What Should a DTC Performance Creative Agency Actually Deliver?

A Benchmarking Illusion

If you're a DTC founder vetting performance creative agencies right now, you've probably noticed something frustrating… nobody agrees on how many ads your account needs (or backs up their assertions with reasoning).

One agency promises 50 ads a month. Another says 15. A third talks about "concepts" without defining what that means. When you ask what results should look like, you get a case study that sounds impressive but feels completely disconnected from your brand's vertical and/or account health.

As the founder of Y'all, a DTC performance creative and marketing agency, I've chatted with many brand founders, CMOs, and marketing directors about this very thing. The industry doesn’t have actual standardized benchmarks, partly because determining the right amount isn't straightforward!

So, with that in mind, I want to set what I think is a solid standard. What creative volume actually matters, what structured testing looks like in 2026, and why the relationship between creative and media buying determines whether you scale or stall.

The Ad Volume Problem

Let's start with the question every founder asks first. How many ads should I be getting per month?

I 100% get why people ask it, volume feels measurable. Three hundred ads sounds better than thirty. Unfortunately, the way most agencies approach volume doesn't work with Meta's algorithm.

Meta's Andromeda update rolled out throughout 2024 and into 2025, fundamentally changing how the platform evaluates ads. The update rewards brands for creating genuinely distinct ads rather than minor variations. Meta now groups similar ads and would, for example, treat five product-shot ads with slightly different headlines as essentially one ad.

Pre-Andromeda, an agency could deliver 50 ads that were really one hero video with 49 headline tweaks. That worked when Meta's algorithm treated each variant as a separate learning opportunity.

Now, those 50 deliverables register as a single ad in the algorithm's eyes. Meta's visual recognition models can identify when images with different text overlays are essentially the same creative, and when the system detects low diversity, it looks past any differences in the ads and just serves them to the same audience.

To use an extreme, but very real example, six genuinely different ads now outperform 75 variations of the same concept. Beyond wanting your agency to deliver ads that Meta will read as unique, the only way to scale an account without CAC taking a huge hit is to adapt to this new methodology. On top of that, you can't pull learnings from ads that Meta isn't treating as unique… or you can and will get bogged down with the illusion of learnings.

What "Diverse Creative" Actually Looks Like

When most agencies talk about creative diversity post-Andromeda, they're still thinking in terms of singular format tweaks. That's incomplete, and (you guessed it) it's costing you money.

Sam Lund, a partner at Y'all and our creative director, thinks about it as a matrix rather than a checklist. The way we think about it at Y'all is across three layers at once: Placements, Visual angles, and Messaging angles. 

The point of running all three layers in the same ad set is deceptively straightforward. You're giving Meta's algorithm enough fuel to find the right combination for the right person at the right time. When an ad set only has variations of one concept, Meta has nothing to work with. It picks one and basically goes hunting people down to force them to look at the same ad over and over (which is exactly how CPMs spike).

When you do this right, the whole is greater than the sum of the individual ads. You're running ads that work together as a system. Someone who pauses on a top-of-funnel hook gets served a different angle further down. A static reinforces a video. A creator video reinforces a brand visual. Meta has a network to draw from, and it gets cheaper and smarter the more diversity you feed it.

Volume That Actually Maps to Spend

So, if volume alone isn't the answer, what is?

Creative output should scale with ad spend, that is, how many net-new, Andromeda-optimized, visually unique ads (not including variants!) are needed.

Here's how we think about volume at Y'all. These are rough guidelines; every account has its own creative shelf life, account history and behavior, and eccentricities, but this is the baseline we tend to work off of when first approaching an account for an audit:

With how much I've already talked about it, you shouldn't be surprised to see that "variants" is missing from this table! We always baseline volume based on visually distinct ads.

Testing Without Traditional Variants

"We test everything" might be the most common claim in agency pitches, and it's also usually the most meaningless. Real message testing post-Andromeda doesn't look like the old A/B test logic of "keep everything the same and only swap the hook." That approach actually breaks now, because if your ads look too similar, Meta groups them as one and randomly forces spend to whichever one gets a slightly higher CTR in the first few cents of spend. You end up with a winner that wasn't really a winner, and no real learning to apply forward.

Here's how we structure our tests.

Inside each ad set, ads created through the matrix we mentioned earlier (placements × visual angles × messaging angles) give you something you can actually pull learnings from. Rather than testing one variable against another, you're loading the ad set with enough diversity that Meta can show you which combinations resonate with which people.

Each test should feed into a larger learning. We benchmark top spenders against the account average. We track which messaging angles unlock new audiences, which visual styles drive lower CPMs, which formats are converting versus just spending. Without documenting this, you're spending money to discover the same insights over and over (and over).

High Standards

So what should a DTC performance creative agency actually deliver?

• True creative diversity. Ads that vary across style, format, and message type. Not 30 versions of the same concept with different headlines.
• Volume matched to spend. Net new, Andromeda-optimized ads at the cadence your spend level demands, scaling as your budget grows without sacrificing strategy.
• Structured testing. A documented framework for loading ad sets with enough genuine diversity to generate clean learnings, and a system for compounding those insights into future creative decisions.

Within 90 days, you should have documented patterns about what works for your brand that guide future creative decisions and reduce reliance on guesswork.

None of this is revolutionary, but it is oddly absent from many agencies. The reason it feels rare is that most agencies have built their business around the volume of creative rather than the diversity of the ads (or more importantly, how the ads do).

So What?

Your creative agency is making decisions every week that directly impact your customer acquisition costs. CAC has been climbing across DTC for the past few years thanks to platform saturation, signal loss, and the Andromeda shift, and the margin for error has basically disappeared.

You can't afford to spend six months with an agency before discovering they've been inflating deliverable counts with meaningless variants. You can't build a scalable acquisition channel on guesswork disguised as structured testing.

Use this article to gut check against what you are being pitched, and what you are seeing in your accounts. If you'd like an audit, or to chat about anything DTC performance creative and marketing, let's connect.