Google Ads AI Video Dubbing in Asset Studio Explained (2026)

If you run video ads for more than one country, you already know the real cost of going international is never the media spend. It is the production. A 30-second video that performs well in English usually needs a completely separate version for every other market you want to enter, a new voiceover session, sometimes new actors, definitely new timing since a Spanish or German script rarely lands in the exact same seconds as the English one, and a review pass to make sure nothing reads wrong culturally. Most advertisers solve this the only way that has ever worked: they either skip international video entirely and run static or text-only ads instead, or they budget for a dubbing vendor and wait one to three weeks per language while the invoice climbs.

Google Ads just started chipping away at that bottleneck directly inside the platform. A new feature called Video Ads Dubbing, rolling out inside Asset Studio, uses AI to take a video ad you already have and generate localized audio tracks for it in 33 languages and locales, without a reshoot, a new voice actor booking, or a separate creative brief per market. Search Engine Land reported the rollout on September 19, 2026, and ContentGrip's coverage confirms the same core mechanics: one approved video goes in, translated and dubbed versions come out, and the whole workflow lives inside the same Asset Studio panel advertisers already use for image and video generation.

This is a genuinely different feature from the Gemini Omni video generation Google also added to Asset Studio this year, which is worth being precise about since the two are easy to confuse. Gemini Omni, covered in detail in Miraflow's earlier breakdown, builds an entirely new video ad from a text brief and your brand assets when you have no video yet. Video Ads Dubbing does the opposite job. It takes video you already shot, already approved, already know performs, and gives it new language tracks so the same creative can run in dozens of markets instead of one. One tool creates video from nothing. The other multiplies video you already trust. Advertisers running international campaigns will likely end up using both, but for very different reasons, and this post is about the second one.

Be warned upfront that this is an early, limited rollout, not a fully documented general-availability feature yet. Google has not published a complete public list of the 33 supported languages, has not spelled out full eligibility rules for which advertisers get access first, and has not detailed exactly what editing controls exist once a dubbed track is generated. Both outlets covering the story make the same point: important mechanical details, like whether the system attempts to match mouth movements, how much line-by-line editing you get before a dub goes live, and the full account-level eligibility criteria, are simply not documented publicly yet. Everything in this post that is a confirmed fact is sourced to that reporting or to Google's own existing Asset Studio documentation. Everything that is reasoned production advice, how to prepare a video so an auto-dubbing system has the best possible source material, is exactly that: informed advice based on how AI dubbing and translation pipelines actually work, not a leaked Google specification.

TL;DR

• Google Ads is rolling out Video Ads Dubbing inside Asset Studio, an AI feature that generates localized audio tracks for an existing video ad in 33 languages and locales, without a reshoot or a separate voiceover session per market.
• Access is currently limited to select advertisers in an early rollout. Google has not published the complete language list, full eligibility rules, or the exact editing controls available once a dub is generated.
• The feature appears to draw on the same category of AI dubbing and translation technology behind Google's own open-source Ariel project, which uses Gemini for transcription and translation alongside Cloud Text-to-Speech, though Google has not officially confirmed Asset Studio's dubbing feature and Ariel share the same underlying system.
• The campaigns that stand to benefit most are Video action campaigns, Demand Gen, and Performance Max asset groups that already include video, since all three can plug a dubbed track back into an existing asset group without rebuilding the campaign.
• How you shoot and edit the original video matters enormously for dub quality: clean separation of dialogue from music and sound effects, minimal burned-in on-screen text in the source language, clear and unhurried pacing, and a script written in complete sentences all make an AI dub sound dramatically more natural.
• Compared to hiring a traditional localization or dubbing vendor, Asset Studio's version trades some control and polish for a massive advantage in cost and speed, covered in the comparison table further down. It is a strong first pass for testing new markets, not necessarily a full replacement for a flagship brand campaign's dubbing needs.
• Miraflow's Cinematic AI Video Generator and AI Music Generator are useful earlier in the pipeline, for producing the original video and voiceover that then gets dubbed at scale once it reaches Google Ads.

What Google Ads' Video Ads Dubbing Actually Does

Strip away the AI framing for a second and look at what advertisers have always needed to do to run a video ad in more than one language market. You need a translated script that still fits the pacing of the original video, a voice actor or text-to-speech voice that sounds natural in the target language, a mix that blends the new dialogue back in with the existing music and sound effects at the right levels, and a final render that syncs the new audio to the picture. Every one of those steps used to require either a specialized vendor or a genuinely skilled in-house production team, and every additional language multiplied the work almost linearly. Ten languages meant something close to ten separate production cycles.

Video Ads Dubbing folds that entire chain into a single request inside Asset Studio. According to the reporting from Search Engine Land and ContentGrip, an advertiser selects an existing, already-approved video asset, and the tool analyzes the audio, separates the speech from other audio elements as best it can, translates the dialogue into a chosen target language, generates a new spoken audio track in that language, and produces a localized version of the ad that can be used in place of, or alongside, the original. The advertiser is not asked to write a translated script, cast a voice, or book studio time. The system produces a full candidate dub from one input video and one language selection.

The headline number in both outlets' coverage is 33 languages and locales. That phrase, languages and locales, matters more than it might first appear. A language like Spanish is not one dubbing target, it is several, since Spanish spoken in Mexico, Spain, and Argentina carries different vocabulary, cadence, and cultural reference points even though the grammar is largely shared. Describing the feature's coverage as languages and locales rather than just languages suggests Google is thinking about regional variants specifically, not simply checking off major world languages once each. That is a meaningfully more useful scope for an advertiser actually trying to reach a specific national market rather than a broad language family, but since Google has not published the full 33-item list yet, do not assume your specific target locale is definitely included until you see it appear as an option in your own account.

What is genuinely unclear, and worth being honest about rather than guessing, is the fidelity of the dub itself. Neither Search Engine Land nor ContentGrip's reporting confirms whether the system attempts any form of lip-sync adjustment, whether it tries to match the video's mouth movements to the new language's phonemes the way high-end dubbing studios and some dedicated AI dubbing tools do, or whether it simply replaces the audio track cleanly underneath the existing picture and accepts that lip movement will not match perfectly, the same trade-off most efficient dubbing has always made outside of premium film localization. ContentGrip's coverage explicitly flags this as an open question the current documentation does not answer. For an advertiser deciding how to use this feature, the safe assumption right now is that it behaves like practical, high-volume dubbing rather than frame-perfect lip-sync, which actually shapes some of the production advice later in this post, since a video where the speaker's face fills the frame in tight close-up will show audio-visual mismatch more obviously than a video that cuts between b-roll, product shots, and a speaker who is not always in frame.

The Technology Behind It: The Ariel Connection

Google has not put out a technical paper explaining exactly what model or pipeline powers Asset Studio's Video Ads Dubbing, and neither has any outlet covering the story confirmed a direct, official link between this specific Google Ads feature and Google's existing open-source dubbing project. That gap is worth naming honestly rather than papering over with a confident-sounding claim this post cannot actually back up.

What is verifiable is that Google already maintains a public, open-source project called Ariel, built by the Google Marketing Solutions team, described in its own repository as a cutting-edge AI dubbing solution designed to enhance the global reach of digital advertising. Ariel's documented pipeline transcribes a video's speech, translates it, and generates a new dubbed audio track using a stack built on Google Cloud services, Vertex AI running Gemini models for transcription and translation, Cloud Text-to-Speech for generating the new spoken track, and Cloud Translation API in the mix as well. That is, functionally, the exact same job Video Ads Dubbing performs inside Asset Studio: analyze a video's speech, translate it, generate a new audio track in the target language, output a localized version of the ad.

The honest way to describe the relationship, given the current public information, is that Video Ads Dubbing appears to sit in the same technical family as Ariel, both are Google-built systems solving the identical advertising localization problem using generative AI translation and text-to-speech, but Google has not published a statement saying Asset Studio's consumer-facing dubbing feature literally runs Ariel's codebase in production. It would not be surprising if Asset Studio's dubbing feature is a productized, tightly integrated evolution of the same underlying approach Ariel demonstrated publicly, built by the same internal team for the same underlying goal. But calling that a confirmed fact right now would be getting ahead of what Google has actually said. Treat the Ariel connection as strong circumstantial context for how this kind of AI dubbing likely works technically, transcription, translation, and text-to-speech generation stacked together, rather than a documented specification of Video Ads Dubbing's exact internals.

This matters practically, not just as a footnote for accuracy's sake. If the underlying approach really is transcribe, translate, then re-voice, the same three-stage pipeline Ariel documents openly, that tells you exactly where dub quality is most likely to break down, and exactly where you have leverage as the advertiser preparing the source video. A transcription stage struggles most with overlapping dialogue, background music that masks speech, and multiple speakers talking over each other. A translation stage struggles most with idioms, brand-specific terminology, and lines that depend on wordplay in the source language. A text-to-speech stage struggles most with matching the natural pacing and emotional tone of the original delivery. Every one of those failure points is something you can reduce at the source, before the video ever reaches Asset Studio, which is exactly what the preparation checklist further down in this post walks through.

Which Campaign Types Benefit Most

Video Ads Dubbing is not a standalone product. It generates a dubbed asset that then needs to live somewhere inside an actual Google Ads campaign, which means the real question for most advertisers is less "should I use this" and more "where in my existing account does a dubbed asset actually plug in cleanly." Three campaign types stand out clearly.

Video action campaigns, Google's format built specifically around driving a conversion action from a video ad on YouTube, are the most natural fit. These campaigns already center on a single hero video asset per ad group, so replacing or adding a dubbed variant is a direct, contained change rather than something that ripples across a dozen linked assets. An advertiser running the same video action campaign concept across, say, the UK, Germany, and Brazil can plausibly run one underlying video with three dubbed audio tracks feeding three region-targeted ad groups, instead of maintaining three separately produced videos.

Demand Gen campaigns, which run across YouTube, Discover, and Gmail with a broader creative mix than Video action campaigns, benefit for the same core reason, plus one more. Demand Gen already leans on multiple asset variations being tested against each other within one campaign, so a dubbed track is just one more variant in a system that is already built to rotate and test creative combinations, rather than a special case that needs its own separate campaign structure.

Performance Max asset groups that already include video are the third clear fit, and arguably where the compounding value shows up most, because Performance Max already spans Search, Display, YouTube, Discover, Gmail, and Maps from one asset group. A Performance Max campaign expanding into a new country typically needs the same asset group logic rebuilt with localized text assets, localized images, and ideally a localized video, all at once. Being able to generate the video half of that localization inside the same Asset Studio panel where the text and image assets are already being adapted removes one entire production dependency from a process that otherwise has several. This connects directly to two other recent Performance Max changes worth knowing alongside this one: the Asset Studio prefill update, which already auto-populates asset groups straight from Asset Studio, and the AI video resizing feature, which auto-generates missing aspect ratios for video assets. Google has clearly been building Asset Studio into a single hub that handles more of a video asset's lifecycle end to end, generation, resizing, and now localization, rather than three separate disconnected tools.

One campaign type worth naming as a poor fit, at least based on what is documented so far, is a pure Search campaign with no video component, obviously, since there is nothing to dub. Less obviously, a Shopping campaign built entirely around a Merchant Center product feed also has no natural place for a dubbed video unless it is running as a linked video within a broader Performance Max asset group, since standard Shopping listings do not carry a video creative slot at all.

Step by Step: Preparing Your Video Assets for Auto Dubbing

This is the part most coverage of this feature skips entirely, and it is the part that actually determines whether a dubbed video sounds like a real localized ad or an obviously synthetic one. An AI dubbing pipeline, whatever its exact internals, can only work with the audio and structure you actually feed it. A messy source video produces a messy dub no matter how good the underlying model is. Here is a real, practical checklist for preparing a video ad specifically for auto-dubbing, reasoned from how transcription, translation, and text-to-speech pipelines actually behave, not guessed at random.

Step 1: Record or mix dialogue on its own separate track from music and sound effects, and keep it that way through final export if you can. The single biggest lever you have over dub quality is how cleanly the speech can be isolated from everything else in the mix. A transcription model has to first identify what is speech at all before it can translate anything, and background music with a similar frequency range to a human voice, or sound effects layered directly under a line of dialogue, make that separation meaningfully harder. If your video was produced with separate stems, keep the final mixed export available as a backup, but consider whether a version with the dialogue track slightly more isolated, or briefly ducked music under key lines, gives the dubbing system a cleaner signal to work from.

Step 2: Write the original script in clear, complete sentences, and avoid dense wordplay or idioms the ad's core message depends on. A translation stage handles a straightforward, literal sentence far better than a pun, a culturally specific idiom, or a line where the humor depends entirely on a double meaning in the source language. If your hook line only works because of an English wordplay, that line is very likely to lose its punch, or land oddly, once translated, no matter how good the underlying model is. Consider whether your core message can be restated in a way that survives translation cleanly, while keeping a separate, more idiomatic English-only version for markets where you are not localizing.

Step 3: Minimize burned-in, on-screen text in the source language, especially text that repeats or reinforces the spoken line. This is easy to overlook because burned-in captions and on-screen callouts are common in modern video ad editing, used to reinforce a key phrase, a price, or a call to action visually. The problem is that a dubbed audio track changes the spoken language, but it does nothing to the picture itself. A viewer watching a German-dubbed version of your ad with English text baked into the video frame gets a jarring mismatch, hearing one language while reading another laid directly over the footage. Where possible, keep essential text as a separate overlay layer added after dubbing rather than burned permanently into the base video file, or minimize how much of your message depends on on-screen text at all, letting the voiceover carry more of the actual communication.

Step 4: Keep speaking pace natural and slightly unhurried rather than packed tight against the video's runtime. Different languages carry the same meaning in different numbers of syllables and different natural speaking speeds. German and Finnish sentences commonly run longer than the equivalent English sentence for the same idea, while some Romance languages can run either direction depending on the specific phrasing. A script that is already timed to the exact second in English, with no breathing room, gives a text-to-speech generation stage nowhere to fit a translated line that naturally needs slightly more or less time, which is one of the more common ways an auto-dubbed line ends up sounding rushed or unnaturally compressed.

Step 5: Reduce reliance on tight, sustained close-ups of a speaker's face, especially for lines with real emotional weight. Given that neither Search Engine Land nor ContentGrip's reporting confirms lip-sync adjustment is part of this feature, assume for now that mouth movement will not perfectly match the new dubbed audio. A wide shot, a cutaway to product footage, or a voiceover played over b-roll hides that mismatch naturally, since the viewer is not staring directly at lip movement while the new audio plays. A tight, sustained close-up on a talking face makes any audio-visual mismatch immediately obvious. This is not a reason to avoid close-ups in your original video entirely, but it is worth knowing that heavily close-up-driven video is the format most likely to show visible seams once auto-dubbed, and it is worth reviewing those specific moments most carefully before publishing a dub.

Step 6: Review every generated dub before it goes live, and check pronunciation of brand names, product names, and any numbers or measurements specifically. A text-to-speech system generating a new voice track in an unfamiliar language has no inherent way to know how your brand name is supposed to be pronounced in that market, and mispronounced brand names are one of the fastest ways to undermine trust in an ad rather than build it. The same goes for prices, units of measurement, and any region-specific figure, since a currency symbol or measurement unit that is not localized alongside the language can read as sloppy even when the dub itself sounds fluent.

Step 7: Treat the auto-dubbed version as a strong first draft for testing new markets, not an automatic final asset for your flagship campaign. Given that this is an early rollout with unconfirmed editing controls, the safest workflow right now is generating a dub, reviewing it carefully against the checklist above, and using it to validate whether a market is worth deeper investment before committing budget at full scale. If a dubbed version performs well in a smaller test spend, that is a strong signal the market and the message both work, at which point investing in a more polished, professionally produced version for that market's larger campaigns becomes a much easier budget decision to justify.

Common Mistakes

Assuming every language on your wishlist is already available. Google has not published the complete 33-language list publicly, and account-level access is still an early rollout. Check what actually appears as an option inside your own Asset Studio panel rather than planning a full international campaign around a specific locale you have not confirmed is live yet.

Feeding the system a video with burned-in captions in the original language. As covered above, this is one of the most common and most avoidable mistakes, since it produces an obvious language mismatch between what the viewer hears and what they read on screen. Check your existing video library for this before assuming a video is dub-ready.

Skipping the review pass because the output sounds fluent at a glance. A dub can sound perfectly natural sentence by sentence and still mispronounce your own brand name, get a price wrong, or land a cultural reference oddly. Fluency and accuracy are not the same thing, and only a native or fluent reviewer in that target market can actually confirm the second one.

Treating this as a replacement for cultural review, not just language translation. A literal translation can be linguistically correct and still miss the mark culturally, a color, a gesture, or a reference that reads fine in the source market can carry a completely different connotation elsewhere. Auto-dubbing solves the language problem. It does not solve the cultural fit problem, which still needs a human familiar with that specific market.

Applying the same close-up-heavy video to every dubbed market without reviewing lip-sync mismatch first. A talking-head video that looks perfectly fine auto-dubbed into one language can look noticeably off in another, depending on how differently that language's phonemes move the mouth. Review each dub individually rather than assuming one pass is representative of all of them.

Forgetting to localize the parts of the ad this feature does not touch. Video Ads Dubbing changes the audio track. It does nothing to a landing page, a call-to-action button rendered as an image, or a companion text asset running alongside the video in the same asset group. A beautifully dubbed video sending traffic to an English-only landing page still loses most of the value the localization was supposed to create.

Video Ads Dubbing vs Hiring a Localization Vendor

The realistic comparison most advertisers actually need to make is not "AI dubbing versus doing nothing." It is "AI dubbing versus the traditional path of hiring a dedicated localization or dubbing vendor," since that vendor path is what most serious international video campaigns have relied on until now. Neither approach is strictly better across every dimension, which is exactly why the comparison is worth laying out plainly rather than picking a winner.

The practical read on that table is straightforward. A dedicated vendor still wins decisively on control, polish, and accountability, a real studio session gives you a human director's ear on delivery, a dedicated linguist checking cultural fit line by line, and someone to hold responsible if something goes wrong before launch. That is exactly why flagship brand campaigns, anything running significant paid media weight in a market where getting it wrong is expensive, will likely keep working with vendors for the foreseeable future, dub quality issues aside.

What Asset Studio's version changes is the economics of testing a market at all before deciding whether that vendor investment is worth making. A vendor engagement for even a handful of languages easily runs into thousands of dollars and one to three weeks of turnaround per language, a real cost and time commitment that makes testing five or six new markets speculatively a genuinely risky bet for a mid-sized advertiser. Generating a dub inside Asset Studio, reviewing it against the checklist earlier in this post, and running a modest test budget in a new market changes that calculation entirely. A market that clearly does not respond, even with a solid dub, can be deprioritized without having spent real vendor budget finding that out. A market that responds well becomes a much easier case to bring a full vendor-produced campaign to later, backed by real performance data rather than a hunch.

Where Miraflow Fits: Producing the Original Creative That Gets Dubbed

Video Ads Dubbing solves the localization half of the problem. It does nothing to help you make the original video and voiceover in the first place, which is exactly where a tool like Miraflow fits naturally into the workflow rather than competing with what Google Ads is doing.

Miraflow's Cinematic AI Video Generator, built on models including Veo3 and Veo3.1, lets a small team or a solo marketer produce a polished, cinematic-quality source video from a written prompt, without booking a production crew or a studio day. That matters directly for the preparation advice covered earlier in this post, since a video generated with clean, deliberate pacing and clear separation between dialogue and background elements from the start is exactly the kind of source material that dubs well later. Rather than treating video production as a bottleneck that has to be solved before localization can even begin, generating the original creative inside Miraflow and then feeding the finished export into Asset Studio's dubbing tool turns the entire pipeline, idea to finished localized ad, into something a lean team can realistically execute without outside vendors at either end.

The audio side matters just as much as the visual side here, and this is where Miraflow's AI Music Generator fits in specifically. Since Step 1 of the preparation checklist above is about keeping dialogue cleanly separated from music and sound effects for a better dub, generating your background score directly as its own clean track, rather than pulling from a stock library you have to fight to isolate later, gives you real control over exactly how that separation is structured from the very first edit. Miraflow's music tool supports both a quick simple mode for a fast instrumental backing track and a more detailed custom mode with control over duration, tempo, and structure, useful for matching a score's length and energy precisely to a video's actual runtime and pacing, the same runtime and pacing consideration that Step 4 of the checklist flags as important for how well a dub fits.

For advertisers whose video ads are cut down from longer source footage rather than built from scratch, Miraflow's AI Clipping tool is worth a mention too. It analyzes a full-length video, identifies and ranks the strongest moments with a viral score, and automatically produces vertical, captioned clips ready for short-form placement, a useful step before those clips ever reach Asset Studio's dubbing tool for language expansion. And for advertisers who want to keep their AI voiceover workflow consistent across markets even outside of Google's own dubbing pipeline, it is worth reading Miraflow's complete guide to AI voice cloning alongside this post, since a consistent, clonable brand voice matters whether the actual dubbing happens inside Asset Studio or through a separate tool entirely.

None of this replaces the actual dubbing step Google Ads now offers directly. It simply means the video and audio you feed into that dubbing step can be meaningfully better prepared, which, per the checklist earlier in this post, is the single biggest factor in how good the resulting dub actually sounds.

Frequently Asked Questions

Is Video Ads Dubbing available in every Google Ads account right now?

No. Based on current reporting, it is an early, limited rollout available to select advertisers inside Asset Studio, not a fully general-availability feature yet. If you do not see it in your account, that reflects the staged rollout rather than a setup mistake on your end.

What languages does Video Ads Dubbing actually support?

Google has described the coverage as 33 languages and locales, but has not published the complete public list yet. The safest approach is checking which specific languages actually appear as options inside your own Asset Studio panel rather than assuming a specific target market is included.

Does Video Ads Dubbing cost extra on top of my normal Google Ads spend?

Both outlets covering the rollout describe it as free for advertisers who currently have early access. Google has not indicated a separate pricing tier for the feature as of this writing, though that could change as it moves toward broader availability.

Is Video Ads Dubbing the same thing as Gemini Omni in Asset Studio?

No, and this is a common point of confusion. Gemini Omni generates an entirely new video ad from a text brief and your brand assets when you have no existing video. Video Ads Dubbing takes a video you already have and generates new localized audio tracks for it. One creates video from nothing, the other multiplies video you already trust across languages.

Does the dubbed audio match the speaker's lip movements?

This has not been publicly confirmed either way. Current reporting does not establish whether the system attempts any lip-sync adjustment, so the safer assumption is that it behaves like efficient, high-volume dubbing rather than frame-perfect lip-sync, similar to most dubbing outside of premium film localization.

Can I edit a generated dub before it goes live?

The exact editing controls available have not been fully documented publicly yet. Treat every generated dub as something to review carefully before it runs, checking brand name pronunciation, numbers, and cultural fit specifically, rather than assuming it is either fully editable or fully locked.

Should this replace my dubbing vendor entirely?

For most serious, high-spend flagship campaigns, probably not yet, since a dedicated vendor still offers more control, polish, and direct human accountability. It is a strong, fast, low-cost way to test whether a new market is worth that deeper vendor investment in the first place, which is a genuinely valuable use even if it never fully replaces a vendor relationship for your top campaigns.

Which Google Ads campaign types can actually use a dubbed video today?

Video action campaigns, Demand Gen campaigns, and Performance Max asset groups that already include a video asset are the clearest fits, since all three can plug a dubbed track into an existing asset group structure without a rebuild. A pure Search or feed-only Shopping campaign has no natural place for a video asset at all.

Conclusion

Video Ads Dubbing is a genuinely practical answer to a problem that has quietly limited international video advertising for years: producing a separate, fully localized video for every market you want to enter is expensive and slow enough that a lot of advertisers simply never bother, running text or static ads internationally instead of the video creative that actually performs best. Folding translation, voice generation, and audio localization into a single request inside Asset Studio, whether or not it turns out to be built directly on the same technology as Google's open-source Ariel project, meaningfully lowers the cost of finding out whether a new market is worth a real production investment.

The honest caveats matter here just as much as the feature itself. This is an early, limited rollout without a fully published language list, documented eligibility rules, or confirmed editing controls, and the smartest way to use it right now is as a fast, low-cost way to test new markets and validate a message, backed by careful preparation on the source video and a real human review of every dub before it runs, not as a fully hands-off replacement for a dedicated localization vendor on your highest-stakes campaigns.

Getting the most out of a feature like this starts before it ever reaches Google Ads, with a clean, well-produced original video and a properly separated audio mix. Miraflow's Cinematic AI Video Generator and AI Music Generator are a genuinely useful starting point for producing that original creative, and tools like Text2Shorts and the AI Image Generator round out a full content pipeline for advertisers building video assets from scratch rather than adapting existing footage. Anyone building out a broader video ad workflow can explore the full toolset from the Miraflow homepage, and the Miraflow blog has more breakdowns of recent Google Ads changes worth reading alongside this one, including the removal of manual language targeting, the Performance Max asset A/B testing rollout, and the real-time policy reviews update, since Google has clearly been rebuilding several pieces of the video and localization workflow inside Google Ads at once, not as one isolated announcement.