A fictional incident-response software company loads nine headlines into a responsive search ad. Two name the category. The rest cover benefits, integrations, and calls to action. With every headline left free, one valid combination is:
Start a guided trial | Keep every owner aligned | Connect Slack and Teams
Nothing in that line is false. It simply fails the most basic test: a searcher cannot tell what is being advertised. The usual reaction is to pin a complete three-headline sentence. That restores order, but it also throws away most of the flexibility the ad format was built to use.
The better answer is narrower: pin the requirement, not your favorite full ad. Put the message that truly must survive in a guaranteed position, give that position interchangeable alternatives only when they are genuinely equivalent, and write every remaining headline so it can stand beside any other eligible headline without creating a dangling phrase, a contradictory offer, or an unclear audience.
Pinning is therefore not a universal performance tactic. It is a message-control tool. The job is to use the smallest amount of control that prevents a material failure.
Open the original RSA Pinning Combination Lab to switch between free, hybrid, and corrected versions of the three fictional sets used below.
What headline pinning controls now
Google currently allows up to 15 headlines and four descriptions in one responsive search ad. The creation flow requires at least three headlines and two descriptions, while an individual serving can contain fewer assets; Google describes a minimum of one headline and one description, with up to three headlines in the traditional headline area. Headline fields are limited to 30 characters and description fields to 90. These are format rules, not a promise that every supplied field will appear. See Google's current responsive search ads documentation.
The practical pinning rules are more specific than “fixed copy” suggests:
| Question | Current documented behavior | Practical consequence |
|---|---|---|
| What happens when one headline is pinned to H1? | That headline is eligible only for H1, and other headlines are prevented from taking its place. | One exact H1 becomes mandatory whenever the ad serves. |
| What if two or three headlines are pinned to H1? | Google can choose among the headlines assigned to that same position. | Use this only when each alternative independently satisfies the requirement. |
| Can unpinned headlines still appear? | Yes, unless every available headline position has been filled by pinned pools. | A single H1 pin can preserve substantial freedom in H2 and H3. |
| Is H3 safe for mandatory wording? | No. Google says pinned H1, H2, and Description 1 continue to show in their positions, while H3 and Description 2 are not guaranteed. | Do not hide a required qualifier, price condition, or disclaimer in H3. |
| Does a preview prove every possible output? | No. Google says the preview shows potential ads, not all combinations, and some formats can shorten text. | Audit the asset system, not just the preview cards you happen to see. |
There is another current wrinkle. Under what Google calls enhanced flexibility, an unused RSA headline can appear at the beginning of the description or as a link-based asset, and unused text from another active RSA in the same ad group can also be used in certain link-based spaces. Google says assets pinned to H1, H2, or Description 1 remain in their designated positions. That makes two reviews necessary: the ad-level combination review and an ad-group-level review for stray text that would be awkward when reused elsewhere. The same Google RSA page documents those placements and restrictions.
Pin the requirement, not the preferred reading
Before touching the pin icon, classify the message you are trying to protect.
An essential requirement is text whose absence would make the ad materially wrong: an exact compliance statement, a qualifying condition on an offer, the product category when every other line is generic, or a segment label needed to avoid attracting the wrong buyer. A strong preference is different. Brand first, benefit second, CTA third may be the team's preferred narrative, but the ad is not necessarily false or unintelligible if those ideas move.
That distinction matters because pinning charges a real flexibility cost. Pinning one exact line to H1 is a small restriction. Pinning one exact line to H1, another to H2, and another to H3 reduces a three-headline presentation to one sequence—and H3 can still disappear. Full pinning may be justified, but “this is how the copy deck reads” is not the same as “the customer must see this every time.”
The strongest RSA copy is modular. Each headline should carry a complete unit of meaning. It should not need the line before it to finish its grammar, the line after it to limit its promise, or a particular audience headline to stop it from contradicting another offer.
Three failures recur:
- Omission: every surviving headline is plausible, but the product, audience, or required term disappears.
- Dependency: one asset is a fragment or an unqualified claim, such as “Save 25%” separated from “On annual plans.”
- Collision: two individually valid assets refer to incompatible segments, packages, or conditions and should never have shared one RSA.
The next three sets show why those failures require different repairs.
Set 1: preserve the category with one pinned pool
RelayOps is a fictional incident-response platform. Its benefit headlines are specific enough for people who already know the product, but several combinations can omit the category entirely.
| Headline | Characters | Role | Recommended status |
|---|---|---|---|
| Incident response software | 26 | Category | Pin to H1 |
| Incident management platform | 28 | Equivalent category | Pin to H1 |
| Resolve incidents faster | 24 | Outcome | Free |
| Keep every owner aligned | 24 | Collaboration benefit | Free |
| Automate status updates | 23 | Feature benefit | Free |
| Built for on-call teams | 23 | Audience | Free |
| Connect Slack and Teams | 23 | Integration | Free |
| See the incident timeline | 25 | Product capability | Free |
| Start a guided trial | 20 | CTA | Free |
All free
Start a guided trial | Keep every owner aligned | Connect Slack and Teams
Problem: the combination is readable, but the essential category is absent. The ad could describe project management, customer support, or dozens of other tools.
Incident response software | Resolve incidents faster | Start a guided trial
Clear: this combination happens to include the category, but leaving all assets free does not require it.
Minimal control
Pin both category alternatives to H1 and leave the other seven headlines free:
Incident response software | Automate status updates | Start a guided trial
Incident management platform | See the incident timeline | Built for on-call teams
Both H1 alternatives independently identify the product. H2 and H3 can still combine benefits, integrations, audience cues, and CTAs. This is the useful version of Google's advice to pin multiple unique assets to the same position: the alternatives are not cosmetic rewrites; each fulfills the same business requirement.
The flexibility cost can be illustrated with a deliberately simplified count. If exactly three distinct headlines are shown and we ignore all other serving behavior, nine all-free headlines create 9 × 8 × 7 = 504 ordered three-headline sequences. Two H1-pinned category alternatives plus seven free headlines create 2 × 7 × 6 = 84. The reduction is substantial, but the ad still has 84 synthetic sequences rather than one. These numbers are teaching arithmetic, not a prediction of how many combinations Google will actually serve.
Unnecessary full pinning
Incident response software [H1] | Resolve incidents faster [H2] | Start a guided trial [H3]
This is clear, but it preserves only one three-headline sequence. It also treats the H3 CTA as though it were guaranteed, which Google's documentation says it is not. Full pinning has solved more than the actual problem.
Set 2: keep the offer and its qualifier in one headline
BrightBoard is fictional team-planning software. Its first draft splits a promotion into separate assets:
| Weak free pool | Characters | Repaired pool | Characters | Repaired status |
|---|---|---|---|---|
| Save 25% | 8 | Save 25% on annual plans | 24 | Pin to H1 |
| On annual plans | 15 | Team planning software | 22 | Free |
| Start month to month | 20 | Plan across every project | 25 | Free |
| Cancel anytime | 14 | Keep owners and dates clear | 27 | Free |
| Team planning software | 22 | Move work from spreadsheets | 27 | Free |
| Move projects in minutes | 24 | Compare plans and pricing | 25 | Free |
| Keep work in one place | 22 | Start a guided trial | 20 | Free |
| Try every core feature | 22 | Built for growing teams | 23 | Free |
| See plans and pricing | 21 | One view for every deadline | 27 | Free |
All free
Save 25% | Start month to month | Cancel anytime
Problem: the promotion applies to annual plans, yet the visible combination places it beside a monthly-plan message and omits the qualifier. Even if every statement has a defensible interpretation on the pricing page, the assembled claim is misleadingly broad.
On annual plans | Keep work in one place | See plans and pricing
Problem: “On annual plans” is a dependent fragment. The reader has to guess what happens on annual plans.
A two-position workaround would pin “Save 25%” to H1 and “On annual plans” to H2. Current documentation supports H1 and H2 as guaranteed pinned positions, but this still consumes two premium slots to express one claim. It also leaves a fragile copy system that becomes unsafe the moment someone removes the second pin or reuses the fragment elsewhere.
Repair the asset before pinning it
Combine the promise and condition:
Save 25% on annual plans [H1] | Team planning software | Compare plans and pricing
Save 25% on annual plans [H1] | Move work from spreadsheets | Built for growing teams
Now the offer is complete inside one 24-character headline. One H1 pin protects it, while the remaining assets can rotate without changing the deal. The monthly-plan and broad “Cancel anytime” lines have been removed from this offer-led RSA because they invite a second question the ad does not need to answer.
This is the key copy rule: pinning can control position, but it cannot turn two weak fragments into one robust message. When a claim and its qualifier fit in the same field, keep them together. When they do not fit, use a guaranteed description position or rewrite the claim; do not assume H3 will complete it.
Set 3: pinning cannot reconcile incompatible audiences
Northstar Workspace is a fictional client portal sold in both enterprise and solo-consultant packages. The first draft combines both motions in one RSA:
| Mixed headline pool | Characters |
|---|---|
| Secure client workspace | 23 |
| For enterprise IT teams | 23 |
| Built for solo consultants | 26 |
| Free setup for 10+ seats | 24 |
| No seat minimum | 15 |
| Book a security review | 22 |
| Start online in minutes | 23 |
| Centralized admin controls | 26 |
| Simple client approvals | 23 |
All free
For enterprise IT teams | Built for solo consultants | No seat minimum
Problem: the ad addresses two different buyers in one breath.
Free setup for 10+ seats | No seat minimum | Start online in minutes
Problem: “10+ seats” and “no seat minimum” may both be true for different packages, but together they blur which condition applies.
Category pin, conflict intact
Pinning “Secure client workspace” to H1 does not repair the pool:
Secure client workspace [H1] | For enterprise IT teams | Built for solo consultants
Secure client workspace [H1] | Free setup for 10+ seats | No seat minimum
The product is now clear, yet the audience and terms still collide. This is not a pinning problem. It is an asset-set problem.
Split the pool
Create separate ad groups when the audience messages must stay isolated. Separate RSAs in one ad group can still share unused text through enhanced flexibility; their traditional headline pools alone are separate.
| Enterprise RSA | Characters | Solo-consultant RSA | Characters |
|---|---|---|---|
| Client workspace for IT teams [H1] | 29 | Workspace for solo consultants [H1] | 30 |
| For teams with 10+ seats | 24 | No seat minimum | 15 |
| Free guided setup | 17 | Start online in minutes | 23 |
| Centralized admin controls | 26 | Simple client approvals | 23 |
| Book a security review | 22 | Keep each client organized | 26 |
| Manage every client space | 25 | Share files and decisions | 25 |
| Set roles and permissions | 25 | Organize files by client | 24 |
| Consolidate client access | 25 | Try the solo plan | 17 |
| See enterprise pricing | 22 | See solo pricing | 16 |
Each H1 now carries category and audience together. Every free enterprise line can sit beside every other enterprise line; the same is true for the solo set. Pinning preserves the essential audience message, but segmentation does the work that pinning could not.
A practical pinning decision table
| Situation | Better action | Why |
|---|---|---|
| Exact wording must appear in every ad | Pin the exact text to H1, H2, or Description 1 after checking length and policy requirements. | Those are the documented guaranteed pinned positions. |
| The category or audience must appear, but several phrasings are acceptable | Pin two or three genuinely equivalent, unique assets to the same guaranteed position. | The requirement survives while the system retains a choice. |
| A claim depends on a qualifier | Put the claim and qualifier in the same asset where possible, then pin that complete asset only if it must always appear. | Position control is weaker than self-contained meaning. |
| The team merely prefers a certain story order | Rewrite assets so they work in any eligible order before adding pins. | A preferred narrative is not automatically a serving requirement. |
| Two assets describe different packages, geographies, or audiences | Separate the asset pools, ads, or ad groups. | Pinning one line does not stop incompatible free assets from meeting elsewhere. |
| A CTA would be nice in the third headline | Leave several CTAs free unless one is genuinely mandatory. | H3 is not guaranteed, so the ad must still work without it. |
One exact pinned asset is appropriate when exactness matters more than variation. Multiple same-position pins are appropriate only when every alternative preserves the same meaning. “Official website,” “Visit the official site,” and “BrandName official site” may be equivalent for one requirement; “Save 25%” and “Flexible monthly plans” are not.
Ad Strength is feedback, not the verdict
Pinning can reduce Ad Strength because it removes eligible combinations. Google recommends using pinning sparingly and supplying several unique alternatives to a pinned position when the requirement allows it. At the same time, Google explicitly says the Ad Strength rating does not determine whether an ad is eligible to serve and is not used to calculate Ad Rank, Quality Score, or auction wins. Read the current Ad Strength documentation as creative-coverage guidance, not as a substitute for business outcomes.
External performance evidence also resists a universal rule. In April 2026, Optmyzr published an observational analysis of roughly 20,000 active customer accounts. Partial pinning led several aggregate metrics in the full dataset, but the smaller group of 268 accounts using all three strategies produced a more mixed pattern, including different leaders for ROAS and conversion rate. The authors label the work observational, note customer-selection bias, and say the findings are directional rather than population-wide absolutes. That makes the study useful as a testing hypothesis—not proof that partial pinning will improve every account. See Optmyzr's 2026 RSA performance study.
The sensible editorial position is therefore asymmetric:
- Do not remove a necessary pin just to improve Ad Strength.
- Do not add a pin because a broad industry average appears favorable.
- Do remove unnecessary pins that exist only to preserve a copywriter's preferred sentence.
- Judge the remaining strategy against the campaign's actual conversion and value goals.
Audit the combinations before launch
A preflight review should be semantic, not just visual.
First, label every asset by job: category, brand, audience, benefit, proof, offer, qualifier, CTA, or compliance. Then mark which ideas are required and which are merely useful. Required ideas need a guaranteed position or a different ad structure; useful ideas need compatible neighbors.
Next, read combinations in pairs before triples. Pair testing exposes most collisions faster: “Save 25%” + “Start month to month,” “For enterprise IT teams” + “Built for solo consultants,” or two near-duplicate benefits that waste scarce space. Then test three-line sequences, including sequences that omit the category, brand, offer, or CTA. Ask four questions of every sample:
- What is being sold?
- Who is it for?
- Is every promise still correctly qualified?
- Would the ad remain coherent if H3 disappeared?
Do not approve a set merely because you found several good combinations. The important question is whether the eligible pool still contains a materially bad one. Preview screens are useful for spotting layout issues, but Google's own documentation says they do not display every combination.
Finally, review the rest of the ad group. With enhanced flexibility, unpinned text from another active RSA can appear in certain link-based placements. A clean ad can still inherit an awkward neighbor if the ad group mixes different promotions or audiences.
Measure pinning without pretending one asset acted alone
After launch, use the reporting layers for different questions.
The ad-level asset report shows the text, asset type, position-pinning information, and performance metrics. Google says full asset performance statistics are available for dates on or after June 5, 2025. It also warns that asset-level metrics are non-summable and that ratios such as CTR, CPA, and ROAS are directional because every asset appears as part of a combination. Use the report to find assets with no delivery, verify pin settings, and form hypotheses—not to declare one headline independently causal. See Google's ad-level asset report guidance.
The combinations report answers a narrower question: which assembled combinations received impressions and which appeared most often. Google cautions against converting a common RSA combination into a static ad and assuming it will perform the same. Use the combinations report to catch live readings that escaped preflight and to confirm whether your intended message structures are actually appearing.
When the account has enough volume to evaluate a structural change, an ad variation can compare a modified RSA treatment with the original across a defined scope and traffic percentage. Google describes ad variations as suitable for testing responsive search ads or one change across multiple Search campaigns. Keep the test interpretable: change the pinning system, not the pinning system plus offer, landing page, bid strategy, and audience at the same time. See Google's ad variations overview.
A useful result is not “the pinned headline had the best CPA.” A more defensible result is “the treatment that guaranteed the category preserved qualified conversion volume at an acceptable cost,” or “removing the full three-position pin improved value without creating unsafe combinations.” Those are strategy-level conclusions matched to the way RSAs actually assemble assets.
The final rule
Start with language architecture, not the pin icon. Remove fragments. Join claims to their qualifiers. Separate audiences and offers that should never meet. Then pin the smallest complete message that must survive, using H1, H2, or Description 1 when it truly has to appear every time.
A strong RSA is neither completely uncontrolled nor mechanically fixed. It has a protected spine and flexible, compatible parts. The spine keeps the ad honest. The parts give the system room to adapt without turning the message into nonsense.
Sources
- About responsive search ads — Google Ads Help. Current format, pinning, position guarantees, preview limitations, and enhanced-flexibility behavior; checked September 9, 2026.
- About Ad Strength for responsive search ads — Google Ads Help. Ad Strength purpose, serving-eligibility distinction, and pinning guidance; checked September 9, 2026.
- About the ad-level asset report for responsive search ads — Google Ads Help. Pinning fields, full performance statistics, and asset-metric limitations; checked September 9, 2026.
- About the combinations report for responsive search ads — Google Ads Help. Combination impressions, common combinations, and interpretation warning; checked September 9, 2026.
- About ad variations — Google Ads Help. Scope and use of ad-variation testing; checked September 9, 2026.
- We Analyzed ~20,000 Google Ads Accounts. Here’s What Actually Drives RSA Performance (Hint: It’s Not Ad Strength) — Optmyzr, April 6, 2026. Observational RSA performance analysis and stated caveats; checked September 9, 2026.
