Google Ads copy in 2026 is harder to write than it was three years ago — and easier to get wrong. The platform has changed significantly: Responsive Search Ads (RSAs) are the default format, AI-generated assets are being inserted automatically, and the relationship between your copy and your Quality Score is less transparent than it used to be. Yet Choco Media‘s paid media team still sees the same google ads copywriting mistakes made repeatedly by smart advertisers. This post is for marketers and founders who want to know what actually moves the needle on RSA performance right now — headline formulas, asset variety, Quality Score signals, and how to write copy that holds up when Google’s automation is filling in the blanks.
RSAs give Google up to 15 headlines and 4 descriptions to combine. In theory, more assets mean more testing and better performance. In practice, advertisers often load all 15 slots with slight variations of the same message — and wonder why Performance Planner shows low ad strength. The issue isn’t the format. It’s the writing strategy behind it.
If you manage your own Google Ads account or are evaluating an agency’s output, this post will give you a concrete framework for auditing what’s there and improving it.
Why Google Ads Copywriting Changed After Smart Bidding Became Default
The shift to Smart Bidding — particularly Target CPA and Maximize Conversions — changed how copy works. Previously, a well-structured manual campaign could compensate for weak creative through tight bid management. Now, the algorithm handles bid decisions, and your copy’s job is simpler and harder at the same time: generate clicks from the right people and suppress clicks from the wrong ones.
This changes what “good copy” means. High CTR is no longer always the goal. Copy that attracts irrelevant clicks will lower your conversion rate, which degrades Quality Score and pushes up CPCs. The best RSA headlines in 2026 are self-qualifying — they attract the person most likely to convert and repel those who aren’t a fit.
- Include pricing in headlines if your offer is priced above the average competitor — this pre-qualifies budget
- Name the specific problem or outcome, not the category (“reduce checkout abandonment” vs. “e-commerce tools”)
- Use qualifier language like “for B2B teams” or “for growing agencies” to filter for intent
- Avoid clickbait patterns — they inflate CTR but crater conversion rate and waste Smart Bidding learning
What Quality Score actually rewards now
Quality Score is a composite of Expected CTR, Ad Relevance, and Landing Page Experience. The copy-specific elements are the first two. Expected CTR is Google’s prediction of how often your ad gets clicked when shown — based on your account history, format, and keyword match. Ad Relevance measures how closely your headlines and descriptions match the intent behind a search query.
The practical implication: keyword insertion and strong message-query match still matter. But the way you achieve that match has shifted. Rather than keyword-stuffing, write for the intent behind the keyword — and use dynamic keyword insertion (DKI) selectively for high-volume generic terms, not as a substitute for real copy thinking.
RSA Headline Formulas That Consistently Perform
After running dozens of RSA tests across different verticals, we’ve settled on a set of headline patterns that give Google meaningful variety without diluting the message. The goal is to cover different decision triggers — problem awareness, solution framing, credibility, and urgency — without repeating the same angle with different words.
- Problem/pain headline: “Ads Burning Budget? Here’s Why” — meets the user at the point of frustration
- Outcome headline: “Lower CPL Without Cutting Spend” — specific, measurable, desirable outcome
- Proof/credibility headline: “Used by 200+ E-Commerce Brands” — social proof in the headline slot
- Qualifier headline: “For Growing B2B SaaS Teams” — narrows the audience to improve relevance
- Feature/differentiator headline: “No Long-Term Contracts” — removes a common objection
- Urgency/scarcity headline: “Onboarding Open: 3 Spots Left” — works best when it’s true
- Keyword-match headline: “[Keyword] Agency in Helsinki” — maintains query relevance for broad match
When you load 15 headlines, aim for at least 2-3 from each trigger category. Google will test combinations, but it can only find winners if the inputs have meaningful variation. Fifteen slight variations of your value proposition is not variation — it’s noise.
Description field strategy
You get 4 description slots at 90 characters each. Most advertisers write two and leave two blank. That’s a missed opportunity. Use all four, each with a different function: one for the primary value proposition, one for proof/credibility, one that addresses a common objection, and one with a clear CTA. Google will test combinations, and the more you give it to work with, the better the learning data.
Ad Strength: What It Means and What It Doesn’t
The Ad Strength indicator (Poor / Average / Good / Excellent) is frequently misunderstood. Advertisers optimise for Excellent ad strength as if it directly correlates with performance. It doesn’t — at least not in a simple way. Ad Strength is Google’s assessment of asset variety and volume, not a prediction of conversion rate.
We have run accounts where an RSA with “Good” ad strength significantly outperformed an “Excellent” ad strength RSA because the Good-strength ad had tighter message-query alignment on a specific campaign objective. Ad Strength is a useful signal for checking whether you’ve provided enough variety, but it should not override your copy strategy.
“Ad strength tells you whether Google has enough material to work with. It doesn’t tell you whether that material is actually good. Those are different problems.”
Use Ad Strength as a floor check — aim for at least Good, then focus your optimisation energy on conversion rate, CPC trends, and asset-level performance data (which shows impressions per headline).
- Check asset-level reporting in the “Assets” tab under individual ads — this shows which headlines get the most impressions
- Low-impression headlines may be too similar to higher-performing ones — prune or replace them
- Pin headlines sparingly — pinning a headline forces Google to show it, which reduces testing flexibility
Pinning: When It Helps and When It Costs You
Pinning allows you to fix a specific headline to position 1, 2, or 3 within the ad. Agencies overuse it because it gives a sense of control. The cost is reduced combination testing — when you pin headline position 1, you’re telling Google to always show that headline first, regardless of what combination would perform best for a given query.
There are legitimate use cases for pinning. If your brand name must always appear in position 1 for brand protection, pin it. If legal requires a specific disclosure in every ad shown for a regulated product, pin it. Beyond those cases, pinning is usually a way to impose human preferences onto a system that often performs better without them.
- If you pin, always pin multiple headlines to the same position to preserve some testing variety
- Never pin your only strong proof-point or CTA to a single position — you’ll suppress its ability to appear in other high-value slots
- Review pinned assets quarterly — what made sense 12 months ago may no longer be the strongest message
Our own approach: we pin brand name in position 1 on brand campaigns and leave everything else unpinned. On non-brand campaigns, we let the algorithm decide position unless there’s a hard compliance requirement. For a deeper look at the full account structure that supports this, see our post on how we audit a Google Ads account — the copy layer is only as effective as the structural layer beneath it.
Writing for Broad Match in 2026
Broad match has gone from a liability to a viable option because Smart Bidding now uses first-party signals (audience lists, remarketing data, customer match) to control which queries trigger your ads. The copy implications are significant: if your ads will appear across a wider range of queries, they need to be more intent-forward, not keyword-forward.
Headlines written primarily for keyword density perform poorly with broad match because they become irrelevant quickly across the expanded query space. Headlines written for the underlying problem or desire perform better because they stay relevant even as the triggering query varies.
- Include audience-specific language in descriptions (“for small marketing teams”, “for Finnish B2B brands”) to help Google match your ad to the right user signals
- Layer broad match with audience signals, not just negative keywords
- Monitor the search terms report weekly — broad match requires more active query management than exact
- Use one broad match ad group per campaign, not as the sole bidding strategy across all ad groups
Negative keyword management as a copy support system
Negative keywords are part of your copy strategy, not separate from it. If your copy is perfectly written but your ads appear for irrelevant queries, your conversion rate suffers and your copy appears to underperform. We see this consistently in account audits — our 90-minute paid media audit process treats negative keyword hygiene as a first-pass diagnostic before we even look at headlines.
How Google’s Automated Assets Change Your Copy Strategy
In 2025, Google began more aggressively auto-generating ad assets — including headlines — from your landing page content and previous ad performance. By default, these AI-generated assets can appear in your ads without explicit approval. For most advertisers, this is turned on and quietly diluting their copy quality.
The generated assets are often grammatically correct but generically written. They pull language from your landing page and flatten it into copy that sounds like every other ad in your category. More importantly, they ignore your brand voice, your specific proof points, and your targeting strategy.
- Check Settings → Ad Rotation and Assets → Automatically created assets — review and, if needed, disable
- If you keep auto-assets on, audit them monthly in the Assets tab and remove any that dilute your message
- The better your manual RSA assets, the less frequently Google falls back to auto-generated ones
This is one area where the platform’s automation works against you unless you actively manage it. We treat auto-asset review as a monthly task in our client accounts — it often surfaces headlines we’d never have approved and gives a useful picture of what Google thinks your value proposition is (which is not always what you intend).
Message Match: The Bridge Between Your Ad and Your Landing Page
Message match is the degree of alignment between what your ad promises and what your landing page delivers. It is one of the highest-leverage optimisations in paid media — and one of the most frequently overlooked. You can write excellent headlines and still lose the conversion if the landing page opens with different language, a different offer, or a different emotional register.
For RSAs specifically, message match is complicated by the fact that different headline combinations will produce different promises. If your ad can combine a “30-day free trial” headline with a “no credit card required” description in one combination, and a “Book a Demo” headline with “Used by 200+ Teams” description in another, your landing page needs to credibly support both. That requires a landing page written at the offer level, not the headline level.
- Write landing page hero copy around the primary value proposition, not a single ad headline
- Ensure the CTA on the landing page matches the dominant CTA pattern in your descriptions
- Use UTM parameters and URL customisers to send different RSA combinations to variant landing pages where volume allows
For a detailed walkthrough of how we approach the ad-to-page relationship, our work on conversion rate optimisation covers the full audit framework — copy alignment is step two, after offer clarity.
A Simple Testing Framework for RSA Copy
RSAs are self-testing by design, but that doesn’t mean you should load all 15 headlines at launch and never return. The most useful testing structure we’ve found: start with 8-10 headlines covering distinct trigger categories, run for 4-6 weeks with adequate impression volume, then check asset-level performance. Remove headlines with consistently low impressions (they’re losing to others), replace with new variants in the same trigger category, and repeat.
- Minimum impressions before evaluating: ~3,000 per headline asset (check the impressions column in asset view)
- Don’t make copy changes during Smart Bidding learning periods — wait until the campaign has exited “Learning”
- Keep a copy log in a Google Sheet: what was tested, when, what was replaced, why — this builds institutional knowledge
- Test one variable at a time if you’re pinning for a controlled test — otherwise, let Google run combinations freely
The frequency mistake we see most often: advertisers write their initial RSA assets, achieve Good or Excellent ad strength, and never return. Copy optimisation is iterative. The market shifts, competitors update their messaging, and the queries triggering your ads evolve. A quarterly copy review is the minimum cadence for accounts spending more than €1,000/month.
If you’re building or auditing a Google Ads account and want a structured view of where copy fits in the broader setup, we’re happy to take a look. You can reach us via the contact page — we work with brands running their own ads and those looking to hand off management entirely.