Performance Max seasonal themes are Google Ads’ newest creative lever: when you create an asset group, you can now select a high-performing existing group, apply a theme like Back to school or Cyber Monday, and let Google’s AI generate themed image variations plus suggested headlines and descriptions — saved as a copy, with your original asset group left untouched.
The option was spotted broadly in accounts in early July 2026 and picked up by Search Engine Roundtable and the PPC trade press within a day of each other. It arrives at a delicate moment: Google’s AI now generates ad creative on request, and two days before this post, Google also began labeling AI-made ads inside My Ad Center. Themed asset groups sit at the intersection of both stories.
This guide covers what the new option actually does, the exact creation flow from Google’s own help documentation, the full official theme catalog, why theming is a fundamentally different tool from Performance Max asset experiments, the rollout timeline, the disclosure-label question, and a practical playbook for using it without letting AI creative drift off-brand.
- 01A third asset-group creation path is rolling out.Alongside New and Copy, Google Ads now offers Apply theme to asset group: pick a high-performing source group, choose a theme, and Google's AI generates themed image variations plus suggested headlines and descriptions. The output saves as a copy; the original is untouched.
- 02This is not the June 8 asset-experiments feature.Asset experiments run a controlled A/B test — a 40-bucket traffic split, a winner declared at 95% confidence, a 4–6 week minimum. Theming has no built-in test framework: it generates a new copy you publish and monitor manually. That contrast is our read of the two features' documented mechanics.
- 03The official catalog is bigger than the press lists.Google's help doc groups themes into Generic, Seasonal, and Holidays/cultural-moments buckets — including entries like Cyber Monday, Hanukkah, and Lunar New Year that most press summaries skip. Treat trade-press theme lists as illustrative, not exhaustive.
- 04It has been rolling out gradually since Q1 — no launch date.The first documented sighting dates to March 31, 2026, with another wave visible by roughly April and broad coverage on July 7. Availability still varies by account, and the interface language must be one of eight supported languages.
- 05The AI-disclosure connection is the open question.Google's July 9 'How this ad was made' labels auto-disclose ads created with Google's own generative AI tools. Themed assets are generated by exactly those tools, so a published themed group would logically qualify — but that link is our inference; Google states it nowhere.
01 — What ShippedThe third path in asset-group creation.
When advertisers create an asset group in a Performance Max campaign, many accounts now see three paths instead of two: New asset group, Copy existing asset group, and the new Apply theme to asset group. The third option is the news: it clones a source group you choose and uses Google’s AI to re-dress it for a season, holiday, or promotion — generated image variations plus suggested headlines and descriptions, reviewable before anything saves.
One distinction has to land before anything else in this post. Google shipped native Performance Max asset experiments on June 8, 2026 — a structured A/B testing framework with a 40-bucket traffic split, a 95% confidence threshold, and a 4–6 week minimum run time. Theming is not that. It shares the same UI neighborhood — both live around asset-group management — and the two features surfaced within a month of each other, which is exactly why advertisers conflate them. But experiments answer did this change cause a lift, while theming answers can I get seasonal creative made faster. Section 04 maps the differences side by side.
If you want the foundation first — what asset groups are and how Google’s AI assembles their headlines, descriptions, images, and video across Search, Display, YouTube, Gmail, Maps, and Discover — our Performance Max campaign guide covers the mechanics in depth.
New asset group
The classic route — assemble headlines, descriptions, images, and video manually for a new audience signal, product line, or offer. Full control, full production cost.
Copy existing asset group
Clone a group as-is and adjust it by hand. Faster than starting from scratch, but every seasonal variation is still your team's design and copywriting work.
Apply theme to asset group
Select a high-performing source group, pick a theme from Google's catalog, and the AI generates themed image variations plus suggested headlines and descriptions. Saves as a copy — the original stays untouched.
02 — MechanicsHow theming actually works.
Google’s own help center documents the flow end to end, which is worth walking through because the details constrain how you can use it. Per Google Ads Help, the sequence is:
- Navigate to Campaigns → Asset Groups and select + Create asset group.
- Choose Apply theme to existing asset group.
- Select the source asset group you want to copy — the natural move is your best performer, since the point is to carry proven structure into a new seasonal moment.
- Pick a theme from Google’s catalog (Section 03 below).
- Review and edit the AI-generated assets — themed image variations plus suggested headlines and descriptions.
- Save. The new group lands as “Copy of [Original Asset Group Name]”, and the source group is not modified.
Two implementation details from the help doc shape what to expect from the output. Image generation runs on Google’s AI Image Editor technology — the same generative image tooling Google has been threading through its ads products. And text generation uses the detected language of the campaign’s final URL, so the themed headlines and descriptions follow your landing page’s language rather than your account settings.
What the help doc does not say is just as important. There is no themed-asset-specific brand-safety, content-moderation, or human-review guidance anywhere in the documentation — generated assets are subject to standard Google Ads policies, and that is the entire stated guardrail. The review step in the flow is your only structural checkpoint, which is why the playbook in Section 07 treats it as non-negotiable.
03 — Theme CatalogThree buckets, from studio settings to Lunar New Year.
Google’s help documentation is the authoritative source for what you can actually pick, and its catalog is meaningfully wider than the lists circulating in trade coverage. The official catalog groups themes into three buckets:
Evergreen settings
Celebration, Sale/Promotion, and Studio setting — non-calendar themes for promo pushes and product-shot refreshes that work any week of the year.
Season resets
Winter, Spring, Summer, and Fall — quarterly creative refreshes that keep an evergreen asset group from visually aging out between holiday pushes.
Cultural moments
Back to school, Halloween, Cyber Monday, Christmas, Hanukkah, New Year, Lunar New Year, Valentine's Day, Easter, Mother's Day, and Father's Day.
A sourcing caution worth keeping straight: press coverage of the feature lists a shorter, overlapping set — Black Friday, Christmas, summer sale, back to school, Mother’s Day, Valentine’s Day — that roughly maps onto the official catalog’s Sale/Promotion and Holidays buckets. Those press lists are paraphrases, not the catalog. Notably, Google’s own documented list includes Cyber Monday rather than the Black Friday label most summaries use, and it carries cultural moments — Hanukkah, Lunar New Year — that no press write-up we reviewed mentions. When you plan a seasonal calendar around this feature, plan from the help doc, not from the coverage.
04 — The DistinctionTheming vs asset experiments — different levers entirely.
None of the outlets covering the theming news drew this contrast, so we built the map ourselves. The two features share an entry point — asset-group creation and management — and arrived 29 days apart, June 8 and July 7 by first broad sighting. But they solve opposite problems: theming is a creative-generation lever — it manufactures seasonal variations fast — while experiments are a statistical-proof lever — they tell you whether a change actually caused a lift. The rows below draw on Google’s theming help doc, our asset-experiments test playbook, and the practitioner coverage; the framework rows reflect our own analysis of the two features’ documented mechanics.
| Dimension | Apply theme to asset group | Native asset experiments (Jun 8, 2026) |
|---|---|---|
| Mechanics | ||
| What it does | Clones a chosen asset group; Google’s AI generates themed image variations plus suggested headlines and descriptions | Runs a structured A/B test of an asset change inside one campaign |
| Where it lives | Asset-group creation flow — the third option under + Create asset group | The campaign experiments framework, applied to Performance Max assets |
| Output | A new group saved as “Copy of [Original Asset Group Name]”; the original is untouched | A winner declared at 95% confidence on a 40-bucket traffic split |
| Operations | ||
| Testing framework | None built in — you publish the themed copy and monitor performance manually (our read of the documented mechanics) | Built in — 40-bucket split, 95% confidence threshold, guided end dates |
| Time to first output | Near-immediate — assets generate within the creation flow; performance reads are on you | 4–6 week minimum per test |
| Concurrency limits | No stated limit on themed copies in the documentation | One experiment per campaign at a time; asset groups locked while a test runs |
| Strategy | ||
| Best use case | Seasonal creative refresh and variation velocity ahead of a holiday or promo moment | Proving whether an asset change actually causes a performance lift |
| Risk profile | AI-generated creative can drift off-brand or off-offer; human review is the only checkpoint | Statistically safe by design, but slow and strictly sequential |
| Review guardrails | None specific in Google’s help doc — standard ads policies only; review every generated asset yourself | Experiment design enforces a control arm against the treatment |
The strategic reading: Google is building out both halves of an automated creative loop. AI-driven bidding already runs the money side of Performance Max largely without human hands; experiments gave the platform a native way to validate asset changes; theming now gives it a native way to manufacture them. The missing piece — and it is conspicuously missing — is a bridge between the two: nothing in the theming flow offers to route a themed copy into an experiment. Until that bridge exists, connecting generation to validation is manual work that falls on the advertiser, which is precisely the workflow gap Section 07 addresses.
05 — TimelineRolling out since March, not launched in July.
Despite the July news cycle, this feature has no launch date — and framing it as “new in July” gets the story wrong. The documented trail runs 98 days:
- March 31, 2026 — Search Engine Roundtable first documents asset-group theming, credited to PPC practitioner Bia Camargo’s LinkedIn screenshots.
- ~April 2026 — a Swedish PPC agency publishes its own sighting, which the July coverage later cites as evidence of the earlier wave.
- July 6–8, 2026 — the broad wave: Search Engine Roundtable’s July 7 piece (crediting Thomas Eccel’s LinkedIn post) and PPC News Feed’s July 8 write-up land within a day of each other, both describing the option as newly visible across many more accounts.
More than three months separate the first sighting from broad visibility — a classic Google Ads gradual rollout, where a feature quietly reaches a slice of accounts, expands in waves, and only makes news when enough practitioners hit it at once. The practical consequences: if you do not see the option today, that is expected, not a signal it was pulled; and if you plan Q4 seasonal creative around it, build a manual fallback in case your accounts join the rollout late.
One projection worth making explicit: features that reach the broad-sighting stage in Google Ads typically continue expanding rather than retrenching, and the timing here — wide visibility arriving just before back-to-school and Q4 planning cycles — suggests Google wants this in advertisers’ hands for the seasonal windows the theme catalog is clearly built around. Treat the second half of 2026 as this feature’s proving season.
06 — AI DisclosureThe disclosure-label question nobody has answered.
Two days before this post — July 9, 2026 — Google began rolling out AI-creation disclosures for ads. A new “How this ad was made” section inside the My Ad Center panel indicates whether an ad was created or edited with AI, accessible globally via the three-dot menu or info icon on ads across Search, YouTube, and Discover. The rollout was announced on Google’s official blog by Keerat Sharma, VP and GM of Ads Privacy and Safety, alongside a restatement of policy: Google continues to prohibit misleading and deceptive ads “whether created with AI or not.”
The mechanics split into two tiers. Ads built with Google’s own generative AI advertising tools get the disclosure added automatically. Ads built with AI from other providers rely on the advertiser manually flagging AI use — an honor system. And in the EU, India, and New York, ads with assets designated AI-created or AI-edited also carry a visible on-ad overlay driven by local legal requirements, beyond the My Ad Center panel. The timing is not accidental: the EU AI Act’s Article 50 transparency obligations for AI-generated content take effect August 2, 2026 — 24 days after the label rollout — and multiple compliance observers frame Google’s move as pre-positioning for that deadline.
Why this matters practically: if the inference holds, every themed asset group you publish becomes a labeled artifact — a user curious enough to open My Ad Center could see that the ad was AI-made, and in the EU, India, and New York the overlay would be on the ad itself. For most performance advertisers that is a non-event. For brands whose positioning leans on craft or authenticity, it is a real creative-strategy input. Our advertiser playbook on Google’s AI-disclosure labels covers what the labels look like, where they surface, and how to audit your exposure.
07 — PlaybookUse it as inspiration, never autopilot.
The most useful guidance in the entire news cycle came from the practitioner who surfaced the feature. Thomas Eccel’s caution is worth quoting in full, because it is the only cautionary voice on record — Google’s own documentation, as noted above, offers no themed-asset-specific review guidance at all:
"Use them to play around with, but do not let it run on autopilot. A holiday theme still has to match your brand and your offer. I would use it as inspiration but not apply it."— Thomas Eccel, PPC practitioner, via Search Engine Roundtable, Jul 7, 2026
That framing — inspiration, not application — is the right default while the feature is this young. Here is how we would slot theming into a Performance Max creative workflow, by scenario:
Fast creative variation for a calendar moment
Theming's sweet spot. Clone your best performer, apply the seasonal theme, then human-review every generated image and line against brand and offer before saving. The AI does the production; you keep editorial control.
Does the new creative actually perform better?
Theming gives you no control group — any performance read against the original is confounded. If you need causal evidence, run the native asset experiment: 40-bucket split, 95% confidence, 4–6 week minimum.
Campaigns where visual identity is the asset
Google's help doc names no themed-asset guardrails — standard ads policies are the only stated backstop. Where the brand look is the moat, treat AI output as a first draft for your designers, or keep production fully in-house.
Theme first, then validate the winner
The bridge Google has not built: generate a themed copy, edit it to standard, and — where the 4–6 week experiment window fits your seasonal timing — validate the edited variant through an experiment before scaling spend behind it.
Three operational habits round out the playbook. First, never skip the review step — it is the only structural checkpoint between generation and a live ad, and the detected-language behavior means multilingual accounts should check copy especially carefully. Second, name and annotate themed groups the moment they save — the default “Copy of” naming will bury itself in any account with real asset-group volume. Third, log themed groups as AI-generated in your own records — if the disclosure inference from Section 06 holds, you will want an inventory of which live assets carry Google-generated creative. If you would rather have a senior team run this whole loop — generation, review, validation, and the seasonal calendar around it — that is exactly what our paid media service does for Performance Max accounts.
08 — ConclusionA creative lever, not a testing lever.
Theming manufactures seasonal creative — proving it works is still your job.
Apply theme to asset group is a genuinely useful addition to Performance Max: cloning a proven asset group and letting Google’s AI re-dress it for Cyber Monday or Back to school collapses what used to be a design-team production cycle into a review-and-save flow. The original stays untouched, the catalog covers the commercial calendar, and the rollout has been widening since March.
But keep the feature map straight. Theming generates; experiments prove. Nothing in the theming flow tells you whether the seasonal variant outperforms the original — that still requires the June asset-experiments framework or your own disciplined measurement. And the disclosure-label question hangs over all of it: themed assets are Google-generated AI creative arriving in the same month Google started labeling AI-made ads, days before the EU’s transparency obligations bite. We think the label connection is logical; Google has not confirmed it.
The practical stance is Eccel’s: use it as inspiration, never autopilot. Review every generated asset, keep your own inventory of AI-made creative, and where the timing allows, close the loop Google left open — theme, edit, then test.