Targeting moved into the ad. Creative is the audience now.

Media buyers used to express an audience with settings.

Age bands, interests, layered exclusions, a saved audience with a name that made sense to whoever built it.

Delivery on Meta now leans broad.

The system works out who to show an ad to largely from the creative and the early response to it.

The audience is inferred rather than declared.

Which relocates the entire skill. A hook about being tired at 4pm finds tired people.

A hook about a spreadsheet finds people with a spreadsheet problem.

Those are audience decisions written as sentences.

So an account with four creatives is not an account testing four ads.

It is an account addressing four audiences and hoping one of them is large.

Take your best performing ad and read its first line. Then describe the person it would find. That description is your real targeting document.

Published benchmarks put the share of ad creatives that become winners at roughly 5 to 8 percent.

That comes from Motion's analysis of 550,000+ Meta ads. One in thirteen to one in twenty.

Six creatives a month against that rate is a lottery with six tickets.

The monthly review where you argue about which hook was weak is a meeting about sampling error.

Nothing in a better brief fixes it.

Sharper writing moves the odds a little. Twenty-six attempts moves them a lot.

Only one of those is available to a team that produces six.

The useful reframe for a review is that the villain is arithmetic rather than anybody's creativity.

That is checkable, and it moves the conversation to the only lever that works.

Film one talking-head take of up to three minutes.

Upload it and Cutroom transcribes the read, drafts the cut, places b-roll against the words and returns a finished 9:16 MP4 with captions burned in.

You direct on the transcript. Highlight a phrase and a clip lands over exactly those words.

Delete a line and the cut rebuilds. Change the pace and it re-cuts the whole thing.

For Meta specifically, the variant that matters is the opening.

Same body, twelve different first sentences, each aimed at a different situation the viewer might be in.

That is twelve audience definitions produced from one filming session, at about 110 credits each.

There is no timeline, no keyframes and no layers, so a variant is a decision rather than an afternoon.

Your own product footage sits in the workspace library and drops onto the words that name it.

Exports are 9:16, which covers Reels and Stories. Feed and right column placements are made elsewhere.

The second effect of a cheap attempt is not more ads. It is faster deletions.

An ad that cost six hundred dollars has a defender in the room.

It runs until somebody is willing to lose face.

At about 110 credits, nobody defends anything.

The creative that is not working goes off on Wednesday afternoon with no conversation.

That is the behaviour every testing guide asks for and almost nobody manages.

Basic is $39.99 for 2,500 credits, roughly 22 ads. Premium is $79.99 for 5,000, roughly 45.

A $15 top-up adds 1,000 credits when a week runs hot.

The trial is 7 days and 300 credits with no card, which is two finished ads to run against whatever is currently winning.

Set a kill rule before anything goes live and write it down.

Exports are 9:16 MP4. That covers Reels and Stories placements, which is where vertical video earns its keep.

Square and widescreen assets for feed placements are produced elsewhere.

The source is one talking-head take capped at three minutes, and one project makes one finished ad.

Static image creative, carousels and catalogue ads are outside this entirely.

The b-roll search covers millions of free clips for the world around the claim.

Your product, your interface and your own footage go into the workspace library.

From there they land under the exact words that describe them, with one highlight.

The caption treatment is your look, across six packs plus colour, weight, size and position.

In a muted feed that treatment is most of what a viewer registers as brand, and it costs no design round.

Somebody has to be on camera, because the transcript is the interface.

That person is also the reason broad delivery can read the ad at all. A face making a specific claim is the strongest signal the system gets.