How to Write High-Converting Google Ads Copy: RSA Frameworks, Headline Strategies, and a Testing Approach That Actually Works

The average Google search results page now has three to four paid ads at the top, all fighting for the same click. Most of them say almost exactly the same thing. “Free Quote.” “Trusted Experts.” “Call Us Today.” They blur together into beige wallpaper, and the searcher’s eye jumps straight to the organic result — or the one ad that actually said something specific enough to earn attention.

Your Google Ads copy is either doing real selling work, or it’s decorating a page. There’s no middle ground. After managing over $50M in Google Ads spend, we’ve seen campaigns with identical budgets and targeting produce wildly different results based almost entirely on the copy. Here’s what actually separates the ads that convert from the ones that quietly drain your budget.

Why Most RSA Copy Underperforms (And It’s Not the Algorithm’s Fault)

When Responsive Search Ads became the default format in 2022, a lot of advertisers treated them like expanded text ads with extra slots. They wrote 15 headlines that all said essentially the same thing — just in slightly different word orders — then handed the reins to Google’s machine learning and hoped for the best.

The problem: Google’s system can only combine what you give it. If all your headlines are slight variations of “Award-Winning Digital Marketing Agency,” you haven’t given the algorithm any meaningful variation to test. You’ve given it 15 synonyms and called it RSA copywriting.

RSA copy best practice starts with a simple rule: every headline should make a different argument. Some argue relevance (“Google Ads for SaaS Companies”). Some argue differentiation (“No Long-Term Contracts”). Some argue urgency (“Campaigns Live in 48 Hours”). Some argue social proof (“Trusted by 300+ B2B Brands”). Cover different dimensions of why someone should click, and let Google figure out which combination resonates best for each query and audience signal.

The Three-Layer RSA Headline Framework

Every RSA you write should have headlines spread across three distinct layers. Think of it less like filling in a template and more like building a case a lawyer would make — you’re covering every angle of “why click this ad.”

Layer 1: Relevance Headlines (5–6 headlines)

These prove you’re a match for the search. They mirror the searcher’s intent as closely as possible without being robotic about it. This is where you lean on keyword insertion thoughtfully, not lazily.

Examples for a Google Ads agency:

• “Google Ads Management for B2B”
• “Managed Google Ads Campaigns”
• “PPC Management for SaaS Brands”
• “Google Ads Agency — Free Audit”

These won’t win any copywriting awards, but they serve a critical function: they signal to the searcher in under a second that you understand what they were looking for. Relevance headlines anchor the RSA in the search context.

Layer 2: Differentiation Headlines (5–6 headlines)

This is where most advertisers leave the most money on the table. Differentiation headlines answer the question the searcher is asking internally: “Why you specifically, and not the other four ads on this page?”

Specificity is everything here. “Great Results” is not differentiation. “Average 4.2x ROAS Across Ecommerce Clients” is. “Experienced Team” is noise. “Ex-Google Employees Managing Your Account” is a reason to click.

Push yourself to go further than you’re comfortable with. The headline that feels too bold to you is usually the one that converts best. Some examples:

• “No Retainers — Pay Per Performance”
• “Average 67% Reduction in Wasted Spend”
• “Dedicated Account Manager, Not a Ticket Queue”
• “We’ve Managed $50M+ in Google Ads Spend”

Layer 3: CTA and Urgency Headlines (3–4 headlines)

These close the loop. They tell the searcher exactly what happens when they click and create a low-friction next step. The best CTA headlines name the action and either reduce perceived risk or create a soft urgency.

• “Get Your Free Account Audit Today”
• “See Where Your Budget Is Being Wasted”
• “Book a 20-Min Strategy Call — No Pitch”
• “Start with a Free Audit — No Commitment”

Notice those CTAs are doing more than saying “Click Here.” They name a specific deliverable and pre-empt the objection (“no commitment,” “no pitch”). That’s the difference between a CTA headline that pulls people in and one that just fills a slot.

Description Lines: The Most Underused Real Estate in Google Ads

Here’s something we see constantly: advertisers spend an hour agonizing over headlines, then write description lines in four minutes. The descriptions are an afterthought — one generic sentence about services and a phone number copy-pasted from the website footer.

Description lines are where you handle objections and deepen desire. You’ve got up to 90 characters per description line, and Google will often show both. Use them to do the work that headlines can’t fit into 30 characters.

Structure your description lines to complement the headline mix. If your headlines lean on relevance and differentiation, let descriptions handle trust signals and risk reversal. Something like:

• “We’ve managed campaigns for 300+ B2B companies. No fluff, no lock-in contracts — just results you can see in your dashboard.”
• “Free Google Ads audit shows you exactly where your budget is leaking. Takes 20 minutes. No sales pressure, just data.”

One is leading with social proof and a bold positioning statement. The other is reducing friction around the offer. Together, they cover different decision triggers — which is the whole point.

The Pinning Problem: When You’re Fighting the Algorithm

Pinning lets you lock specific headlines to specific positions in your RSA. Position 1 always shows that headline, position 2 always shows another. It sounds great for brand control, and it has legitimate uses — legal disclaimers, brand name consistency, mandatory promotional language.

But here’s what happens when you over-pin: you gut the RSA model. Google’s ability to test combinations shrinks dramatically when you’ve pinned five of your fifteen headlines. You’re essentially running an expanded text ad with extra steps. Your ad strength score will drop, your impressions will narrow, and you lose the core value proposition of the RSA format entirely.

Our rule: pin only what you absolutely must, and never pin more than two headlines. If legal needs a disclaimer, pin that. If your brand name needs to appear in position 1, pin that. Everything else — let the machine test it.

How to Test RSA Copy Without Losing Your Mind (or Your Budget)

Ad copy testing inside RSAs is more nuanced than it was with expanded text ads, where you could just duplicate an ad, change one element, and let it run. RSAs don’t work that way — Google is doing the combination testing internally, so your job shifts to testing at the asset level.

Read Your Asset Performance Labels Correctly

Google rates each headline and description as “Best,” “Good,” “Low,” or “Learning.” Most people look at “Low” and immediately delete that headline. That’s wrong, and we did it wrong for years before we understood it.

“Low” doesn’t always mean the headline is bad — it sometimes means it’s not being served often because similar headlines with more history are winning the rotation. Before you cut a “Low”-rated headline, check how many impressions it’s received. If it’s below a few thousand, it hasn’t had a fair test yet.

Delete “Low” performers only after they’ve had meaningful exposure. The threshold depends on your volume — in a high-spend account, 5,000+ impressions is a reasonable bar before making a judgment call.

Run Campaign Experiments for Structural Tests

If you want to test a fundamentally different angle — say, price-focused messaging versus outcome-focused messaging — use Google’s Campaign Experiments feature to split traffic 50/50. That’s a proper A/B test. Asset-level performance labels tell you which individual elements are winning; campaign experiments tell you which overall ad strategy is winning.

Use both. They answer different questions.

Give Tests Enough Time and Enough Traffic

Three days is not a test. Two weeks with 200 clicks is not a test. Aim for at least 30 days and enough conversions to reach statistical significance — usually 50–100 conversions per variant minimum. If your account doesn’t have that volume, you’ll need to be more patient and rely more on directional data rather than claiming a winner prematurely.

Ad Copy Best Practices That Actually Move Metrics (Not Just “Best Practices” Theater)

Everyone’s list of ad copy best practices includes “use your keyword in the headline” and “include a clear call to action.” That’s true, and also the floor, not the ceiling. Here’s what separates accounts running good copy from accounts running great copy:

Mirror the searcher’s language, not your internal vocabulary. If your customers call it “expense tracking software,” your ad shouldn’t say “expenditure management solutions.” Search the same terms your prospects search. Read their Reddit threads, their G2 reviews, their LinkedIn comments. Then write in that language.

Price anchoring works in headlines when your pricing is a feature. If you’re cheaper than the competition, say it. “Plans from $299/mo” will outperform “Affordable Pricing” every time. If you’re premium, use it as a filter — “Enterprise-Grade Campaigns from $5K/mo” will repel wrong-fit leads and attract the right ones.

Numbers build credibility faster than adjectives. “Fast Results” means nothing. “First Optimizations Delivered in 48 Hours” means something. Replace every vague descriptor in your headlines with a specific number wherever you can honestly put one.

Questions in headlines work — but only when the answer is obviously yes. “Wasting Money on Google Ads?” works because the searcher looking for a Google Ads agency probably suspects they are. “Looking for Better Results?” is too generic to do any work.

Frequently Asked Questions

How many headlines should I write for an RSA?

Fill all 15 headline slots. This isn’t optional — every unfilled slot is a missed opportunity for Google’s algorithm to test a winning combination. If you’re struggling to get to 15, you’re not thinking across different layers of the argument. Go back to the three-layer framework and make sure you’re covering relevance, differentiation, and CTAs.

Should I use keyword insertion in Google Ads headlines?

Yes, but selectively. Keyword insertion is powerful for relevance, but it can produce embarrassing or nonsensical output if your keyword list includes weird long-tail variations. Always preview how the insertion will look across your top keywords before enabling it, and have a strong default fallback text set.

How do I know if my Google Ads copy is actually converting?

Stop optimizing for CTR alone. A high click-through rate on a low-converting headline just means you’re spending more to get people to a page that isn’t closing them. Connect your RSA performance data to your actual conversion events — leads, calls, purchases — and optimize toward the combinations that drive real outcomes, not just traffic.

What’s a good ad strength score for RSAs?

Target “Excellent” — but treat ad strength as a hygiene metric, not a performance metric. Google has explicitly said ad strength doesn’t directly predict performance. What it does tell you is whether you’re following structural best practices: enough headlines, enough variety, description lines filled out. Use it as a checklist, not a KPI.

How often should I refresh my Google Ads copy?

Review asset performance monthly. Don’t change headlines just to change them — only swap out “Low”-rated assets that have had sufficient impression volume, or when you have a new offer, promotion, or angle worth testing. In high-volume accounts, you might iterate faster; in smaller accounts, give tests more runway. Churn for churn’s sake just resets your learning data.