20 checks, in the order that actually moves ROAS. Start with account structure, end with the one thing every other checklist like this one skips: the ad itself.
Most Google Ads "optimization" work is account maintenance: structure, negatives, bidding hygiene. It's necessary, and it's where every guide like this one starts. But it also caps out fast. Perfect structure and a clean negative list make sure the right person sees your ad at a fair price. None of it controls whether that person actually wants to click once your ad is in front of them. Work through the six sections below in order, then read the last one: it's the step every checklist including the ones ranking above this page leaves out.
01
Account & campaign structure
One campaign per intent tier, not per product line
Branded search, generic high-intent search, and top-of-funnel terms have wildly different CPCs and conversion rates. Blending them into one campaign forces one bid strategy to serve three different jobs, and it ends up optimizing for none of them well.
Separate Search from Performance Max for the same terms
Performance Max respects your Search keyword targeting, but it can still surface for branded terms when your Search campaign has a limited budget or looser targeting. Add brand exclusions in Performance Max, or run a dedicated branded Search campaign and watch the search terms insights report for overlap.
Pause anything under roughly 10 conversions a month
Google's automated bidding needs a real conversion signal to learn from. A campaign limping along at 2-3 conversions a month isn't being optimized, it's random. Consolidate it into a bigger ad group or cut it.
02
Keywords, match types & negatives
Move off broad-only once you clear roughly 30 conversions
Broad match is genuinely useful for discovery in a new account with no data. Once you have enough conversion volume for Smart Bidding to work with, layer in exact and phrase match on your proven terms so spend concentrates on what's already confirmed to convert.
Run the search terms report weekly for a new campaign's first month
This is the single highest-leverage, most-skipped task on this whole list. Every irrelevant query it surfaces is money leaking out in real time, and every good one is a new exact-match keyword you haven't captured yet.
Build your negative list on day one, not after the first invoice shock
Start with your own brand's common misspellings if you don't want cheap traffic diluting data, competitor brand names (unless that's a deliberate strategy), and generic informational terms like "free," "jobs," "salary," and "diy."
03
Ad copy & Responsive Search Ads
Fill every RSA asset slot Google gives you
Headlines top out at 30 characters and descriptions at 90, so there's real room to test more angles than most advertisers use. Google can't test a headline combination that doesn't exist: thin RSAs (5-6 headlines instead of the max 15) get less impression share and lower Ad Strength, not as a penalty, just because there's less for the auction to work with.
Only pin an asset when a claim is legally or factually fixed
Pinning a price, a disclaimer, or a legal claim to a specific position is correct. Pinning your "best" headline out of habit stops Google from testing combinations you haven't thought of, and some of those will outperform your favorite.
Match at least 2 headlines to the literal query intent
A generic "Shop Now, Best Prices" headline loses Ad Rank against a competitor whose headline echoes the searcher's exact phrase back at them. Specificity beats enthusiasm every time this gets tested.
04
Quality Score & the landing page
Pull the 3-part Quality Score breakdown, not just the 1-10 number
Quality Score is really three separate diagnostics: Expected CTR, Ad Relevance, and Landing Page Experience, each rated below average, average, or above average against other advertisers over the trailing 90 days. A score of 4 caused by weak landing page experience needs a completely different fix than one caused by weak ad relevance.
Restate the ad's exact promise on the landing page headline
The most common Quality Score killer isn't the page's design, it's a message-match gap. If the ad promises "same-day quote" and the page headline is your company tagline, the disconnect costs you Quality Score and costs the visitor their trust in the same instant.
05
Bidding & budget
Don't switch bid strategies more than once every 2-3 weeks
Every strategy change resets the learning phase, during which performance is genuinely noisier and often worse. Switching again before it stabilizes means you're perpetually judging a campaign mid-relearn, not at its real performance.
Set Target CPA or ROAS from your own trailing 30-day data, not Google's suggestion
Google's suggested targets are calibrated to keep spend flowing, not to your margin. Pull your actual blended CPA or ROAS over the last full month and set the target near that, then let it drift down or up gradually rather than in one aggressive jump.
Only daypart once you have real hourly conversion data
Cutting the 2am-6am window because it "feels" wasteful, with no data behind it, just cuts volume Smart Bidding could have used to learn faster. Check the day and hour report for at least a few weeks of actual conversion timestamps first.
06
Audiences
Add audience layers as Observation before ever using them to restrict
Observation mode shows you how an audience segment actually performs against your campaign without touching delivery. Only promote a segment to targeting-mode restriction once the data itself justifies narrowing.
Give remarketing lists 90+ days of retention as a floor
For anything beyond a low-consideration impulse purchase, a 30-day remarketing window closes before most real buying cycles do. Extend it, especially for higher-ticket offers where the research-to-purchase gap is measured in weeks, not hours.
07
The step every checklist skips: the ad itself
Everything above is a knob on the machine: structure, bids, audiences, Quality Score. They all optimize distribution, making sure the right person sees your ad, at the right price, at the right time. None of them touch persuasion: whether the headline, the offer, and the copy actually overcome a stranger's hesitation in the half-second before they decide to click or scroll past.
That's a different kind of audit: behavioral psychology, offer clarity, message match to the landing page, not a bidding or structure problem at all. It's also the fastest check on this whole list. Drop in your ad and get a scored, ranked read in under a minute.
How often should I actually optimize a Google Ads account?
Check the search terms report and Quality Score weekly for the first month of any new campaign. Once an account is stable, a deeper pass every 2-3 weeks is enough. Anything more frequent usually just means reacting to noise instead of signal.
What's a good Quality Score?
7+ is solid, but the number alone is less useful than the 3-part breakdown behind it (ad relevance, expected CTR, landing page experience). A 5 caused by one weak component with two strong ones is a much easier fix than a 5 caused by all three being average.
Does this checklist still apply on a small budget?
The structure and negative-keyword items apply at any budget. The bidding-strategy and audience-layering items need enough conversion volume to have real data behind them: below roughly 15-30 conversions a month, keep bidding manual or use Maximize Clicks until you clear that threshold.
Performance Max or standard Search campaigns?
Not either/or. Run both, but exclude your own brand terms from Performance Max, or watch the search terms insights report closely, so it isn't just relabeling conversions Search was already winning.
I've done all of this and performance is still flat. What's left?
Almost always the ad itself. Structure, bidding, and Quality Score all cap out at making sure the right person sees your ad at a fair price. None of them control whether that person actually wants to click once it's in front of them. See the section below.