How many Responsive Search Ads (RSAs) are allowed in Google Ads?
The hard limit: 3 enabled RSAs per ad group
In Google Ads, you can have up to 3 enabled Responsive Search Ads per ad group. If you already have three RSAs set to enabled in that ad group, the platform will prevent you from enabling a fourth until you pause or remove one of the existing enabled RSAs.
This is one of those limits that matters operationally because it impacts how you structure creative testing. You’re not deciding whether RSAs “work” (they do, when built correctly); you’re deciding how to get the most learning and performance out of a maximum of three simultaneously serving RSA containers.
The related limit that can surprise you: 50 active text (and other non-image) ads per ad group
Separate from the “3 enabled RSAs” rule, Google Ads also limits you to 50 active text and other non-image/gallery ads per ad group. Most advertisers never get close to this, but larger legacy ad groups (especially those that carried a lot of older formats and experiments) can hit it, and then ad creation/enabling starts failing for reasons that look unrelated at first glance.
In practice, for most modern Search builds, the RSA-specific cap (3 enabled) is the one you’ll run into first—and it’s the one you should plan around when you design your testing approach.
What you should run in real life (best practice, not just limits)
Why “more RSAs” isn’t always better
Google’s own best-practice guidance trends toward keeping each ad group covered with at least one strong RSA and letting the system optimize combinations at query-time inside that RSA. The RSA itself already contains a lot of testable inventory because you can provide up to 15 headlines and 4 descriptions, which creates substantial variation without needing a large number of separate ads.
From 15+ years managing accounts, my rule is simple: don’t multiply ads when you should be improving assets. Most underperforming RSAs fail because the headlines are too repetitive, too generic, or too “feature heavy” without clear differentiation—so the system has limited meaningful choices.
A practical setup that works in most accounts
If you’re unsure where to start, aim for 1–2 enabled RSAs per ad group in most cases, and only use the third slot when you have a clear reason (for example: a tightly controlled promo message, a different value proposition, or a distinct landing page angle that needs its own creative framing). You’re staying within the 3-enabled limit either way, but you’re doing it with intention.
As you build each RSA, focus on “asset diversity” instead of minor word swaps. You want different themes (pricing, speed, quality, social proof, guarantees, inventory depth, location/service area, etc.), because that’s what gives the system real options to match different search intents.
How to maximize performance within the 3-RSA limit (a systematic approach)
Build RSAs like a testing framework, not a copy dump
Think of each enabled RSA as a container for a strategy. If all three RSAs are saying the same thing, you’re not gaining anything—you’re just splitting learning and making reporting noisier. A cleaner approach is to make each RSA “win” a different job, such as one RSA oriented around core benefits, another around proof/authority, and a third around a time-bound offer (when you truly need that separation).
Inside each RSA, use the available headline and description slots to cover the questions a high-intent searcher is silently asking: “Is this for me?”, “Why you?”, “What will it cost?”, “How fast?”, “What’s the catch?”, “How do I start?” You can’t control the exact assembly every time, but you can control whether any assembled version still reads like a coherent, persuasive ad.
Use Ad Strength as a build quality check (but don’t chase the label blindly)
Ad Strength is useful as a diagnostic because it pushes you toward more complete, more varied assets. As you add more unique headlines and descriptions (up to the RSA limits), you generally expand the system’s ability to find good combinations and improve relevance.
That said, “Excellent” is not the business goal—profitable conversions are. I’ve seen “Excellent” RSAs that were too broad and lowered lead quality. Use the indicator to catch obvious problems (thin assets, repetition, missing variety), then judge final success with conversion quality and incremental performance.
When you hit the limit: the only fixes that matter
If Google Ads won’t let you enable another RSA in an ad group, treat it like a quick triage:
- Count enabled RSAs in the ad group: if there are already 3 enabled, pause one to free a slot.
- Check total active non-image/text ads: if the ad group is unusually cluttered and near the 50 active ad limit, reduce active ad count.
- Decide what you’re trying to test: don’t pause the “wrong” RSA—replace the one with the least differentiated message or the one constrained by excessive pinning.
The strategic move is usually not “add another RSA,” but “retire the weakest message” and rebuild it into a stronger, more differentiated RSA or into better assets inside the remaining ads.
One final tip: avoid over-pinning unless you truly need it
Pinning can be necessary for compliance, brand requirements, or must-say elements, but heavy pinning can reduce the system’s flexibility and limit performance gains from RSA assembly. Use pinning surgically—only where it protects meaning or legal requirements—then let the rest of the asset set do its job.
Let AI handle
the Google Ads grunt work
Let AI handle
the Google Ads grunt work
In Google Ads, you can have up to 3 enabled Responsive Search Ads (RSAs) per ad group (and you may also run into the separate 50 active text/non-image ad limit in older ad groups), which is why many teams get better results by keeping 1–2 strong RSAs and focusing on improving asset variety—up to 15 headlines and 4 descriptions—rather than adding near-duplicates that split learning. If you want help staying efficient within those limits, Blobr connects to your Google Ads account and uses see-what-changed, always-on analysis plus specialized AI agents—like a Headlines Enhancer to refresh underperforming RSA assets with new angles, or a Best URL Landing Matcher to align ads with better-fitting landing pages—so you can iterate thoughtfully without turning RSA testing into manual busywork.
How many Responsive Search Ads (RSAs) are allowed in Google Ads?
The hard limit: 3 enabled RSAs per ad group
In Google Ads, you can have up to 3 enabled Responsive Search Ads per ad group. If you already have three RSAs set to enabled in that ad group, the platform will prevent you from enabling a fourth until you pause or remove one of the existing enabled RSAs.
This is one of those limits that matters operationally because it impacts how you structure creative testing. You’re not deciding whether RSAs “work” (they do, when built correctly); you’re deciding how to get the most learning and performance out of a maximum of three simultaneously serving RSA containers.
The related limit that can surprise you: 50 active text (and other non-image) ads per ad group
Separate from the “3 enabled RSAs” rule, Google Ads also limits you to 50 active text and other non-image/gallery ads per ad group. Most advertisers never get close to this, but larger legacy ad groups (especially those that carried a lot of older formats and experiments) can hit it, and then ad creation/enabling starts failing for reasons that look unrelated at first glance.
In practice, for most modern Search builds, the RSA-specific cap (3 enabled) is the one you’ll run into first—and it’s the one you should plan around when you design your testing approach.
What you should run in real life (best practice, not just limits)
Why “more RSAs” isn’t always better
Google’s own best-practice guidance trends toward keeping each ad group covered with at least one strong RSA and letting the system optimize combinations at query-time inside that RSA. The RSA itself already contains a lot of testable inventory because you can provide up to 15 headlines and 4 descriptions, which creates substantial variation without needing a large number of separate ads.
From 15+ years managing accounts, my rule is simple: don’t multiply ads when you should be improving assets. Most underperforming RSAs fail because the headlines are too repetitive, too generic, or too “feature heavy” without clear differentiation—so the system has limited meaningful choices.
A practical setup that works in most accounts
If you’re unsure where to start, aim for 1–2 enabled RSAs per ad group in most cases, and only use the third slot when you have a clear reason (for example: a tightly controlled promo message, a different value proposition, or a distinct landing page angle that needs its own creative framing). You’re staying within the 3-enabled limit either way, but you’re doing it with intention.
As you build each RSA, focus on “asset diversity” instead of minor word swaps. You want different themes (pricing, speed, quality, social proof, guarantees, inventory depth, location/service area, etc.), because that’s what gives the system real options to match different search intents.
How to maximize performance within the 3-RSA limit (a systematic approach)
Build RSAs like a testing framework, not a copy dump
Think of each enabled RSA as a container for a strategy. If all three RSAs are saying the same thing, you’re not gaining anything—you’re just splitting learning and making reporting noisier. A cleaner approach is to make each RSA “win” a different job, such as one RSA oriented around core benefits, another around proof/authority, and a third around a time-bound offer (when you truly need that separation).
Inside each RSA, use the available headline and description slots to cover the questions a high-intent searcher is silently asking: “Is this for me?”, “Why you?”, “What will it cost?”, “How fast?”, “What’s the catch?”, “How do I start?” You can’t control the exact assembly every time, but you can control whether any assembled version still reads like a coherent, persuasive ad.
Use Ad Strength as a build quality check (but don’t chase the label blindly)
Ad Strength is useful as a diagnostic because it pushes you toward more complete, more varied assets. As you add more unique headlines and descriptions (up to the RSA limits), you generally expand the system’s ability to find good combinations and improve relevance.
That said, “Excellent” is not the business goal—profitable conversions are. I’ve seen “Excellent” RSAs that were too broad and lowered lead quality. Use the indicator to catch obvious problems (thin assets, repetition, missing variety), then judge final success with conversion quality and incremental performance.
When you hit the limit: the only fixes that matter
If Google Ads won’t let you enable another RSA in an ad group, treat it like a quick triage:
- Count enabled RSAs in the ad group: if there are already 3 enabled, pause one to free a slot.
- Check total active non-image/text ads: if the ad group is unusually cluttered and near the 50 active ad limit, reduce active ad count.
- Decide what you’re trying to test: don’t pause the “wrong” RSA—replace the one with the least differentiated message or the one constrained by excessive pinning.
The strategic move is usually not “add another RSA,” but “retire the weakest message” and rebuild it into a stronger, more differentiated RSA or into better assets inside the remaining ads.
One final tip: avoid over-pinning unless you truly need it
Pinning can be necessary for compliance, brand requirements, or must-say elements, but heavy pinning can reduce the system’s flexibility and limit performance gains from RSA assembly. Use pinning surgically—only where it protects meaning or legal requirements—then let the rest of the asset set do its job.
