---
title: Paid social creative testing for apps — levers, hook/hold rate, 3-day tests
slug: paid-social-creative-testing-framework
summary: The levers of paid social (Meta/TikTok/YouTube) for apps, hook-rate and hold-rate targets, and Ivy's 3-day creative test (4 creatives per themed ad set, $50/ad set) run on highest-converting weekdays.
domain: mobile-apps
tags: [user-acquisition, paid-social, creative-testing, meta-ads, tiktok-ads, metrics]
related: [app-ad-creative-concepts, meta-ads-budget-math-for-apps, competitor-ad-research-for-apps, meta-ads-readiness-checklist, ad-creative-volume-and-hit-rate, ugly-ads-and-brand-caution, meta-creative-testing-by-budget]
sources: [yt:WaOrPQVLz08, yt:EyoYQU6JyV8, yt:jzlPf100vy4]
updated: 2026-10-24
---

# Paid social creative testing for apps

## TL;DR
- Levers: campaign setup, audience (**keep broad**), placements, creative (video vs static), optimization (**start trial**), attribution, scaling and testing strategy, onboarding/paywall, store listing, budget. **Creative fixes most problems.**[^1][^2]
- Targets: **hook rate 30–40%**; hold rate depends on definition (see below).[^3]
- Test: **3-day window**, **4 creatives per ad set**, ad sets split by **theme**; simplest version: one ad set at **~$50/day**.[^4]
- Run the 3 days on your **highest-converting weekdays**; kill what doesn't perform.[^5]
- Budget for 2–3 months of losses before it works.[^6]

## Details
**Metrics (add as custom columns in Meta Ads Manager).**[^7]
- *Hook rate* — target 30–40%. Steve doesn't define it on screen ("just ChatGPT this"); the usual definition is 3-second video views ÷ impressions.
- *Hold rate* — two definitions: 15-second views ÷ impressions (low numbers expected, as Ivy used), or 15-second ÷ 3-second views (then ~30–40%). Pick one and be consistent.[^3]

**Ivy's testing method** (as relayed by Steve; she reached ~$20K/day at positive ROAS):[^4][^6]
- 3-day test period, **$200–250/day** total across several ad sets.
- **4 creatives per ad set**; each ad set = one theme (e.g. for Cal AI: "calorie deficit"; or women-only vs men-only).
- Simple version: one ad set, ~4 creatives, $50/ad set/day.
- If a creative doesn't perform within 3 days, drop it.

**Pick the 3 days from your data.** One of Steve's apps converts best Sun–Tue; health & fitness tends to peak early in the week; games may peak Thu–Sat. Use your trial-activation-by-day data.[^5]

**Store page alignment.** Matching ad creative to screenshots works well, but is a margin optimization once at scale.[^2]

## Disagreements / caveats
- Post-Andromeda (Meta, 2025–26) much of the targeting happens through the creative itself; Ethan Ethier (Built With Science) found spending *less* time in Ads Manager and more on content was the biggest unlock — consistent with "keep audiences broad"[^8].
- Expect a low hit rate (one operator: ~1 in 85 videos) and test cheaply in similar-converting geos (~$10/day vs ~$1,000/day in the US, per Yuliya Lennox) — see [[ad-creative-volume-and-hit-rate]][^9].
- Ivy's numbers are one founder's method at high spend; Steve promised a deeper setup video later.[^5]
- Steve still recommends Apple Ads first because it has far fewer levers.[^1]

## See also
- [[ad-creative-volume-and-hit-rate]] — how many creatives to expect per winner
- [[app-ad-creative-concepts]] — what to put in the 4 creatives
- [[meta-ads-budget-math-for-apps]] — 50 events/week constraint
- [[competitor-ad-research-for-apps]]

[^1]: yt:WaOrPQVLz08 @ 00:00 — Steve P. Young
[^2]: yt:WaOrPQVLz08 @ 01:35 — Steve P. Young
[^3]: yt:WaOrPQVLz08 @ 02:37 — Steve P. Young
[^4]: yt:WaOrPQVLz08 @ 03:49 — Steve P. Young
[^5]: yt:WaOrPQVLz08 @ 04:51 — Steve P. Young
[^6]: yt:WaOrPQVLz08 @ 02:37 — Steve P. Young
[^7]: yt:WaOrPQVLz08 @ 03:49 — Steve P. Young
[^8]: yt:EyoYQU6JyV8 @ 23:57 — Ethan Ethier
[^9]: yt:jzlPf100vy4 @ 28:32 — Yuliya Lennox