Most Google Ads accounts I audit aren't broken. They're 80-90% of the way there, built by someone competent, running profitably, ticking along. The problem is that last 10-20% is where a huge amount of money hides.
I recently consulted with an ecommerce company selling custom dental appliances. They're spending $10,000-$15,000 per month on Google Ads, running traditional Shopping and Search campaigns (no Performance Max), and hitting a respectable 3x-3.5x ROAS overall. Solid account. No disasters. But I found 11 specific areas where they were leaving money on the table. And I'd bet most of you reading this are making at least a few of the same mistakes.
Let me walk you through every single one.
Capture Every Branded Search
This account was sitting at 94% impression share on branded search terms. Sounds good, right? It's not. When your brand campaigns are delivering 10x+ ROAS, there is zero reason to let 6% of your branded traffic slip away. That's qualified traffic actively searching for your business, and you're handing it to competitors.
The fix is simple: lower your brand campaign's Target ROAS to something like 13x-15x. Yes, lower. By reducing the ROAS target, you're telling Google's algorithm it can bid more aggressively. The goal is 99-100% impression share on brand terms.
If someone is searching for your brand, you should be showing up. Every. Single. Time.
While you're at it, check whether you're losing any impression share to budget. This account was losing roughly 4% to budget constraints. When your ROAS targets are being met, you should be aiming for 0% lost impression share. A simple 4-8% budget increase can capture that lost traffic instantly.
Tighten Up Non-Branded Efficiency
On the flip side, non-branded campaigns often benefit from raising ROAS targets. This account was running at 3x-3.5x overall, but I recommended testing a 4x-4.5x target on non-branded campaigns specifically.
Yes, this might reduce volume. But if you're operationally constrained (as many ecommerce businesses are), you want every dollar working harder. The key metric to monitor here isn't ROAS alone, it's contribution margin. Track what happens to your actual profitability over 1-3 months as you adjust these targets.
Your Responsive Search Ads Are Probably Too Complex
Here's one I see constantly: advertisers stuffing 12-15 headlines into their Responsive Search Ads because Google gives them the option. This account had 15 headlines per RSA. Sounds thorough, but the maths tells a different story.
With 15 headlines, you're creating over 4,000 possible ad combinations. Now think about how much traffic a typical ad group receives. At that volume, impression distribution drops from roughly 1,400 impressions per combination down to under 100. Google simply cannot test 4,000+ combinations with any statistical significance at normal traffic levels.
My recommendation: strip your RSAs back to 7-8 headlines and just 2 descriptions. This focuses Google's testing, and I've seen significant conversion rate increases when accounts make this change. Keep your current winning ads as the control, then test these simplified versions against them.
Implement a Proper Ad Testing Cadence
This particular account hadn't run proper ad tests since 2022 and 2024. That's far too long between iterations. Your market changes, your competitors change, what resonates with buyers changes.
You should have a regular testing process: keep your best-performing ad as the control, create a challenger with one meaningful difference, and let them run head to head. Consider setting a monthly reminder to review ad performance and decide whether to iterate. You can even automate parts of this with AI workflows if you want to save time.
Modernise Your Shopping Campaigns
This account was running a legacy three-tier priority Shopping structure, something a previous agency had built over three years ago. It was primarily designed to filter brand terms using high, medium, and low priority campaigns with manual CPC bidding.
The world has moved on.
Here's what I'd do: switch the brand Shopping campaign from manual CPC to Target ROAS bidding, matching roughly the same ROAS the campaign currently achieves. This alone typically improves both efficiency and volume because Google's algorithm can react to signals you simply cannot manage manually.
For the general (non-brand) Shopping campaign, the bigger move is testing Performance Max.
Give Performance Max Another Chance
I know, I know. Many of you tested PMax back in 2023 or early 2024 and it didn't blow your socks off. This account last tested it in late 2023/September 2024. But Performance Max has improved significantly since then, and dismissing it based on old results is a mistake.
Here's how to test it properly:
Run it as a campaign experiment against your Standard Shopping general campaign. Use full assets, not a feed-only setup. This is important because a full-asset PMax campaign reveals which channels actually work for your business: Video, Gmail, Discover, and more.
Give the experiment 6-8 weeks minimum for statistical significance. Even if PMax doesn't outperform Standard Shopping (which it very well might), you'll learn something valuable about which channels resonate with your audience. That intelligence alone is worth the test.
Split Test Your Shopping Images
With a small product catalogue (this account had roughly 4 products), image testing in Shopping is incredibly manageable, yet almost nobody does it.
Most ecommerce advertisers use the standard product-on-white-background image and never question it. But lifestyle images and in-use product shots can dramatically change click-through rates. The testing process is straightforward: create duplicate product listings with different images and monitor CTR differences. You only need 30-50 clicks per variation to spot meaningful differences.
If you have a small catalogue, there's really no excuse not to run this test. It's low effort with potentially high reward.
Run a Keyword Gap Analysis
This is one of my favourite recommendations because it nearly always uncovers opportunities in mature accounts. Here's the process:
Download your search terms report from Standard Shopping, filtered to converting terms only.
Download your full keyword list from Search campaigns.
Upload both CSVs to ChatGPT (or your AI tool of choice) and ask it to identify converting Shopping terms that aren't targeted as keywords in Search.
That's it. This process typically finds 100+ keyword opportunities in mature accounts. Terms that are already converting through Shopping but aren't being specifically targeted in Search, meaning you're missing the chance to show tailored ad copy and dedicated landing pages for those queries.
Test AI Max for Search
If you're running traditional phrase and exact match Search campaigns, you should be testing AI Max. It's a setting within your existing Search campaigns that allows Google to discover new search patterns and keyword variations you might not have thought of.
The beauty of it is that it's easily reversible. Enable it, run it for 1-2 months, and if performance declines, simply turn it off. But for mature accounts that may have gone stale on their keyword lists, this can surface entirely new pockets of demand.
Replace Image Retargeting With Video
Image-based display retargeting has been on the decline in Google Ads for a while now. If you're still running traditional image remarketing campaigns, it's time to consider shifting that budget to video-based retargeting instead.
The options here are YouTube (landscape format) and YouTube Shorts (portrait format). If you decide to test Performance Max with full assets as I recommended above, you'll naturally start to see whether video resonates with your audience. If it does, you can then launch dedicated video retargeting campaigns with confidence.
The Testing Framework That Ties It All Together
None of these recommendations should be implemented blindly. Here's the framework I'd suggest:
Test duration: 1-3 months for ROAS target changes; 6-8 weeks minimum for campaign experiments
Use Google's campaign experiments feature for A/B testing wherever possible
Monitor contribution margin, not just ROAS. A higher ROAS doesn't always mean more profit
Review weekly: Check impression share and budget constraints every week without fail
The overall assessment of this account was that it's well-built, roughly 80-90% optimised, with no major structural issues. The improvements I've outlined here are incremental efficiency gains rather than dramatic overhauls. But in an account spending $10,000-$15,000 per month, incremental gains compound into tens of thousands of dollars over a year.
If your account has been managed by the same team (or the same approach) for more than a year without meaningful changes, I'd encourage you to run through this same checklist. You might be surprised by what you find.
Conclusion
A mature Google Ads account that's "doing fine" is often hiding significant untapped potential. The 11 optimisation areas covered here apply broadly to most ecommerce advertisers:
Brand campaigns should target 99-100% impression share when ROAS is strong
Non-branded campaigns may benefit from higher ROAS targets to improve efficiency
Responsive Search Ads perform better with fewer headlines (7-8) and descriptions (2)
Ad testing needs a regular cadence, not once every two years
Shopping campaigns should use Target ROAS bidding, not manual CPC
Performance Max deserves a fresh test with full assets and proper experiment setup
Shopping images should be split tested, especially with small catalogues
Keyword gap analysis between Shopping and Search typically reveals 100+ opportunities
AI Max for Search can uncover new demand in mature accounts
Video retargeting should replace declining image-based remarketing
A structured testing framework with weekly reviews ties everything together
The key takeaway: focus on profitability and efficiency rather than raw volume, and use proper testing methodology to validate every change.