What's the absolute minimum number of creatives I need per week?

Eight new variants per week is the floor for any DTC brand spending above $5,000/month on Meta in 2026. Below 8, you're starving Andromeda's creative-diversity model and you'll see CPA inflate as you scale. If you can only ship 4, you're not 'doing less' — you're paying a 35–45% CAC tax for the privilege.

Should I ship more static or video creatives?

Most accounts win with a 40/40/20 split: 40% static, 40% short-form video (8–30 seconds), 20% UGC or founder-face. Pure-video accounts plateau around $30K/month spend because static still wins certain placements (FB right-rail, IG explore). Pure-static accounts cap at $15K/month because Andromeda's video-completion signal is too valuable to skip.

How do I know if my agency is shipping enough?

Ask for a 90-day creative log: total assets shipped, broken down by static/video/carousel. Then ask how many were tested live (vs. shipped but never run). If the test-rate is below 70% of shipped, your agency is volume-faking — making assets that never see real audience exposure.

What if my budget is too small for 12 creatives a week?

If your monthly Meta spend is under $4,000, the formula gives you ~5 creatives/week — and that's actually fine. The volume rule kicks in hardest above $8K/month spend, when Andromeda's delivery system is making most of the budget decisions for you. Below that threshold, you can win with manual targeting and a smaller creative pool.

How fast should I kill losing creatives?

Use a 4-day kill rule. If CPA at day 4 is greater than your target multiplied by 1.4, kill the variant. Most founders flinch and keep losers running 'just in case.' Don't. The kill rate for healthy testing is 50–65%; if your kill rate is below 30%, you're not testing — you're optimising in place.

How Many Creatives Per Week To Scale Meta Ads in 2026

If you’re running Meta ads in 2026 and your CAC keeps climbing, the question you should be asking isn’t “is my audience burned out?” or “is my offer wrong?”

It’s: how many creatives did I ship last week?

If the answer is less than 8, you have a creative volume problem. Everything else is downstream.

This post breaks down the formula, the math behind it, and the testing protocol that makes the volume actually work.

Why Meta Ads Now Run on Creative Diversity

In 2022, you could win Meta with three good creatives and tight targeting. The platform rewarded the buyer who picked the right interest stack and let the algorithm pace.

That model died with the rollout of Andromeda — Meta’s new ranking model that prioritises creative-level signal over audience-level signal. The buyer’s job shrank. The strategist’s job grew.

Three things changed:

1. Targeting got broader. Advantage+ Shopping and broad-targeting campaigns now drive the majority of D2C spend. The algorithm is choosing the audience inside the campaign. The buyer’s input is the creative pool and the kill rules.

2. Creative fatigue got faster. Frequency caps that used to hit at week 3 now hit at week 1 for high-spend accounts. The same hook that worked Monday is dead by Friday.

3. The Andromeda model rewards diversity. When Meta has more creative variants to choose from, it can match the right creative to the right user. Less variants = a worse match = higher CPA.

The buyer who ships 6 creatives a month and the buyer who ships 40 are not running the same campaign anymore. They’re playing different sports.

The Formula

Weekly creative count = Monthly ad spend in USD ÷ 800

Examples:

$4,000/month spend → 5 creatives/week → 20/month
$8,000/month spend → 10 creatives/week → 40/month
$16,000/month spend → 20 creatives/week → 80/month
$40,000/month spend → 50 creatives/week → 200/month

Where does the ÷ 800 come from? It’s the rough cost — at the typical kill-rate of 50–65% — to find one creative that performs above target. Multiply your target CAC by the kill-multiple, and you get the approximate spend per “winning variant” cycle. At ~$800/win, the equation falls out.

The formula is a guide, not a prescription. Adjust by ±25% based on:

Category fatigue speed — wellness/supplements fatigue faster than premium apparel; bump volume +20%
Offer maturity — new product launches need 1.5x more testing than established SKUs
Account age — accounts under 60 days old need more variants because the algorithm hasn’t found stable winners yet

What Happens Below the Formula

We’ve watched this exact pattern in 5 different client accounts over the last 18 months. A brand spending $10K/month on Meta ships 4–6 creatives in month one. Performance is fine. They decide creative volume isn’t the issue.

They try to scale spend to $15K/month. CPA inflates 35–45% in the first 14 days of scaling.

The diagnosis is almost always the same: not enough creative diversity for Andromeda to find the right match at the higher spend level. The system is forced to over-deliver the same 4 creatives to colder audiences. Frequency spikes, fatigue accelerates, CPA climbs.

The fix is rarely “fix the targeting.” It’s “ship 30 more creatives in the next 14 days.”

Brands that accept this and rebuild their creative ops typically recover within 30 days. Brands that don’t either roll back to the lower spend level (capping their growth) or churn 2–3 agencies looking for someone who can “fix the algorithm.”

There’s nothing to fix in the algorithm. The bottleneck is upstream.

What Happens Above the Formula

Going above the formula doesn’t break anything — it just diminishes returns unless you have a structured testing protocol.

Most agencies that ship 100+ creatives a month do it without a system. Half the assets never get fair exposure (under 1,000 impressions before being killed or buried). Half the data is noise.

If you’re shipping above the formula’s recommended volume, you need:

A clear test framework (4-day windows, defined budget per variant, kill thresholds)
A scaling rule (e.g., promote any variant with CPA < target × 0.85 to a dedicated scaling CBO)
A creative-debt tracker (assets shipped but never tested → kill the production process that’s making them)

Without those, more volume = more noise. With them, more volume = more learning per dollar.

The 4-Day Testing Protocol (Inside Hook & Frame Clients)

This is the testing structure we use across all 12+ active accounts. It works at $4K/month spend and at $80K/month spend.

Day 0 — Launch

Create one CBO campaign labeled “Creative Test — Week of [date]”
3 ad sets inside the CBO
4 creative variants per ad set (12 total)
Daily budget = (weekly testing budget ÷ 4 days)
Audience: broad / Advantage+ / minimal targeting

Day 2 — First kill

Kill any variant where CPA > target × 1.4
Typically kills 5–7 of 12
Don’t second-guess. The data is noisy at 48h but the bottom-5 outliers are almost always real.

Day 4 — Promote winners

For any variant where CPA < target × 0.85, copy into a dedicated “Scaling — [creative name]” CBO
Set daily budget to 120% of what it was getting in the test CBO
Pause the variant in the test CBO

Days 5–14 — Ride and watch

Promoted creatives typically run for 7–14 days before drifting past target × 1.2
When CPA drifts, refresh with a new variant from the next test batch

Repeat indefinitely. A week without a launched test CBO is a week of compounding creative debt.

What the Volume Actually Looks Like

A common pushback: “12 variants a week sounds like a lot. What are we actually making?”

Here’s a real production breakdown from one of our active D2C accounts (Indian skincare, $8K/month spend, 32 variants/month):

14 static images — bold-text overlays, product close-ups, ingredient explainers, before/after stills
12 short-form videos (8–30s) — UGC clips, motion graphics, founder voice-overs on b-roll, kinetic typography
5 founder-face videos (30–60s) — talking head on plain background, captioned, light grade
3 carousels — 5–8 slides each, mix of education and social proof

Total production time: ~12 hours/week split across one designer, one motion artist, and a shared UGC asset library refreshed monthly.

Most D2C brands try to do this with one in-house designer and zero motion capability. That’s the underlying reason they ship 4 creatives instead of 32.

In-House vs. Agency: When Does Each Win?

Quick math on the build-vs-buy question at common spend levels:

Monthly Meta spend	Creatives needed/week	In-house cost (fully loaded)	Agency cost	Winner
$4K	5	~$3K (1 designer part-time)	~$1.5K	Agency (low volume)
$10K	12	~$8K (designer + motion + UGC budget)	~$3K	Agency
$25K	30	~$12K	~$5K	Agency
$60K	75	~$18K (full team)	~$8K	Agency or tie
$150K+	180+	~$35K (manager + 3 producers + UGC)	~$15K	In-house at scale

The crossover point is around $100–150K/month spend. Below that, an agency wins on $/creative because they’re amortising production team costs across multiple clients. Above that, the volume is enough to justify fixed in-house overhead.

The Real Question

If you’re stuck at sub-formula creative volume, the question isn’t “should I hire an agency?”

The question is: do you treat creative as production or as service?

Production means: same process every month, predictable output, hit-rate matters more than individual asset quality, kill rules are mechanical not emotional.

Service means: each asset is bespoke, approval cycles are long, every creative needs to be “right” before it ships, kill decisions are debates not rules.

Production wins in 2026. Every benchmark, every client account, every category we operate in tells the same story.

If you want me to look at your current creative ops and tell you which side of that line you’re on — submit your best Meta ad here and I’ll record a 5-minute Loom for you personally. Free, no follow-up unless you ask: submit your ad for a free teardown.