RSA Optimization: Why Ad Strength Doesn't Matter and What Actually Does

Google wants you to believe that Ad Strength matters. Here's what actually moves the needle.

A study of 93,055 RSAs shows that Ad Strength has zero impact on Ad Rank or Quality Score. Zero. It's a compliance score — a measure of how well you followed Google's formatting recommendations, not how well your ad performs. I've seen "Poor" strength ads outconvert "Excellent" ones by 4x because they were tightly targeted instead of generically diverse.

The things that actually matter: 2 RSAs per ad group, keywords in only 20-30% of headlines, strategic pinning decisions, and a weekly testing cadence using a feature most advertisers don't even know exists.

RSA Performance Data: What 93,055 Ads Tell Us

With ETAs sunset, 92% of advertisers now run RSAs. Google optimizes for click volume (revenue for Google), not conversion quality (revenue for you). That structural conflict shapes everything below.

Optmyzr analyzed 93,055 RSAs across 13,671 accounts. Key findings:

• RSAs deliver 4x the impressions of ETAs with an average 7% conversion improvement
• Two RSAs per ad group is the sweet spot — one gives too few combinations, three+ fragments data
• Use all 15 headline and 4 description slots. Every empty slot is a missed testing opportunity

Related reading: For the full breakdown of Google's incentive structure, see Google Ads Efficiency Playbook: The Foundation.

Ad Strength: The Metric That Doesn't Matter

This is the thing most advertisers get wrong:

Ad Strength does NOT impact Ad Rank.

Ad Strength does NOT impact Quality Score.

Ad Strength does NOT correlate with actual performance.

What Ad Strength Actually Measures

Ad Strength is a compliance score — did you fill all slots, vary your headlines, include keywords? Each Ad Strength improvement yields approximately a 3% click lift. Not conversion lift. Click lift. Clicks Google charges you for.

A law firm running pinned headlines like "24/7 Emergency Criminal Defense Attorney" might get "Poor" Ad Strength but convert at 3x the rate of a "Good" strength ad with generic variations. I've seen this pattern repeat across dozens of accounts.

Focus on these instead: conversion rate within target CPA, conversion volume at acceptable cost, CTR as a secondary signal, and cost per conversion trending down.

The Keyword Headline Rule: 20-30% Maximum

Google recommends loading up your headlines with keywords. The data says otherwise.

The Optimal Keyword Ratio

Include keywords in only 20-30% of your headlines (3-4 out of 15).

Why Not More Keywords?

Keywords in every headline produces spam like:

Emergency Plumber Houston | Houston Emergency Plumber | 24/7 Plumber Houston

Compare with a diversified approach:

Emergency Plumber Houston | Fixed in 60 Minutes or Free | 4.9 Stars - 500+ Reviews

One keyword headline for relevance, one benefit for value, one trust signal for credibility.

Description Strategy

For your 4 descriptions (90 characters each), assign each a unique angle: primary value proposition + CTA, trust and credibility, specific offer or differentiator, and urgency or social proof. Example: "No overtime charges. Same flat rate 24/7. Get your free quote in minutes."

The Pinning Debate: Data vs Dogma

Pinning forces specific headlines or descriptions into fixed positions. Google discourages it. The data tells a more nuanced story.

The 93K RSA Pinning Study

From the Optmyzr study: fully pinned ads get 3.9x fewer impressions but deliver higher CTR and conversion rates. Volume vs. quality — know which you're optimizing for.

Pin when: legal/compliance requirements, proven headline winners from data, brand name must appear in position 1, specific promotions, or local market where location must be visible.

Do not pin when: launching without historical data, collecting initial data, or testing new service offerings.

The Hybrid Approach

What works best for most service businesses: pin position 1 to your strongest keyword headline, pin description 1 to your primary CTA, and leave everything else unpinned. This gives you control over the most important position while letting Google test the rest.

Multi-pin tip: Pin 2-3 headlines to the same position (e.g., "Emergency Plumber Houston," "24/7 Plumbing Service Houston," "Houston Emergency Plumber - Fast"). Google rotates among them while keeping position 1 always relevant.

Ad Copy Testing Framework

Traditional A/B testing is dead for RSAs. Google rotates combinations automatically, so you test differently now.

The Ad Variations Feature (Hidden Gem)

Location: Campaigns > Experiments > Ad Variations

Most advertisers don't know this exists. It lets you test across multiple campaigns simultaneously with true 50/50 traffic splits and statistical significance indicators.

Setup: Create a variation, select target campaigns, choose your modification type (Find and Replace for testing CTAs/USPs, Update URLs for landing pages, or Update Text for adding/removing/pinning assets). Name clearly, set max 84-day duration, 50/50 split, and launch.

What to Test (One Element at a Time)

One variable per test. Priority order: CTA language, primary value prop, trust signals, pricing language, urgency elements, landing page, display path.

Statistical Significance

• At least a few hundred clicks and 1,000 impressions per variation
• Blue star icons indicate statistical significance
• Run for minimum 2 weeks, ideally 4 weeks
• Review at campaign, ad group, and individual ad levels — a CTA that wins for emergency plumbing ("Call Now") may lose for kitchen renovation ("Get Your Free Design Consultation")

The Weekly RSA Testing Cadence

Consistent testing compounds. Here's the weekly rhythm I recommend:

Friday Morning Protocol (1-1.5 Hours)

1. Review experiments (15 min) — Check Ad Variations for significance, document winners, end completed tests
2. Analyze asset performance (20 min) — Review the Combinations Report (Ads > Assets > Combinations). High impressions does NOT equal high conversions — check both.
3. Launch next test (20 min) — Change ONE variable from the winning version
4. Review ad group health (15 min) — Confirm 2 RSAs per group, all slots filled, replace poor-performing assets
5. Document (15 min) — Record results, update testing backlog, plan next week

The Testing Flywheel

Month 1: Test CTAs. Winner: "Get Your Free Quote" (+12% CTR). Month 2: Test value props with winning CTA. Winner: "Same-Day Service" (+8% CVR). Month 3: Test trust signals. Winner: "500+ 5-Star Reviews" (+15% CVR). By month 4, every RSA element has earned its place through data, not intuition.

RSA Optimization Quick-Reference Checklist

Setup: 2 RSAs per ad group, all 15 headlines filled, all 4 descriptions filled, keywords in 3-4 headlines only, headlines across 5 categories, ad rotation set to "Do not optimize."

Headlines: Each highlights a different point, mix of lengths, 2+ with specific numbers, 2+ with clear CTAs, location in at least 1 (local businesses).

Descriptions: Unique angle per description (value, trust, offer, urgency), primary CTA in 2+ descriptions, specific details only, character limits maximized.

Pinning: Only pin with historical data, consider multi-pin for position 1, leave 2+ positions unpinned.

Testing: One experiment always running via Ad Variations, one variable per test, minimum 2-week duration, statistical significance before concluding, weekly Friday review cadence.

Common RSA Mistakes (and Fixes)

RSA Strategy by Business Type

Measuring RSA Success

Primary metrics: Conversion rate (higher = better), cost per conversion (lower = better), conversion volume (higher within budget), impression share (80%+ for non-brand). Secondary: CTR and Quality Score (7+ is good). Ignore Ad Strength.

The Combination Report

Access via Ads > Assets > Combinations. Use it to spot headlines that never appear (replace them), catch high-impression combinations with low conversion rates, and find hidden winners that convert well despite fewer impressions. Always cross-reference with conversion data — Google shows what gets clicks, not what converts.

Display Path Optimization

Display paths (www.yoursite.com/path-1/path-2) are 15 characters each and don't need to match actual URLs. They help users understand what they'll find.

Strategies: Service + Location (/plumbing/houston), Service + CTA (/emergency/free-quote), Service + Differentiator (/roofing/licensed-pros), Service + Offer (/dental/new-patient-99).

Test display paths via Ad Variations — a 5% CTR improvement means 5% more traffic at the same budget.

The RSA + Quality Score Connection

Ad Strength doesn't impact Quality Score, but RSAs influence QS through Expected CTR (higher-CTR headline combinations improve this) and Ad Relevance (keyword headlines drive this). Landing Page Experience isn't directly affected, but message match between RSA and landing page matters.

The feedback loop: Better RSAs > Higher CTR > Better QS > Lower CPC > More budget > More testing data > Even better RSAs. That's why the weekly cadence matters.

Average Quality Score across all advertisers: 6.2 out of 10. One point better reduces CPC by approximately 16%. Aim for 7+ — below 5 means auditing your keyword-headline match and landing page alignment.

Up Next

Your ad copy is only as good as the page it sends traffic to. Landing Pages for Service Businesses: Message Match, Single CTA, and the 1-Second Rule covers how to build landing pages that convert the traffic your RSAs generate.

Your RSA Checklist

Run through this before you close this tab:

• 2 RSAs per ad group. Not 1, not 3. Pause the extras.
• All 15 headline slots filled. All 4 descriptions filled. No empty slots.
• Keywords in 3-4 headlines only (20-30%). The rest: benefits, CTAs, trust signals, differentiators.
• Stop chasing Ad Strength. It has zero impact on Ad Rank or Quality Score. Measure conversion rate and cost per conversion instead.
• Pin strategically — or don't pin at all. If you have historical data showing a winning headline, pin it to position 1. If you're still learning, leave everything unpinned.
• Set ad rotation to "Do not optimize." This ensures both RSAs get equal data during testing.
• Launch an Ad Variations experiment. One variable. 50/50 traffic split. Minimum 2 weeks. Start with CTA language — it typically has the highest impact.
• Set up a Friday morning review cadence. 1-1.5 hours. Check running experiments, review asset performance, launch the next test, document results.
• Cross-reference the Combinations Report with conversion data. High impressions does not mean high performance — Google shows what gets clicks, not what converts.
• Test display paths. Small CTR wins from /free-quote vs. /emergency-service compound over time.