The short answer
Comparing ads launched in the last 90 days with ads that have run for a year, the shift in 2026 is clear: story narration is 35% of new produced videos against 5% of year-old winners, AI-generated video has gone from zero to 21%, and 42% of new winners target unaware audiences. What isn't changing: 7 in 10 ads still bridge a problem to a solution.
How we measured 2026 trends
Most trend lists are opinion. This one comes from the Ad Radar board on October 9, 2026: 2,443 live Meta ads from 749 brands, each with $50k or more in estimated spend, tagged on ten dimensions.
A snapshot has no history, but it has dates. Every ad carries its start date from the Meta Ad Library, so we split the board into three groups:
701
ads launched in the last 90 days
1,404
ads live 90 to 364 days
338
ads live a year or more
10.4%
top-tier share of the newest group (7.4% for year-old ads)
The newest group shows what brands are launching and scaling right now. The year-old group shows what has survived. Comparing them shows the direction of travel.
One caution runs through the whole piece. Year-old ads are survivors: formats that last naturally pile up there (product demos are the clearest case). Where a difference could be survivorship rather than a real shift, we say so.
Trend 1: Story narration takes over
The single biggest change in new creative is narrated stories.
| Video type | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| Story narration | 35% | 16% | 5% |
| Demo of the mechanism | 16% | 22% | 39% |
| Voiceover listicle | 18% | 18% | 13% |
| UGC talking head | 7% | 7% | 23% |
| Expert authority | 8% | 15% | 10% |
| Interview clip | 7% | 3% | 3% |
Share of tagged videos in each group (130, 355 and 79 ads).
A third of newly launched produced videos are now stories told by a narrator: a first-person account, a scene, a turning point, a product. A year ago this type was rare among winners. Part of the decline of demos and talking heads in new launches is survivorship (both last long, so they accumulate in the old group), but the jump in stories is too large to explain that way.
The new stories are working, too. Among the newest ads, story narration reaches the top tier at 15.2%.
I met a combat medic in Afghanistan who had more energy at 40 than most guys had at 25.
- Est. spend
- $5.9M
- Days live
- 48
- Format
- Video, 69s
Already in the top tier seven weeks after launch. A narrator, a character and a contrast in the first line. This is the shape of the 2026 story ad.
Our footage + voiceover ads piece shows why narrated stories outperform other voiceover structures, and story hooks covers how they open.
Trend 2: AI-generated video arrives, but doesn't win yet
| Production format | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| AI-generated | 21% | 13% | 0% |
| Footage + voiceover | 27% | 30% | 29% |
| Founder or expert | 24% | 25% | 20% |
| UGC selfie | 14% | 11% | 21% |
| Product demo | 8% | 16% | 30% |
| Creator skit | 6% | 3% | 0% |
Share of tagged videos in each group (140, 383 and 87 ads).
AI-generated video went from absent among year-old winners to one in five new produced videos. That is the fastest-growing production format on the board. The results don't match the volume yet:
| Format, last 90 days | Ads | Top-tier share | Median est. spend |
|---|---|---|---|
| Footage + voiceover | 38 | 21.1% | $333K |
| UGC selfie | 19 | 15.8% | $411K |
| Founder or expert | 33 | 15.2% | $161K |
| AI-generated | 29 | 3.4% | $139K |
Within the same launch window, so with no age advantage, AI ads reach the top tier at a fraction of the rate of filmed formats and carry less spend. Brands are testing AI heavily. Few are scaling it. We break that down in AI UGC ads.
Trend 3: More ads for cold audiences
| Reader awareness | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| Unaware | 42% | 39% | 25% |
| Problem-aware | 19% | 23% | 32% |
| Solution-aware | 25% | 22% | 26% |
| Most aware | 9% | 10% | 12% |
Ads written for people who don't yet know they have a problem have risen from a quarter of year-old winners to 42% of new launches. This fits how Meta now delivers ads: with broad targeting and a retrieval system that reads the creative itself to decide who sees it, the ad has to find its audience on its own. Ads that start from scratch with the viewer do that. See Meta Andromeda explained and unaware audience ads.
There is a catch. In the newest group, unaware ads reach the top tier at 6.6%, below the 10.4% average for new ads. Writing for cold audiences is the trend. Doing it well is still hard.
Trend 4: Videos get longer and shorter at the same time
| Video length | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| Under 30 seconds | 16% | 14% | 10% |
| Over 2 minutes | 30% | 27% | 24% |
| Median length | 66s | 64s | 61s |
Both ends are growing. New launches include more sub-30-second cuts and more ads past two minutes than the year-old group, while the median barely moves. The pattern matches the formats above: story and expert videos run long, and brands cut short versions for testing and for placements that reward them. The full numbers are in best video ad length for Facebook.
Trend 5: Hooks move away from the offer
| Hook pattern | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| Problem/Benefit Opener | 26% | 31% | 34% |
| First-Person Story | 22% | 17% | 16% |
| Specific Number Opener | 19% | 17% | 14% |
| Question Opener | 18% | 15% | 11% |
| Offer-First Opener | 7% | 11% | 17% |
Offer-first hooks ("50% off today") have fallen from 17% of year-old winners to 7% of new launches. First-person stories and questions are rising. The 2026 opener sounds like a person talking, not a promotion.
I never thought I would like short press-ons at my age, but here we are.
- Est. spend
- $8.4M
- Days live
- 56
- Format
- Video, 59s
- Variants
- 3
A top-tier ad under two months old. A first-person admission that doubles as an objection, which is what many new openers look like.
Trend 6: Fewer images in new launches
Images are 15% of ads launched in the last 90 days, against 23% of year-old winners. Part of that is survivorship: image ads tend to run slightly longer, so they accumulate. Part is the move toward story video. The images that do launch still perform: new image ads reach the top tier at 14.4%. If your competitors are all-in on video, a strong image program is a gap you can use. See video vs image ads.
Trend 7: Product pages gain, dedicated landers slip
| Landing page | Last 90 days | 90 to 364 days | 1 year+ |
|---|---|---|---|
| Product page | 53% | 50% | 45% |
| Dedicated landing page | 25% | 33% | 39% |
| Homepage | 9% | 4% | 9% |
| Listicle | 6% | 5% | 3% |
| Quiz | 4% | 5% | 3% |
More new ads send traffic straight to a product page. Fewer use a dedicated landing page. This fits trend 1: a story ad does the presell inside the video, so the click can go where people buy. Listicle pages are edging up from a small base. For the page side, see product page vs landing page.
Trend 8: Interviews and skits enter the mix
Interview clips are 7% of new produced videos against 3% of year-old ones, and creator skits went from none to 6%. Both are small groups (9 and 8 new ads), so this is an early signal. The interview format has a decent record on the board. Skits haven't produced a top-tier ad yet. See podcast and interview ads and skit ads.
What isn't changing
The deeper structure of a winning ad looks the same in every group:
- Problem-to-solution bridge is the hook mechanism in 70% of new ads and 72% of year-old ones.
- Root cause is the leading angle in all three groups (45% new, 52% year-old).
- Proof levels barely move: strong evidence sits at 5 to 6% throughout.
- Supplements remain the largest product form (41% of new ads with a product tag).
Formats change. The argument inside them doesn't. A new story ad and a three-year-old demo are both showing a problem, explaining why it happens and presenting the product as the fix.
Creative priorities for the rest of 2026
Today we're packing an order for Madison.
- Est. spend
- $7.1M
- Days live
- 84
- Format
- Video, 53s
- Variants
- 4
A top-tier ad under three months old that opens behind the scenes. A small story about one customer's order, which is the 2026 formula applied to a familiar product.
- Make your next concept a narrated story. One person, one problem, one turning point, under two minutes. It is the fastest-growing format and it is winning.
- Use AI in the workflow, not on screen. Scripts, hooks, variants and B-roll, yes. AI people as the main act, test with low expectations. See AI ad creative tools.
- Write at least one concept for unaware buyers, and give it the length it needs to teach.
- Retire the offer-first opener on cold traffic. Keep offers for retargeting and images.
- Cut every winner two ways: a long version and a sub-30-second version.
- Keep a strong image program. Fewer competitors are launching them, and they still reach the top tier.
- Keep the core argument. Problem, cause, product, proof. Every trend above is a new wrapper for it.
To track these shifts as they happen, the Ad Radar board lets you sort winners by days live and filter by format, so you can see what launched this month in your niche. If you work in Claude, the Ad Radar MCP server (tools like winning_patterns and search_ads) lets you ask the same questions from inside a chat.
Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.