How we stand up a demand-testing funnel — as a system
A reusable machine that turns a creator's audience into scored, tagged, personalized leads and tests real demand before we build a product. Built once by hand for Ruben; now a system any client can run in days. This walks through what it covers and how it actually works — so you can tell me where to make it better.
Status: template built · validating against Ruben's live funnelFor: the Studios teamAsk: where would you improve it?
↳ Leave feedback right on the doc
Hit the 💬 on any section, type your thought, then Copy all feedback (bottom-right) and send it to Linart. Point at sections by their § number. This whole thing is here to be improved.
§1
What it is, and what it covers
One system, one job: take a stranger in a creator's audience and turn them into a lead we actually understand — then use that to test whether a paid product has real demand, before we build it.
✓ What the system covers
The opt-in page that captures interest on a specific concept
Email capture into Kit (create + tag), fully wired
The segmentation survey that scores each person into a persona
Writing answers back to Kit as tags + custom fields
Personalizing the offer/page per persona (headline swaps, matched offers)
The confirmation email (generic, or per-persona when personas exist)
A reusable code template so the next client is days, not weeks
— What it deliberately doesn't cover
The go/no-go call on a concept — that's Dan's demand-testing playbook
Deciding which client or concept to run — that's a team decision
The persona definitions themselves — designed fresh per client, no template
Sales-to-close after a hot lead raises their hand
Revenue attribution (Stripe→Kit) — a known gap, tracked separately
The system is the machine. The strategy and the taste stay human.
§2
Watch a lead flow through it
Once a funnel is live, this is what happens every time someone raises their hand — with no one on our team touching it. Hit play and follow one person through.
Runtime · one lead, start to finish
ready Press play to send one person through the live funnel.
The point: from "stranger" to "scored, tagged, personalized lead," a human on our team touches it zero times. That's the automation. Our people show up earlier — designing the survey and the offer — and later, on the leads worth a real conversation.
§3
Who does what
The whole design is one line: people do judgment, the machine does the repetition. Here's the split.
The team (human)
direction · taste · decisions
Which concept we test and the promise it makes
The persona design — who we're really talking to (fresh per client)
Every taste pass on copy, design, and voice
The go/no-go call when the demand data lands
Approving every publish, deploy, and send
The leads worth a real human conversation
The system (AI)
execution · assembly · capture
Builds, wires, and deploys every page
Sets up Kit tags, fields, and the capture plumbing
Generates the survey build spec and the answer→tag mapping
Drafts every first version — survey, emails, headline variants
Runs QA against every known failure mode
Scores, tags, and personalizes every lead, live, 24/7
§4
How we build it for a new client
Standing up a client's funnel runs through these phases. Click any one to see what happens and who owns it. The orange dots are human gates; the teal dots the system runs.
§5
What to expect
~60%
of every build is reusable boilerplate — only integrations + design change per client
days
to stand up a new client's funnel once the template is proven — not weeks
0
humans in the loop per lead at runtime — the team's time goes to design + decisions
A new client engagement, start to finish
You give the concept + audience source + brand. The system mines the audience language and proposes personas; a human designs the final ones. The system builds the survey spec, the pages, the Kit setup, and drafts the emails — pausing for a taste pass and an explicit go before anything publishes. Then it runs, scoring and personalizing every lead, while the team watches the demand signal and makes the go/no-go call. Every learning gets captured so the next client is faster.
§6
Where it is right now — and what I want from you
Current state
The reusable template is built (isolated sandbox, nothing live touched)
We're validating it by rebuilding Ruben's funnel from requirements alone, then auditing it against his live funnel
Persona derivation for Ruben is the next step
What would make this better?
Tell me where you'd push it:
Anything the system shouldn't own — where does human judgment need to enter that it doesn't?
Anything missing from the scope in §1?
Where would this break with a client that isn't Ruben?
How to send feedback: use the 💬 buttons to comment on any section, then hit Copy all feedback (bottom-right) and drop it to Linart in Slack. Every note shapes the next version.