Blog/Creative
All posts

Author

CM

Connor McKit

Published

April 19, 2026

Read time

13 min

Tags

CreativeAgency opsProduction

Creative at 7-figure spend: the real production math.

Honest numbers on volume, hit rates, and production cycle time from running creative engines against $1M/month+ accounts. Why most agencies under-invest, where AI tools actually help, and what the weekly creative review nobody protects is worth.

A lot has been written about creative testing. Most of it was written by people running $20K/month accounts, and most of it breaks in specific ways once you're running $1M/month. This is a field report on what creative production actually looks like at 7-figure monthly spend, including the volume, the hit rates, the production rhythm, and the specific things that go wrong that you won't read about in the usual “creative is king” posts.

We're writing this because the gap between “creative matters a lot” and “here is the operational engine that produces good creative at scale” is where most 7-figure accounts leak margin. The top-line advice is right. The operational specifics are where the money is, and very few agencies we've seen have written down what actually works.

The short version

  • At 7-figure monthly spend, you need to ship 30+ new ads a month per active audience-offer pair.Most agencies ship 6–10. That gap is the agency's margin and the client's performance ceiling.
  • Creative hit rate is lower than you think. 1 in 7 concepts becomes a winner; 1 in 3 concepts becomes a “workable.”If you're planning capacity off a 50% hit rate, you're planning a deficit.
  • The production pipeline, not the creative director, is the bottleneck. Most agencies have ideas faster than they can ship finished variants. Brief-to-shipped cycle time is the single most important production metric to measure and optimize, and nobody measures it.
  • Creative variants are not a substitute for creative concepts.AI-generated variants on a single concept don't replace net-new concepts. They scale the winner you already have. That's valuable but it's a different problem than finding the next winner.
  • The weekly creative review is the highest-leverage hour in a 7-figure agency's week. Agencies that skip it lose 10–15% of their performance to pattern-blindness within a quarter.

Finding 1

The volume number that actually works.

Saturation is the physical reality that forces creative volume at scale. Every ad eventually exhausts its ability to incrementally perform against an audience. At $20K/month that exhaustion takes six to eight weeks. At $1M/month it takes ten to fourteen days. The difference is audience density: you're hitting the same eyeballs more often at higher spend.

Rough rule of thumb we've validated across different verticals at 7-figure spend: you need roughly one new ad shipped per $30K–$40K of monthly spend, per audience-offer pair, per month. That gives you enough variants in rotation to refresh the winners as they decay and to have challengers tested before the winners actually die.

30–35

New ads/month per $1M/month bucket

10–14 days

Typical winner lifespan

3–5 days

Typical ad test cycle to read

4 weeks

Forward creative runway floor

“Forward runway” is the specific metric that separates a functioning creative engine from a dysfunctional one. It's the number of weeks of tested, ready-to-launch ads sitting in the hopper, waiting for the current winners to decay. A team with 4 weeks of runway can absorb a creative collapse without a performance hit. A team with 1 week of runway is one bad week from a panic cycle.

Finding 2

Hit rates are lower than the case studies say.

If you read enough agency case studies, you come away with the impression that most ads are winners. They're not. The published case studies are the winners by selection bias. The dead-on-arrival concepts don't get tweet threads.

Our honest distribution across 7-figure-spend accounts, measured across thousands of ads:

~14%

Concepts that become winners (outperform account average by 20%+)

~30%

Concepts that become workable (match account average within 20%)

~56%

Concepts that die in testing (underperform by >20%)

Those percentages are averaged across verticals we've worked in. Some verticals run tighter (home services concepts hit at closer to 20%), some run looser (aspirational lifestyle categories closer to 10%). But the shape is always more dead-on-arrival than agency marketing would have you believe.

Two operational implications:

  1. Kill speed is as important as test speed. Concepts that are DOA need to die before they eat real spend. Our kill criteria at 7-figure spend: any ad crossing $1,500 in spend with a CPL 40%+ worse than the account average, assuming reasonable impression volume, dies. That's a $1,500 learning cost per dead concept. At 1 in 7 hit rate, you're spending $9,000 in dead-testing spend per winner you identify. Budget for that; don't be surprised by it.
  2. Winner concentration is a feature, not a bug. When you do find a winner, it should be carrying 40–60% of the account's spend until it decays or a better challenger emerges. Trying to run 10 equal-weight ads at 7-figure spend is ego, not strategy. Concentrate where the math says to.

Finding 3

Brief-to-shipped is the metric that matters.

Every agency we've worked with measures the wrong things in creative production. They measure how many briefs they wrote, how many meetings they had, how many concepts “are in development.” The single metric that actually correlates with creative engine health is: how long does it take from brief written to variant live in Ads Manager?

Healthy shops we've seen: brief-to-shipped under 7 days for statics, under 14 days for UGC video, under 21 days for produced video.

Unhealthy shops we've seen: brief-to-shipped measured in months, often stuck in rounds of internal review before anything touches the platform. If your agency has concepts from six weeks ago still in review, your engine is broken. The winner you would have identified four weeks ago is now four weeks closer to irrelevant.

The specific things that slow shops down

  1. Too many rounds of internal review before client review. Agencies that do 3–4 internal rounds before the client sees anything have process inflation. One internal review pass, one client review pass, ship. Trust the brief and trust the team.
  2. Client review cycles that aren't on a schedule.A client that reviews creative “when they have time” is a client that loses 3–7 days per round to calendar Tetris. Put creative review on a standing Tuesday/Thursday weekly cadence; bank-deposit predictability.
  3. Designer-is-a-bottleneck-for-everything syndrome.If your one designer is producing every static, every video thumbnail, every LP asset, and every email, they're the bottleneck for four departments. At 7-figure spend, you cannot single-thread creative production through one human.
  4. No reusable templates or modular components. A shop that re-designs from scratch every time ships 2–3x slower than a shop with a library of proven modular components that get remixed. The library is an asset. Build it.
  5. Revision requests from non-decision-makers. If creative gets to client and the account manager requests changes that then get reversed by the client the next day, the AM is imposing their taste on the pipeline. Kill that.

Finding 4

AI creative tools: what they're actually for.

AI creative tools (AdCreative.ai, Pencil, Creatopy, the raft of 2024–2026 startups) are positioned as the solution to the creative volume problem. They are a solution to one specific piece of the volume problem, and treating them as a complete solution is a mistake.

What they're genuinely good at, based on heavy hands-on use:

  1. Scaling a known winner into 20 variants cheaply. Once you have a concept that performs, AI tools can produce copy-color-layout permutations fast. Real time savings; real impact on fatigue rotation; worth the subscription.
  2. Generating first-draft static ad concepts from a brief.Usually 2 out of 10 AI-generated concepts are worth refining. That's real productivity if your bottleneck is blank-page syndrome. It's fake productivity if your bottleneck is production capacity.
  3. Rapid copy A/B variants on existing creative. Headlines, descriptions, CTAs — AI tools produce usable alternates quickly. Worth using.

What they're genuinely not good at, yet:

  1. Generating net-new winning concepts.The hit rate on purely AI-generated concepts is closer to 1 in 15 than 1 in 7 in our experience. If your feed is 100% AI-generated, you're testing more concepts and finding fewer winners. The math works out worse, not better.
  2. UGC-style video.The uncanny-valley tax on fully-AI-generated UGC video is real and meaningful on performance. Performance ad audiences are trained on a certain authenticity signal, and current AI video still reads as synthetic. It will change; it hasn't yet.
  3. Brand consistency at scale.AI tools will happily produce variants that drift from the client's brand in ways a human would catch. If you're running AI tools without a human QA layer, expect occasional embarrassing ships.

Finding 5

The weekly creative review nobody wants to protect.

Every agency we've worked with has tried a weekly creative review at some point. Most of them have let it slide. The meeting gets moved for urgent client asks, then moved again, then canceled, then forgotten. The consequence is slow and invisible until one quarter your winners all start looking like each other and nothing new is landing.

The specific format that has worked across every shop we've seen protect it:

  1. 60 minutes, standing weekly meeting.Buyer, strategist, creative lead. Same day, same time. Doesn't move for client calls.
  2. First 15 minutes: last week's data. What shipped, what launched, what died, what's emerging. Data-driven, not opinion-driven. Buyer pulls the numbers before the meeting.
  3. Next 15 minutes: pattern spotting. What themes are winning? What themes are dying? Why? Is there a cohort of winners that all share a specific hook, format, creator? Is there a cohort of losers that all share a specific weakness?
  4. Next 20 minutes: next week's bets. Given what we just learned, what 8 new concepts are we briefing this week? What 3 are we killing? What is everyone responsible for by Friday?
  5. Last 10 minutes: runway check.Are we still 4 weeks out on the creative hopper? If we're at 2 weeks or less, what's the catch-up plan?

The specific discipline is the meeting happens weekly even on weeks where nothing dramatic changed. The cumulative pattern recognition is the entire point. You can't see the week-over-week drift if you only look when there's a fire.

Finding 6

The economics: why creative investment is the hardest sell.

If creative is this important, why do most agencies under-invest? The honest answer is retainer structure. Most agency retainers are set up so that the agency collects a management fee as a percentage of spend and then eats creative production out of that fee. At $15K/month management fee on $1M/month spend, every dollar the agency spends on creative production is a dollar out of their margin.

This creates a specific perverse incentive: the best thing for the client's account is the worst thing for the agency's P&L.

The ways we've seen sophisticated agencies handle this:

  1. Separate creative retainer. Creative production is billed as a separate line item, often at cost-plus-a-margin. Removes the perverse incentive; makes creative production a line item the client can scale with spend.
  2. Pass-through creative with a coordination fee. Client pays the production vendor directly; agency charges a coordination fee for briefing, reviewing, and shipping. Less margin but cleaner structure.
  3. Per-ad production pricing. Agency ships X ads per month at a fixed cost per ad. Predictable for the client, profitable for the agency, and easy to scale up or down based on performance.
  4. Internal AI-assisted studio.The agency invests in its own AI-assisted production capacity as a margin play. Higher upfront cost, lower marginal cost per ad over time. The shops doing this well are quietly eating the lunch of shops that aren't.

The structure that consistently loses money: all-inclusive management fee that has to absorb creative production with no transparent pricing to the client. The client never sees the real cost of creative, the agency can't afford to produce enough, and performance degrades over time in ways neither side understands.

What this means

The operator's checklist.

If you're running or building a creative engine for 7-figure monthly accounts, here's the compressed list:

  1. Measure brief-to-shipped cycle time.In days. Weekly. Post it on the wall. If it's growing, the engine is silting up and needs maintenance before the next quarter.
  2. Plan capacity off a 1-in-7 hit rate, not whatever hit rate the case studies sell. A winner costs you ~$9K of dead testing spend to find. Budget accordingly.
  3. Protect the weekly creative review like a board meeting.It doesn't move for client emergencies; the client emergencies are what the weekly review prevents.
  4. Track your creative runway in weeks. Four weeks minimum. Less than two weeks is a flashing red light.
  5. Use AI tools to scale winners, not to find them. The concept hunt is still a human job. The variant production is where AI earns its keep.
  6. Re-structure the retainer so creative is a visible line item. Hidden creative costs inside management fees is a slow failure mode for both sides.

The short version: a functioning creative engine is what separates the agencies that hold 7-figure accounts from the ones that lose them at renewal. The engine is less glamorous than the creative — it's briefs, cycle times, runway, kill criteria, review cadences. But the engine is what produces the creative. Most shops skip the engine and get beaten by shops that didn't.

Hit-rate numbers and cycle-time ranges are averages across agency books we've worked on directly, normalized across different verticals. Specific agency structures will see tighter or wider variance depending on audience maturity, vertical, and creative strategy. We'll update this post when we're wrong.

What Zeke is

The AI client reporting system this research points toward.

Branded AI reports, source-linked QA, client context, AM talking points, and client-ready monthly narratives. Founder pilot $497/mo. Starter $197/mo. Growth $297/mo. Scale $497/mo. No per-seat pricing.

More from Zeke

08 / Pilot

Run one client through it.

The brief has to earn the renewal before anyone expands.

Bring one client. We’ll show the brief that keeps them.

The founder pilot starts with one client: the inputs you already have, the booked-revenue attribution, the Client Intelligence Brief Zeke produces, the QA flags, and the talking points your AM takes into the renewal call.