Meta Ads Creative Trends 2026: What's Actually Getting Clicks
META ADS · 9 MIN READ

Meta Ads Creative Trends 2026: What's Actually Getting Clicks

The expensive-looking ad is not the winning ad. What performs now looks like something a person made, says one thing, and says it in the first two seconds.

By Tariq Sallam·September 2026

I have spent a lot of the last year comparing creative that cost £2,000 to produce against creative shot on a phone in twenty minutes. The phone footage wins more often than anyone commissioning the £2,000 would like.

This is not an argument for low effort. It is an argument about which effort matters: the angle, the first two seconds and the clarity of the offer, rather than the production.

Quick Info

Best-performing format
Vertical video, 9 to 20 seconds
Most underused format
Plain static with one clear claim
What matters most
The first two seconds
Production level that wins
Native, made-by-a-person, well lit
Concepts per month
Three to five, on different angles
Most common fault
No identifiable offer

What is working: the formats

Short vertical video

Nine to twenty seconds, 9:16, shot on a phone, captioned, sound optional. The dominant format across almost every account I run, and the one most small businesses still do not produce.

Talking head with a claim

One person, one point, straight to camera. Cheap, fast to make, and it consistently outperforms produced brand video in service businesses because it reads as a person rather than a company.

Plain static with one line

Underrated and cheap to test. A single strong claim on a clean background, or a photograph with a short overlay. Frequently the best cost per lead in an account, and almost nobody tests it any more.

Screen recording or demonstration

Showing the thing working. For software, tools and processes this beats describing it by a wide margin.

Before and after

Where your work is visible, this remains the most persuasive format available. Trades, cleaning, landscaping, dentistry, design.

What is not working

—High-production brand film. Beautiful, expensive, ignored. It looks like an advert and people scroll past adverts.
—Stock footage with a music bed. Reads as generic instantly.
—Carousels for anything other than product ranges or step sequences.
—Long text overlays nobody can read at scrolling speed.
—Ads with no identifiable offer. The single most common fault I see, and the easiest to fix.
—Anything relying on sound to make sense.

The ad that looks most like an advert performs worst. That has been true for three years and marketing budgets have not caught up.

The angles worth testing

An angle is the argument, not the aesthetic. Most accounts test the same argument twelve ways and conclude creative testing does not work.

01The problem: name the frustration precisely enough that the right person recognises themselves.
02The outcome: show the after state, not the process.
03The objection: address the thing stopping them. Price, time, trust, disruption.
04Proof: a customer saying it, ideally on camera, ideally badly filmed.
05Price transparency: publishing a number when competitors do not is a genuine differentiator.
06Demonstration: show it working.
07Comparison: how this differs from the obvious alternative.

Seven angles is seven weeks of testing at one a week. That is a year's creative plan, and it will teach you more than any amount of audience experimentation.

The first two seconds

Almost all the variance between a winner and a loser sits in the opening. The rest of the ad only matters to people the opening kept.

—Start with the problem or the outcome. Never with a logo or an establishing shot.
—Show a face or movement immediately.
—Put the claim in the first line of the primary text as well, because a proportion of people read before they watch.
—Captions from frame one, because most viewing is sound-off.
—No slow build. There is no build; there is a scroll.

Customer-made creative

Content that looks like a customer made it continues to outperform brand-made content, and it does so because it looks unproduced rather than because of who filmed it.

01Ask happy customers for a thirty-second phone video. Some will say yes and it costs nothing.
02Give them three questions rather than a script: what was the problem, what did we do, what changed.
03Do not edit out the imperfections. The imperfections are why it works.
04Get permission in writing to use it in advertising.
05Run it against your best produced asset and let the numbers decide.

If you cannot get customer footage, a member of staff speaking plainly to camera is the next best thing and is available today.

AI in creative production

Useful for variation and drafting, weak at producing the thing that wins. Generated visuals still read as generated to a large proportion of viewers, and generated copy defaults to fluent generic.

—Good for: resizing and reformatting, generating twenty headline variants to choose from, background cleanup, rough storyboards.
—Poor for: the hero asset, anything requiring your actual product, anything requiring a real person's face.
—Best combination: real footage, AI-assisted editing and variation.

The AI creative piece covers Meta's own generative tools in more detail.

A creative testing routine that works

01Produce five concepts a month, each on a different angle.
02Run three to six in one ad set and let delivery choose. Do not build one ad set per creative.
03Judge after a week and meaningful spend, on cost per qualified result.
04Take the winning angle and make three variations of it next month.
05Retire ads when cost per result drifts up over a week.
06Keep a document of what won and why. Most accounts relearn the same lesson three times.

The point of the routine is that it does not depend on inspiration. Five concepts a month, forever, is what separates accounts that improve from accounts that plateau.

Production notes that actually matter

—Light. Face a window. This is the entire difference between amateur and unusable.
—Sound, if there is speech. A £30 clip-on microphone fixes it.
—Steady enough. A phone propped against something is fine.
—Vertical. Filming horizontally and cropping wastes most of your frame.
—Keep the raw footage. You will want to recut it three more ways.

Frequently Asked Questions

Do I need professional video for Meta Ads?

No, and it often performs worse. Well-lit phone footage with clear audio and a strong first two seconds beats produced brand film in most accounts.

How many creatives should I test at once?

Three to six in one ad set. Splitting them across separate ad sets fragments your budget and slows learning.

How often does creative need refreshing?

Three to five new concepts a month. Rising cost per result with no other account changes is nearly always creative fatigue.

Does static still work?

Yes, and it is underused. A single clear claim on a clean background is cheap to test and frequently competitive on cost per lead.

Is AI-generated creative worth using?

For variation, resizing and headline drafting, yes. For the hero asset it still reads as generated and underperforms real footage.

What is the most common creative mistake?

No identifiable offer. Plenty of ads communicate a mood and never say what the viewer would actually get.

Before You Go

Make five things a month, on different arguments, shot simply, with the point in the first two seconds. That routine beats every clever targeting idea I have tested against it.

And write down what wins. The cost per lead guide covers where creative sits among the other levers.

Five concepts a month. Point first.

WRITTEN BY TARIQ SALLAM
Marketing Consultant. Entrepreneur. Content Creator.

I'm a marketing consultant, entrepreneur and content creator. I help businesses grow through practical marketing, websites, SEO, content and AI.

More About Tariq →

Keep reading