Growth Marketing, Creative System and Optimization Role
creative is now an
engineering problem
Hey, I'm Danny . I spent years leading digital product for Beyoncé. I know what it's like to sit at the intersection of creative, data, and growth for an iconic brand that is well known and aiming for the next level of growth.
I quit that job to found a company backed by Floodgate, Snapchat, and other great investors. As CEO, I built and scaled a resale marketplace to millions of dollars in GMV through creator and content-driven playbooks.
I've spent the last 12 months building the exact type of growth systems this role calls for. I run them for a range of venture-backed startups, as well as my own projects through Kiko, an operating system for creator and content-led growth. I publish my growth philosophy and work on YouTube.
I've been building the operating system for the next era of growth
The next era of growth is creator-driven video. Most teams run creators, UGC, and paid as three separate channels with three separate owners. That's the old model. They're one operating system: one engine that finds what connects and amplifies it everywhere. This operating system helps feed Reddit, SEO, AEO, lifecycle, and in-product surfaces.
This comes to life through a growth roadmap and testing roadmap: executing a series of high-impact experiments over a period of time, searching for outlier channels and opportunities that can move the needle.
The systems I'm building on Codex let me do what I paid multiple full-time employees and teams to do at my last company.
Codex drives these loops for me across a range of go-to-market activities and tasks: agents optimizing paid spend across Meta, TikTok, and Google, workflows that produce creative ideas to drive content production, and automated product marketing videos based on GitHub PRs. These loops are driven by underlying systems of skills and data sources that get better over time. I've been the product manager, the engineer, and the marketer for these systems.
Here are a few production projects I've built and am running for venture-backed companies right now
Shared with permission
A creator go-to-market engine built to dominate a specific category
Reverse engineer each network with first- and third-party data to find winning content and creators. AI uses multimodal models to pre-sort fit and potential. They feed into a cold email outbound engine, which emails and negotiates with creators to capture rates, as well as agents leading the way on conversation management, contracts, and performance tracking. Under the hood it's a stack of agents and AI systems running the whole thing. None of this was possible 12 months ago.
Driving UGC campaigns with grounded research and analysis
Current winners on the networks are the source of inspiration that drives creative strategy for a UGC program. The system pulls a bunch of first-party data and runs it through multimodal analysis systems with my expertise and point of view baked in via dynamic skills. We combine this research with our own thoughts and ideas to drive winning creative briefs for UGC content creators. Every week, it helps figure out what each UGC creator in the program is going to post.
A creative engine for testing thousands of Meta ad creatives per month
To win on Meta, you have to scale creatives. To do that, you need an engine that can continually produce winning ideas. I built the full pipeline to do it: brand DNA research, angle and hook generation from rated research, script kits, AI video production, and an LLM judge that QAs every final for hallucinations, hook match, and overlay quality, driving automatic regeneration loops before anything reaches a human approval gate. Then it publishes to Meta.
How do you take Codex to 50 million active users?
I just want to share my thinking on a sample growth question here. Obviously I would love to get into the details more. This is really just about showing my high-level thinking around the growth roadmap and how to achieve this goal, sitting at the intersection of product and marketing that this role requires.
Codex just announced 8 million active users and the product is amazing. Maybe paid subs is the better goal, but since paid numbers aren't publicly talked about, I'll reference active users here. So how do you get to that next level of incremental growth? To me, the answer is in the jobs Codex solves outside the core developer use case.
There's a much broader market of professionals and students who already know ChatGPT but haven't tapped into the power of general-purpose agents like Codex: people with recurring reports, internal tools, spreadsheets, research, and workflows that an agent can just do for them. I've increasingly used Codex for personal tasks like finding and buying movie tickets to The Odyssey and summarizing my group chats into daily digests.
Some of the most interesting things in Codex are the ones almost nobody uses to the degree they could: automation, scheduled tasks, and Sites. An agent that runs your work on a schedule and reports back with what it learned. A shop owner whose agent watches competitor pricing overnight, a recruiter who wakes up to a sourced and ranked candidate list, a creator whose agent turns every video into a week of clips, a coach whose agent turns session notes into client plans.
High-level philosophy: we're building the compounding learning loop to achieve this goal
The use-case ontology
- Work backward from the customer's life. The problems we solve for them, the use cases we solve for them, and the context in which they use our products
- Research the signals. Organic content category performance across TikTok, YouTube, and Instagram, and the same for paid across those platforms. Combine that with SEO and answer engine optimization research, plus the highest-frequency and highest-value use cases
- Build the thesis from it. Here are the problems we solve, here is the type of audience each problem maps to, and here is when they actually have that problem. This thesis is what ties the whole system together: it flows downstream into the growth engine and the distribution engine, mapping directly to the type of content we want to make, the type of creators we want to work with, the type of messaging we want to run, and the language we use
The growth infrastructure
- Track every signal in and out of the system. Account and conversion data, pixels, first-party data, server-side tracking, organic performance across Instagram, TikTok, and YouTube, and all the performance data from the marketing platforms, pulled into one queryable place, a warehouse or data lake, and made useful, with incrementality holdouts so you know what actually moved the needle
- LLM judges, evals, and self-optimization at every step of the process. Not just pre-scoring creative before spend, but surfacing learnings and insights and proactively managing campaigns to success
- Move fast with a lot of stakeholders. Quickly building internal tools for people to review, give feedback, and collaborate, so the pipeline never stalls waiting on a thread
The distribution engine
- Creator-led growth puts humans at the center of the storytelling. We find the best creators in the world for each use case, from small micro creators to massive ones, across YouTube, TikTok, and Instagram. We brief, produce, iterate, and measure creator content as a performance lever, and build this into a repeatable distribution channel
- Scaling human-led creative: clipping plus UGC. Winning demonstrations, keynotes, and demos become clip libraries paid per verified view, and UGC is especially good for younger professional use cases. We're locking people in for the future as both the harness and the full product experience become sticky on top of the model
- Identifying scale winners through paid. Spark ads on organic winners, then AI variant generation across hooks, personas, markets, and languages, with fatigue detection triggering automatic refresh
Let's talk
I would love to interview with the team, share my work and learnings, and find out more about the role, what your team needs, and how I can add value.
I've been a user of OpenAI products since Davinci-003, and in that time I've used those products to build marketing and engineering systems. OpenAI is the most important company of our generation. It'd be an amazing full-circle moment to bring the expertise I've built using OpenAI's products home to help OpenAI itself.