How Disorder Media scaled creative production and cut cost per purchase in half with SecretSauce

Amit Waissman is a campaign manager at Disorder Media, a performance marketing agency running paid advertising for major international brands. Amit manages over 15 client accounts across e-commerce, SaaS, real estate, tourism, and more - building and optimizing every channel, campaign, and ad set to hit their growth targets.
The problem is that every campaign needs new creatives, but for most of his clients creative production is the slowest part of the operation. Learn how Amit uses SecretSauce to quickly and easily produce ad creatives for his clients, and how those creatives cut his clients’ cost per purchase by 50%.
The creative bottleneck holding back performance
Meta’s recent shift toward creative volume and broad targeting changed the game for performance marketers. The platforms now reward volume and variety, which means agencies need more creatives, more angles, and more iterations.
But many of Amit's clients simply don't produce enough creatives to keep up: "We have a lot of clients that just don't make enough creatives, so we're basically stuck with what they have, trying to squeeze as much out of the existing ones as we can."
However, when a creative fatigues and there's nothing to replace it with, costs climb and performance drops. For Amit, getting new creatives meant picking between slow, expensive, or inconsistent:
- The client's team: Most clients are slow to turn things around. "I see a creative working, but I know asking the client to produce more will take like a month."
- Freelancers: Hiring outside help was expensive and hard to justify for quick tests. A typical round of 5 to 6 static images would run around $1,000.
- Other AI tools: Amit tried ChatGPT, other image generators, and various LLMs, but the workflow was fragmented and the output was inconsistent. "You need to work with two different LLMs to get what you want. You need to go to ChatGPT, write a script, then paste it into another tool to generate the creative. And even then, the output wasn't good enough to run most times."

