Ten creative variants per campaign, not two.
The AI production pipeline generates a full variant set from one locked concept — hook variations, different opens, static and video cuts, platform-native formats for Meta and LinkedIn.
UNIBRIGHT SOLUTIONS · AI VIDEO PRODUCTION · CASE STUDY
7.4x ROAS, 70 Leads, Just 10 Days
Every advertiser knows creative is the biggest lever in paid media. Almost nobody tests enough of it to find out which creative works.
7.4xROAS
70+10-DAY LEADS
200CREATIVES TESTED
10%OF COST ON CREATIVE
01 / THE PROBLEM
The Problem
Every advertiser knows creative is the biggest lever in paid media. Almost nobody tests enough of it to find out which creative works.
The reason is arithmetic, not ignorance. If a video costs real money and real weeks to produce, you test two or three variants per campaign — because that is what you can afford. Three variants is not a test. It is a guess with a spreadsheet attached. You cannot identify a winning hook from a sample of three, so budget goes behind whichever one performed least badly, and the campaign runs until it fatigues.
We had the same problem on our own account. So we solved the production side first, then ran the test we actually wanted to run.
BUILT
What We Built
Hook variants as the primary test axis.
Curiosity, fear and authority openers scripted against the same core offer, because the first three seconds decide the result and that is the cheapest thing to vary.
Storyboard-first, so variants stay comparable.
Every variant in a set shares runtime, structure and offer. Only the tested element changes, which is what makes the read clean instead of noisy.
Video and static from one production cycle.
Stills and image ads pulled from the same render as the video, so a campaign gets both formats without a second brief.
Creative production held to 10% of campaign cost.
Ninety percent of budget stayed in media, where it belongs. The pipeline is what makes that split possible — conventional production would have eaten a multiple of it.
Kill and scale rules set before launch.
Each variant set runs to a defined threshold, losers are cut, budget consolidates behind the winner. Testing volume is worthless without a decision rule attached to it.
05 RESULT
The Results
RESULTDETAIL
7.4x ROASReturn on ad spend across the campaign set
70+ leads in 10 daysRoughly 7 per day, across Meta Ads and LinkedIn Ads
200 creatives in test10 variants per campaign across 20+ campaigns
10% of campaign cost on creativeVideo and image production combined — 90% of budget stayed in media
Winning creatives identified per campaignFound by testing rather than picked upfront, then scaled
MEANS
What This Means
Testing volume is a production problem.
Nobody chooses to test three creatives. They test three because that is what production cost allows. Fix production and the testing discipline changes on its own.
Cheap creative does not mean cheap-looking creative.
The 10% figure is not a quality trade. Specs are built once and reused across a variant set, so the tenth variant costs a fraction of the first while looking identical in build quality.
Winners get found faster.
Ten variants reach a decision in days rather than the weeks it takes to sequence three separate production rounds.
FIX
What This Does Not Fix
Worth being straight about:
Two platforms, one production cycle.
Meta and LinkedIn need different formats and different registers. Both came out of the same variant set.
The two headline numbers belong together.
7.4x return on a 10% creative allocation is the whole argument. Cheap variants did not dilute the return — having enough of them is what found it.
We ran it on ourselves first.
This was our own brand, our own budget, our own leads. We are not selling a method we have not paid for.
WORKS
Why This Works Elsewhere
Any advertiser with a media budget and a creative bottleneck has this problem. The symptom is recognisable: campaigns that run until they fatigue, a creative refresh cycle measured in months, and a testing plan that never survives contact with the production calendar.
The fix is not a better creative brief. It is enough variants to make the test meaningful, at a cost that leaves the media budget intact.
NEXT STEP
The Next Step
Tell us your monthly media spend and how many creative variants you currently test per campaign. If the second number is under five, the first number is not working as hard as it could.
You get back a variant plan, a creative cost as a percentage of media spend, and the testing structure we would run.
Services Provided:
AI video production, PPC creative production, hook and script writing, static and video ad creative, Meta Ads, LinkedIn Ads, creative testing structure and campaign management