The short answer

Lo-fi wins on Meta, but not because it's cheap. Phone-native ads on our board (UGC selfie and creator skits) carry a $482K median estimated spend, and videos that open with a casual word like "Okay", "So" or "Hi" reach the top tier at 17.4%, about twice the rate of other openers. Polished brand spots and synthetic visuals do worst. The winners are produced where it helps and casual where it counts.

What the data can and can't tell us

Ad Radar doesn't tag ads as "lo-fi" or "polished." That's a judgment call, and two people would disagree on half the board. What we do tag is the production format of 610 winning videos, and every video has a transcript. Those give us three proxies that hold up:

  1. Production format. We group UGC selfie and creator skit ads as phone-native. Footage with a voiceover, product demos and founder or expert videos are produced (some are shot on phones, but they are edited and scripted). AI-generated and animated ads are synthetic.
  2. The first word. We check whether the transcript opens with a casual word like "Okay," "So," "Hi," "Hey," "Oh" or "Yeah," the way people start a voice note.
  3. Short brand spots. Videos of 15 seconds or less, which on our board are mostly polished ads from large consumer brands.

All numbers come from the board on October 9, 2026: 2,443 live Meta ads with $50k+ in estimated spend.

Phone-native vs produced vs synthetic

Production groupAdsTop-tier shareMedian est. spendShare above $1MMedian days live
Phone-native (UGC selfie, creator skit)9810.2%$482K33%135
Produced (voiceover, demo, founder)41912.2%$310K18%144
Synthetic (AI-generated, animation)904.4%$232K13%101

Three things stand out.

Phone-native ads spend the most. One in three is above $1M in estimated spend, almost twice the rate of produced ads. When a lo-fi ad works, brands push serious budget through it.

Produced ads win slightly more often. Their top-tier share is the highest of the three groups, and they run the longest. "Produced" here rarely means glossy: the typical produced winner is a narrator over real footage of a dog, a kitchen, a product in use.

Synthetic is the weak group. AI-generated and animated ads have the lowest top-tier share, the lowest spend and the shortest runs. Only 1% have been live for a year. The reasons differ by type and are covered in AI UGC ads and animated ads.

So the honest answer isn't "lo-fi beats polished." It's that real beats fake, and both phone-native and produced formats are real.

The casual first word

The clearest lo-fi signal in our data isn't about the camera. It's about the voice.

121

video winners that open with a casual word

17.4%

of them reach the top tier

8.6%

top-tier share for all other video openers

$324K

their median estimated spend ($231K for the rest)

Ads whose first word is "So" (29), "Hi" (21), "Okay" (20), "Hey" (19) or "Oh" (11) reach the top tier at twice the rate of other videos and carry a higher median spend. The group spans 89 brands, so it isn't one advertiser's habit.

This is a transcript match, not a controlled test, and a casual opener probably travels with other things that help: a real person on screen, a natural setting, a problem told from experience. But that's the point. The casual word is a marker for an ad that sounds like a person.

True ClassicMeta ad · winning
Okay, 5'10, 200 pounds.
Est. spend
$4M
Days live
154
Format
Video, 58s
Variants
8
  • Solution-Aware
  • Question Opener
  • Wrong Question

A top-tier ad with eight variants. "Okay" plus height and weight is how someone starts a try-on video for friends, and the viewer with the same build is instantly in.

Emily Hoskins ReviewsMeta ad · winning
Hi, I think this is the best gift I've gotten all year.
Est. spend
$4.9M
Days live
217
Format
Video, 51s
Variants
3
  • Warning/Shock Opener
  • Offer Pressure

A top-tier ad run from a reviewer-style page. "Hi, I think" is hedged and personal, which reads as an opinion rather than a claim.

SmartCartingMeta ad · winning
So I always give people gifts, but I finally got one for myself.
Est. spend
$2.5M
Days live
142
Format
Video, 21s
Variants
2
  • Most-Aware
  • Offer-First Opener
  • Root Cause

A 21-second top-tier ad. "So" drops the viewer mid-conversation, and the line sounds like something said to a friend, not a tagline.

Where polish loses: the short brand spot

At the polished end of the board sit short spots from large consumer brands: snacks, personal care, marketplaces. These are the most "produced" ads we track, with professional sets, actors and brand-safe scripts.

GroupAdsTop-tier shareMedian est. spendMedian days live
Videos of 15 seconds or less1094.6%$236K98
All videos1,9688.8%$236K119

They get spend, the same median as the average video. They rarely reach the top tier and they run shorter. Brand spots are built for reach and recall, so they aren't trying to win the same game. If you are a DTC brand selling on direct response, copying their look gets you their weakness.

Lay'sMeta ad · winning
The FIFA World Cup is back, America.
Est. spend
$617K
Days live
119
Format
Video, 6s
Variants
2
  • Unaware
  • Specific Number Opener
  • Systematic Breakdown

A polished six-second spot with a cultural tie-in. Solid spend, a four-month run, and no room for a problem, a demo or an offer.

Where polish wins: proof

Polish isn't the enemy. Our product demo ads are among the longest-running formats on the board, and a demo works because the viewer can see the product clearly. Clean light, a steady frame and good audio make a demonstration believable. Founder and expert videos (12.4% top tier) are often shot in a tidy setting with decent sound.

The rule the winners seem to follow: produce the proof, keep the voice casual. The camera can be good. The person should still sound like a person.

A practical guide to the lo-fi look

ElementLo-fi winnersWhat to avoid
Opening line"Okay," "So," "Hi," a mid-thought startA tagline, a brand name, a slogan
SpeakerA customer, founder or expert talking off-scriptAn actor reading a script
SettingThe place the problem happensA studio set or a plain paper backdrop
CameraPhone or simple camera, eye levelCinematic moves, drone shots
LightWindow light, lamp lightVisible studio lighting
Product shotsClear, close, in a real handRendered or floating packshots
CaptionsNative-style burned-in captionsHeavy motion graphics
AudioClean voice, room toneLibrary music under every line

How to apply it:

  1. Script the beats, not the words. Give the speaker the problem, the turn and the offer. Let them phrase it. That is how you get "Okay, 5'10, 200 pounds" instead of a line no one says.
  2. Record the first word. Do several takes that open casually. Test "So," "Okay" and a mid-sentence start as variants.
  3. Spend the budget on proof. Better light and audio for the demo. Not on sets, actors or motion graphics.
  4. Avoid synthetic people. AI and animation can illustrate a mechanism. They shouldn't be your spokesperson.
  5. Don't copy big-brand spots. Their format is built for reach. Yours needs to sell.
  6. Keep polished cuts for retargeting and images, where the viewer already knows you.

For the creator side of lo-fi production, see UGC ads examples and talking-head ads. To see which openers and production styles competitors in your niche are using, search transcripts on the Ad Radar board for words like "Okay" or "So" and sort by estimated spend.

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.

Questions

What are lo-fi ads?

Lo-fi ads look like they were made by a person, not a studio: shot on a phone, natural light, real rooms, someone talking the way they would to a friend. On Meta they usually take the form of UGC selfie videos, talking heads, casual demos and creator skits. The opposite is the polished brand spot with actors, sets and a voice-over script.

Do lo-fi ads perform better than polished ads on Facebook?

On our board, phone-native ads have the highest median estimated spend ($482K) and a top-tier share in line with the board (10.2%). Produced formats, such as footage with a voiceover or a founder video, reach the top tier slightly more often (12.2%). Short polished brand spots and AI-made visuals do worst (4.4 to 4.6%).

Does the way an ad opens matter for lo-fi?

It seems to. Video ads on our board whose transcript starts with a casual word (Okay, So, Hi, Hey, Oh, Yeah) reach the top tier at 17.4%, against 8.6% for all other video openers, across 121 ads from 89 brands. A conversational first word signals a person, not a commercial.

Should I stop making polished ads?

No. Polish helps where the ad has to prove something: a clean product shot, a readable demo, clear audio. What the data argues against is polish that makes the ad look like an ad: actors reading lines, glossy sets and six-second brand spots. Keep the production quality, lose the commercial tone.

Keep reading

See the ads that are already winning.

Membership is $49 every 4 weeks at the launch price. Cancel anytime.