story. It's the account total that tells you the real one, because the traffic PMax used to "steal" is finally counted where it belongs.
If your own account has one asset group covering hundreds of products with no exclusions, this is worth checking directly rather than assuming it isn't happening to you. I walk through how I'd rebuild that structure in my PMax budget and campaign structure guide.
05Why Agencies Avoid Touching a Broken Account
This is also, I think, exactly why a lot of agencies and freelancers avoid touching a broken account once they inherit it. Fixing tracking, structure, or bidding strategy makes the numbers look worse in the short term. The easy move, the one that keeps the client happy for another month, is to leave the broken thing alone and keep collecting the retainer on numbers that were never accurate to begin with.
That's backwards, and it's a pattern worth knowing how to spot from the outside. If you've hired someone to manage your account and the reported ROAS has looked suspiciously stable for months while your actual revenue hasn't moved the way you'd expect, ask directly whether conversion tracking and campaign structure have been reviewed since day one. I've written more on the specific signs that a manager is avoiding structural work in my piece on Google Ads agency red flags.
06What to Expect If You're About to Fix Your Own Account
If you're running Google Ads for your own store and you're about to fix a real structural problem, tracking, bidding strategy, campaign structure, don't expect an immediate jump. Expect a dip while the algorithm relearns on the cleaner signal, and judge the trend after that window closes, not from the first week of new numbers.
The fastest way to know whether your own account is set up this cleanly to begin with is a direct audit, not guessing from the reported ROAS. I go through the same process I used on this account, tracking, structure, bid strategy, in my Claude AI audit workflow breakdown, and the original walkthrough of this exact account is in my Real Client Google Ads Audit post.
Tracking, structure, bid strategy: I'll review your account directly and tell you what I'd fix first, the same way I did here, no chatbot involved.
