Most advertisers write RSA headlines by gut feel. They stare at the screen, type something that sounds decent, and hit save.
The result: 90% of RSAs underperform because Google's machine learning is optimizing across headlines that all say the same thing in different words.
This rubric gives you an objective scoring system. Every headline gets scored against 6 criteria. Minimum threshold: 70/100. Generate 25 candidates per RSA. Keep the top 15.
Based on $38M+ in managed search spend across 90+ accounts. The difference between scored headlines and first drafts: 1.8x to 3.4x ROAS.
This is binary. All three rules pass, or you score zero. No partial credit.
Rule 1 - Visualizable: Can the reader picture the benefit?
Pass: "Monday to Friday, One Charge" (you picture a work week without plugging in) Pass: "From Dull to Gleaming in 30 Min" (you picture the before/after) Fail: "Premium Sound Awaits" (what does "premium sound" look like?) Fail: "Quality Products Available" (no mental image created)
Rule 2 - Falsifiable: Is there a verifiable claim?
Pass: "10,000+ Cars Protected" (can be checked) Pass: "Rated 4.9/5 by 2,800 Detailers" (verifiable number) Fail: "Trusted by Many" (how many? Can't verify) Fail: "Best Quality Available" (subjective, can't disprove)
Rule 3 - Unique: Could a competitor use this headline word-for-word?
Pass: "5-Layer Ceramic with SiO2" (specific to your formula) Pass: "$49 Replaces $600/Year Waxing" (specific to your price point) Fail: "Free Shipping on All Orders" (any store could say this) Fail: "Premium Skincare Products" (every skincare brand could claim this)
Scoring: All 3 pass = 30 points. Any one fails = 0 points. This is intentionally harsh. A headline that fails the filter needs perfect scores on all remaining criteria to reach 70/100 - which almost never happens.
Does the headline contain theme words from the keyword cluster? Does it match the search intent?
| Score | Description |
|---|---|
| 18-20 | Contains primary cluster keyword or close variant. Directly addresses search intent. |
| 14-17 | Contains related theme words. Intent alignment is clear but indirect. |
| 10-13 | Loosely connected to cluster theme. Reader has to make the connection. |
| 5-9 | Tangentially related. Could fit multiple clusters. |
| 0-4 | No connection to cluster keywords or intent. Wrong ad group. |
Example: Cluster theme is "ceramic car coating protection"
Can the reader understand this headline in under 2 seconds? At 30 characters maximum, every word must carry weight.
| Score | Description |
|---|---|
| 13-15 | Instant understanding. No jargon. No ambiguity. A 12-year-old gets it. |
| 10-12 | Clear to the target audience. Might use industry terms the buyer knows. |
| 7-9 | Requires a moment of thought. Slightly clever or abstract. |
| 4-6 | Confusing. Wordplay that doesn't land. Multiple possible interpretations. |
| 0-3 | Incomprehensible or misleading. Reader doesn't know what you're selling. |
Example:
Is this claim backed by data, evidence, or information on your landing page?
| Score | Description |
|---|---|
| 13-15 | Claim is directly verifiable on the landing page (review count, warranty terms, test results). |
| 10-12 | Claim is supported but not directly stated (implied by product specs or description). |
| 7-9 | Claim is reasonable but no landing page evidence supports it. |
| 4-6 | Claim is a stretch. Landing page would need updates to support it. |
| 0-3 | Claim is fabricated or cannot be verified anywhere. |
Example:
Why this matters: Headlines that promise things the landing page doesn't deliver tank conversion rates. The click got them there. The disconnect pushes them away. Google also factors landing page relevance into Quality Score.
Does this headline only work for YOUR product and YOUR landing page?
| Score | Description |
|---|---|
| 9-10 | Could only be used by your brand. Contains specific claims, numbers, or features unique to you. |
| 7-8 | Mostly unique. One or two competitors might also claim this. |
| 4-6 | Generic within your category. Most competitors in your space could use it. |
| 1-3 | Completely generic. Works for any e-commerce brand in any category. |
Example:
Would someone stop scrolling and click? Does this headline create an emotional response - curiosity, relief, urgency, surprise?
| Score | Description |
|---|---|
| 9-10 | Scroll-stopper. Creates immediate curiosity or desire. "I need to know more." |
| 7-8 | Strong pull. Addresses a clear pain point or desire. Likely to earn the click. |
| 5-6 | Decent. Communicates value but doesn't create urgency or curiosity. |
| 3-4 | Flat. Factual but no emotional hook. Reader moves past it. |
| 0-2 | Invisible. Blends into every other ad on the page. |
Example:
Use this template for every headline candidate. Score all 25 candidates before selecting your top 15.
Headline: [Your headline here]
Character count: [Must be 30 or fewer]
1. 3-Rule Filter: __/30 (0 or 30 only)
Visualizable? Y / N
Falsifiable? Y / N
Unique? Y / N
2. Keyword Relevance: __/20
3. Clarity: __/15
4. Proofability: __/15
5. Uniqueness: __/10
6. CTR Potential: __/10
TOTAL: __/100
Decision: KEEP (70+) / CUT (below 70)
Headline: "One Coat. Two Years. Done." Character count: 25/30
| Criteria | Score | Reasoning |
|---|---|---|
| 3-Rule Filter | 30/30 | Visualizable (picture applying one coat). Falsifiable (2 years is measurable). Unique (specific duration claim). |
| Keyword Relevance | 16/20 | Implies durability and coating. Doesn't contain exact cluster keyword but intent is clear. |
| Clarity | 15/15 | Instant understanding. Three fragments, zero ambiguity. |
| Proofability | 14/15 | 2-year warranty on product page. Landing page supports the claim. |
| Uniqueness | 9/10 | Specific to products with 2-year durability. Most competitors can't make this claim. |
| CTR Potential | 8/10 | Creates a contrast between effort (one coat) and result (two years). The "Done." finisher adds decisiveness. |
| Total | 92/100 | KEEP |
Headline: "Your Paint Deserves Better" Character count: 24/30
| Criteria | Score | Reasoning |
|---|---|---|
| 3-Rule Filter | 0/30 | Visualizable: barely (what does "better" look like?). Falsifiable: no ("better" is subjective). Unique: no (any paint protection product). Failed - zero. |
| Keyword Relevance | 12/20 | References "paint" but doesn't address protection or coating. |
| Clarity | 13/15 | Clear message, just vague. |
| Proofability | 5/15 | What claim is being made? Nothing to verify. |
| Uniqueness | 3/10 | Any car care brand could use this. |
| CTR Potential | 6/10 | Creates mild emotional response but nothing specific. |
| Total | 39/100 | CUT |
The 3-Rule Filter failure made recovery impossible. Even with generous scores on everything else, it can't reach 70.
Headline: "Free Shipping + Money-Back" Character count: 27/30
| Criteria | Score | Reasoning |
|---|---|---|
| 3-Rule Filter | 0/30 | Visualizable: no (what's the mental image of free shipping?). Falsifiable: yes (policy is verifiable). Unique: no (thousands of stores offer this). Failed on 2 of 3 - zero. |
| Keyword Relevance | 4/20 | No connection to any product-specific cluster. |
| Clarity | 15/15 | Crystal clear. |
| Proofability | 15/15 | Both policies on landing page. |
| Uniqueness | 1/10 | Literally any e-commerce store. |
| CTR Potential | 5/10 | Nice-to-know but not a reason to click YOUR ad. |
| Total | 40/100 | CUT |
This is the most common "but it's a good headline!" objection we hear. Free shipping and money-back guarantees are table stakes. They belong in the description as supporting evidence, not as a headline competing for one of your 15 spots.
After scoring individual headlines, validate across your 3 RSAs.
The test: Read all 15 headlines from RSA 1 out loud. Then read all 15 from RSA 2. Do they feel like ads for different audiences? If yes, your differentiation works. If they feel like the same ad rewritten three times, start over.
| Total Score | Decision | Action |
|---|---|---|
| 90-100 | Strong keep | Assign to RSA immediately |
| 80-89 | Keep | Solid candidate, include in top 15 |
| 70-79 | Conditional keep | Include only if needed to fill 15 |
| 60-69 | Rework | Has potential but needs revision on weak criteria |
| Below 60 | Cut | Don't force it. Generate a new candidate. |
This rubric scores individual headlines. It tells you whether a headline earns its spot in your RSA.
What it doesn't cover:
The Google Ads AI Agentic System ($4,997+) covers the complete system end to end. The rubric gets you better headlines today. The Agentic System builds the architecture that produces 1.8x to 3.4x ROAS across your entire search program.
This rubric is the quality gate. The Agentic System is the machine that feeds it.