---
title: Meta creative testing setups at $5K, $50K and $1M+ per month
slug: meta-creative-testing-by-budget
summary: Concrete creative-testing configurations by spend level — one ASC with ~10 ads at lean budgets, an ABO testing campaign with one ad set per concept (3–6 ads, $100–200/day) at growth, and 10–20 tests/week at $200/day for ~7 days at scale — plus rules on test length and not mixing new ads with proven ones.
domain: marketing
tags: [meta-ads, creative-testing, budget, abo, cbo, advantage-plus]
related: [meta-push-delivery-and-creative-testing-tool, meta-strategy-by-spend-level, meta-breakdown-effect-judge-in-aggregate, meta-account-structure-2026, meta-ads-per-ad-set-and-budget-minimums, meta-scaling-winners, meta-weekly-cadence-and-attribution, meta-iteration-vs-net-new-creative, paid-social-creative-testing-framework, meta-ads-creative-testing-engine]
sources: [yt:7knQyPYLmfo, yt:aCkulCPAYzw, yt:kn4ByanAe-s, yt:9mUwsvSK0NY, yt:du6rDgRf06g, yt:C0R5YJiWpig, yt:r51Tj5qQGa0]
updated: 2026-10-24
---

# Creative testing by budget

## TL;DR
| Stage (Dara Denney) | Setup | Ads | Budget |
|---|---|---|---|
| **Lean** ≤ ~$5–15K/mo | **One** ASC/Advantage+ sales campaign optimising for purchases | Start with **10**, mostly statics; add 1–2 per week | campaign-level |
| **Growth** ~$1K/day, $30–50K/mo[^21] | Keep the ASC + add a **manual testing campaign**, **one ad set per concept** | **3–6** per ad set (3 at the low end) | **$100–200 per ad set/day**, usually ABO |
| **Scale** $1M+/mo | Still **one** testing campaign (even at $5M/mo), nearly all ABO, sometimes cost caps | **10–20+ tests per week** | ~**$200/day per test for ~7 days or ~$1,000** |
[^1][^2][^3][^4][^5]

- Never start with traffic, engagement or add-to-cart objectives — go straight for purchases.[^6]
- Don't drop new ads into an ad set that is already scaling; test in separate ad sets/campaigns and move winners over.[^7][^8]
- Give tests **5–7 days** (no need for 14); kill early only clear disasters or scale clear early winners.[^7][^9]

## Details
**Lean stage.** One ASC because it's the most consolidated setup — helps exit the learning phase on a small budget, lets you test more creatives, and is easiest for a solo founder.[^10][^11] Static types to start with: benefit call-outs, us-vs-them, headline ads, "ugly" ads, testimonial statics; include existing UGC or founder video that did well on TikTok/Reels, but hold off on lots of new video until you know which messages convert. Only turn ads off if they miss the CPA goal *and* are getting meaningful spend.[^12]

**Growth stage.** Each ad set is one concept test; variations inside take big messaging swings (videos: different hooks and lengths).[^13] ABO vs CBO is a media-buyer choice — Dara prefers ABO to get a read within a few days by throttling spend; some 6–7 figure brands successfully use CBO or a one-campaign method.[^3] The testing campaign isolates unproven ads from the scaling ASC: she has repeatedly seen performance tank after adding a test to a working campaign.[^14]

**Scale stage.** Structure barely changes — volume, creative diversity and team size do. Agencies that demand their own testing campaign do it for reporting, not performance. Cost caps appear because brands know their target CPA and only want the best new ads to spend.[^15][^4]

**Zach Stuck's version (Mars Men).** Budget a concept test at **$300/day** (set as a multiple of AOV and expected purchases over 2–3 days), with ~5 different statics on the same pillar inside; CBO with a minimum spend per ad set so breakouts can take more budget. Testing budget drags blended ROAS down, so cap it in the financial model.[^16][^17][^20]

**Smart Marketer (Jan 2026).** Meta favours ads with history and social proof, so new ads need their own room to breathe; test new creative in separate ad sets/campaigns for ~7 days, then move winners into the main campaign; keep cold and warm ads separate (with exceptions for ads carrying heavy social proof).[^8][^18]

**Ben Heath's budget-based version.** At ~$100/month: ~one meaningful test a month on one variable. At ~$100K/month: carve **10–30%** into separate testing campaigns. Heath Media overall puts 30–40% of $100M+/yr into testing; his default test share is 10–50% of budget (100% if nothing converts). In accounts that have it he now prefers Meta's **push delivery** (default 20% of budget, 7 days) over extra campaigns — see [[meta-push-delivery-and-creative-testing-tool]].[^22][^23][^24]

- **Test length:** Smart Marketer and Dara's scale brands use ~7 days; Zach reads concepts in 2–3 days at $300/day; the app-focused [[paid-social-creative-testing-framework]] (Ivy, via Steve P. Young) uses 3-day tests with 4 creatives at ~$50/ad set. Shorter tests need a higher daily budget per concept to get enough purchases.
- **Separate testing campaign or not:** Dara and Smart Marketer isolate tests from the scaling campaign; Jon Loomer says low-volume accounts should stop "unnecessary creative testing" — tests on shallow numbers are "almost always meaningless" and water down budget — and publishes new batches straight into the one ad set, judging in aggregate ([[meta-breakdown-effect-judge-in-aggregate]]).[^25][^26] Ben sits between: separate testing campaigns only at high spend, in-ad-set tools otherwise.[^22]
- **Ads per test:** Dara's 3–6 per concept ad set vs Smart Marketer's 15–20 per ad set — different jobs (isolated concept test vs consolidated portfolio). See [[meta-ads-per-ad-set-and-budget-minimums]].
- Dara's stage thresholds are ecommerce; she notes 9 out of 10 times a failing account has a creative problem, not a structure problem — but if a setup keeps failing, try another.[^19]

## See also
- [[meta-scaling-winners]] — what to do once a test wins
- [[meta-account-structure-2026]]
- [[paid-social-creative-testing-framework]] — the app-side version
- [[meta-push-delivery-and-creative-testing-tool]] — in-ad-set testing tools
- People: [[people/dara-denney]], [[people/zach-stuck]], [[people/molly-pittman]], [[people/ben-heath]], [[people/jon-loomer]]

[^1]: yt:7knQyPYLmfo @ 00:00 — Dara Denney
[^2]: yt:7knQyPYLmfo @ 01:12 — Dara Denney
[^3]: yt:7knQyPYLmfo @ 07:44 — Dara Denney
[^4]: yt:7knQyPYLmfo @ 12:22 — Dara Denney
[^5]: yt:7knQyPYLmfo @ 13:53 — Dara Denney
[^6]: yt:7knQyPYLmfo @ 01:12 — Dara Denney
[^7]: yt:kn4ByanAe-s @ 31:24 — Molly Pittman
[^8]: yt:kn4ByanAe-s @ 32:58 — Molly Pittman
[^9]: yt:7knQyPYLmfo @ 13:53 — Dara Denney
[^10]: yt:7knQyPYLmfo @ 02:55 — Dara Denney
[^11]: yt:7knQyPYLmfo @ 03:45 — Dara Denney
[^12]: yt:7knQyPYLmfo @ 02:00 — Dara Denney
[^13]: yt:7knQyPYLmfo @ 09:16 — Dara Denney
[^14]: yt:7knQyPYLmfo @ 09:16 — Dara Denney
[^15]: yt:7knQyPYLmfo @ 10:50 — Dara Denney
[^16]: yt:aCkulCPAYzw @ 22:50 — Zach Stuck
[^17]: yt:aCkulCPAYzw @ 23:41 — Zach Stuck
[^18]: yt:kn4ByanAe-s @ 34:34 — Molly Pittman
[^19]: yt:7knQyPYLmfo @ 15:04 — Dara Denney
[^20]: yt:aCkulCPAYzw @ 17:52 — Zach Stuck
[^21]: yt:7knQyPYLmfo @ 06:26 — Dara Denney
[^22]: yt:du6rDgRf06g @ 00:48 — Ben Heath
[^23]: yt:du6rDgRf06g @ 06:23 — Ben Heath
[^24]: yt:9mUwsvSK0NY @ 01:15 — Ben Heath
[^25]: yt:C0R5YJiWpig @ 02:32 — Jon Loomer
[^26]: yt:r51Tj5qQGa0 @ 04:49 — Jon Loomer