NORBANTIS

Norbantis / Work / Revise

Creatives & campaigns · AI writing SaaS · US

UGC avatar ads for an AI writing app, tested one variable at a time.

Revise sells subscriptions to writers in the US. For two and a half months we produced its ad creative and ran its Meta campaign: 8 UGC videos and 6 statics, each built to answer one question.

Client
revise.net ↗
Market
United States
When
2026 · 2.5 months
Scope
UGC creative + Meta Ads

The work

One master, six variants.

The same hook across three AI presenters, then one presenter across three hooks. Tap any video to play it.

Avatar A · Same hook, presenter A
Avatar B · Same hook, presenter B
Avatar C · Same hook, presenter C
Hook · Problem · One presenter, problem-first hook
Hook · Objection · One presenter, objection-first hook
Hook · Proof · One presenter, proof-first hook

The brief

Sell a writing tool to writers who distrust AI.

Revise is an AI writing product whose buyers are experienced writers — people who have heard every promise about AI and don’t want their voice rewritten. The account needed creative that sounded like a peer, not an ad, and enough variation to learn what actually moved subscriptions.

What we did

Scripts, presenters and a campaign built to learn.

  1. Angles and scripts

    Three hooks — the problem, the objection, the proof — each written as a short first-person script for the buyer the data pointed to.

  2. AI presenters, reviewed by hand

    Avatar video with voice, word-by-word captions and edit, cut for feed, stories and reels. Every cut checked before upload.

  3. A live A/B/C test

    The same hook with three presenters on equal budget, so the result said which face the audience trusted.

  4. Statics alongside video

    Six static ads to cover placements and give the algorithm more than one format.

  5. Running the account

    Campaign structure, budgets and weekly reads in Meta Ads Manager, with each new batch starting from what converted.

Results

Both targets met.

The two performance targets agreed with the client at the start were met.

+46%return on ad spendbest two weeks vs. campaign average
−23%cost per new subscribersame window vs. campaign average
#2best month in the account’s historyAugust, with us running the account

Running something similar?

Tell us what you're advertising. We'll come back with angles for a first batch.

Start a project

More work