How Celebrity Performance
Ads Helped GoodScore Scale

Same account. Same auction. Same scoreboard: cost per conversion. The celebrity won.

ClientGoodScore

Timeline90 Days

Outcome50X spend

01THE SETUP

Goodscore already
had a creative machine.

A fast-growing fintech app, running hundreds of UGC, POV and AI-led ads, one of the most aggressive testing programs in the category. But the next challenge was to find scale efficiently.

So the question wasn’t “can a celebrity get reach?”

It was: can one trusted face beat a mature performance engine on cost per conversion?

REACH IS EASY. EFFICIENCY IS THE TEST.

02THE SYSTEM

Data-Backed Face and Script Selection

1shoot day
an 8-hour shoot, designed for maximum output
6scripts
each mapped to a real user intent and structures derived from winning ad formats
3hooks per script
openings tested independently
18live ad assets
cut for Meta feed + YouTube, 16:9 and vertical

The celebrity got no exemption from testing discipline. Every hook launched, measured, killed on data.

03THE HEADLINE

27% of ad spend.
30% of the conversions.

CELEBRITY ADS’ SHARE OF THE TOP-50 COHORT

A bigger share of results than of budget, on both platforms at once.

04AND CHEAPER, TOO

Creative costs down 75%.
Conversions and installs cost less, too.

META · COST INDEXED, REST OF ACCOUNT = 100

100

restcelebrity

cost / conversion

100

restcelebrity

cost / install

The install-cost gap is the tell: trust pre-sells before the click.

05FULL-FUNNEL

The advantage held at every stage.

STOPHook ratea recognised face earns the pause+13%
WATCHCompletion rateviewers stay through the message+27%
ACTClick → conversionclicks arrive pre-sold+15%
COMMITInstall → registrationhigher-intent users stick65% vs 57%

Not a novelty spike at the top of the funnel. The lift compounds all the way to a registered account.

06THE HERO AD

One script took the #1
slot on both platforms.

Scaling an ad usually taxes efficiency. This one absorbed the account’s highest single-ad spend of the quarter, and still converted cheaper than everything around it.

−28%

cost per conversion vs. the account average, in its launch month, while running as the single most-scaled ad on the platform.

Then the same script was re-cut for YouTube, and became that channel’s most-invested asset too.

07THE OTHER CHANNEL

On Google, the win was scale without decay.

4 of 5

top assets by investment were celebrity cuts: the auction voted with budget.

37%

of channel spend absorbed at cost-per-conversion parity. Scale that normally degrades efficiency didn’t.

2.2×

the click-through rate of non-celebrity assets.

Meta became the efficiency lever. Google became the scale lever. One shoot, both jobs.

08WHY IT WORKED

Celebrity as a performance asset, not an awareness asset.

01
Credibility transfer
a face the audience already trusts, matched to a product built on trust
02
Performance-first scripting
scripts mapped to real user intents, not brand monologues
03
Hooks tested like UGC
launched and killed on data, no celebrity exemption
04
Channel-role split
efficiency on Meta, scale on Google, from one shoot