Most operators run ad ops as an output volume game. Make 50 ads, hope something hits. Prompt Claude, hand a spec to the editor, ship whatever comes back. No hypothesis. No tracking. No learnings. When something wins they don't know why. When something loses they don't know why. The whole thing is random output from a random process.
Real operators run it as a system. Research → Strategy → Brief → Creative → Launch → Read → Iterate → Repeat. Every ad has a hypothesis going in. Every test produces data coming out. Every loss becomes the input for the next win.
This is the thinking layer. The reason some operators hit on accident and others hit on purpose.
What this module covers
Two sub-pages, each a deep layer of the thinking system.
- Customer Research — the input. The post-purchase survey that's the master research artifact (11 questions + extraction notes + clustering framework), the Bloom origin story, my own posture-corrector → women's sports bra arc (and the $50k operations collapse that came with it), the emotional valence framework, plus the multi-source research stack (ZigPoll + Reddit + Facebook groups + AI voice agents + comments scraping).
- The Brief & Concept Tracking System — the workflow. Growth Guide for concept tracking, Persona × Angle × Awareness as the three testable variables, hypothesis-driven testing, ideation vs iteration vs imitation ratios, Milanote brief structure for both static and video. Important note: I've scratched this workflow for a cleaner Claude-Notion connector + sheet system that's masterclass-only. The principles still apply.
Reading order (insane-first): Customer Research → Brief & Concept Tracking. Research is the bigger mind-shift for most operators — the brief system is the workflow that operationalizes it.
How this module fits the rest
- Module 3 (Using Claude to Automate Your Entire Stack) is the production layer — Catalyst, Native, Videos, Advertorials. The actual ads. Module 5 is what makes those ads intentional instead of random. Without Module 5, Module 3's production speed becomes a way to ship 50 bad ads faster.
- Module 4 (Additional Revenue Sources) is the channel + measurement + retention layer — reading metrics, multi-channel attribution, ad account bid strategies. Module 5 is what produces the ads that those campaigns scale. Module 4 measures, Module 5 creates the inputs being measured.
- Module 1 (Affiliate & UGC Community Playbook) is the creator network that feeds production. Module 5 is what tells your creators what to make. Without intent, creator content becomes random output from a different supply source.
The single line for the entire module:
If you have intent and thought behind everything you're doing, even when you have a loser, you're winning.
Losing ads with intent become data that improves the next ad. Losing ads without intent become wasted spend with no permanent learning. The operating IQ difference between these two modes compounds every month.
Why most operators skip this
Three reasons:
- It's slower in the short term. Making 5 ads with a real research brief takes longer than making 50 ads without. It's only faster in the long term when those 5 ads produce a winner you can scale and the system tells you exactly why it won.
- It requires reading. Survey responses, Reddit threads, comment sections, your own customer support inbox. AI can outline but can't replace the read-through. Most operators won't sit down for the deep dive.
- It produces no immediate dopamine. A research session doesn't feel like progress the way uploading 50 ads does. The dopamine comes weeks later when your hit rate on iterations is suddenly 3x what it used to be — by which time most operators have moved on.
The operators who do sit down for the deep dive are the ones running portfolios at scale. Every brand at 8-figures has someone (founder or hire) who does the research work that's described in the Customer Research sub-page.
The Brandon-rule on AI in this module
Claude can outline. Claude can structure. Claude can summarize survey responses, write briefs from research, generate hooks at scale. Claude cannot do the deep trenching work — reading the actual responses, lurking in the Facebook groups, sitting with the AI voice agent transcripts, scrolling your own DMs.
The leverage isn't in the prompts. The leverage is in the quality of the input you bring to the prompts.
If every operator has Claude, the only variable left is the research depth going in. That's the moat. This module is how you build it.
Want the full, video version? Every SOP, notes, extra information, plus calls & a chat with me inside our 1,000+ member community — here inside the RunItUp Masterclass → Join here
Join now to lock in your rate, prices are increasing quickly.