Named growth plays
Launch, activation, consensus, retention, pricing. Each has a trigger, a success metric, and a review window.
journoa 
A WORKING PROJECT / BUILD LOG + WAITLIST
An intelligence layer for the B2B funnel.
Journoa reads the data already sitting in your tools, works out which funnel stage is leaking, and generates the play that addresses it. I am building it in the open, and the engine will stay open to use.
ONE EMAIL FIELD. BUILD NOTES FIRST. NO SPAM, NO FAKE URGENCY.
VISIT
LEAD
SIGNUP
ACTIVATION
DEAL
RENEWAL
THE B2B FUNNEL, FIRST VISIT TO RENEWAL. THE PULSING DOT IS WHERE MOST OF THEM LEAK.
Funnel signals
Mapped by data maturity, so a new startup is never judged like a five year old SaaS.
Golden cases
Tested against skeptics, opportunists, and awkward edge cases before it touches your data.
Engine stages
Every recommendation carries a trace. You can see why it said what it said.
BUILT FROM PUBLIC RESEARCH, CASE EVIDENCE, AND INTERNAL PRESSURE TESTS. CUSTOMER OUTCOME CLAIMS WILL BE PUBLISHED ONLY AFTER REAL BETA RESULTS.
HOW IT WORKS
Signals in. Assessment. One recommended play out.
Read the data
Sources are tagged by quality: SDK, CRM, CSV, manual, or inferred. Weak evidence is weighted as weak evidence.
Evaluate the funnel
60 signals become condition checks. Results can be true, false, or unknown. Under 30 samples is marked directional.
Generate the play
The picker ranks 33 plays by evidence and interaction rules. Every play has a trigger, data minimum, success metric, and review date. You approve before send.
WORKS FROM Your formsCSVManual notesProduct eventsHubSpot / plannedPostHog / plannedSlack / plannedResend / planned
Start with nothing connected. Planned connectors are planned, not faked. Journoa never buys traffic or spends your money.
WHAT YOU GET
Every recommendation carries its trace.
The engine moves through 7 stages: classify, signals, conditions, assessments, growth map, pick, render. Each stage writes a trace entry, so the recommendation can be checked rather than trusted blindly.
EXAMPLE DIAGNOSIS. NOT CUSTOMER DATA.
SIGNAL
62 signups reached account creation.
EVIDENCE
14 reached first value.
ASSESSMENT
Setup friction, not weak demand.
NEXT PLAY
Shrink the first step, offer stuck accounts a hand.
CONFIDENCE
Directional, sample still thin.
Missing evidence evaluates to unknown, not false. The engine states what would make it knowable and the cheapest way to learn it.
THE ENGINE KEEPS LEARNING
Seeded by research. Tuned by measured outcomes.
The catalog starts from published research and 13 benchmark rows with source tiers. Outcomes are recorded against preset metrics. Consented records with identifying details stripped tune win rate estimates. New research keeps entering the pack.
The engine stays open to use. It is a project, not a checkout funnel. It reads the minimum data, shows what it used, and deletes on one call.
Research seeding: published research and 13 benchmark rows with source tiers.
Outcome recording: each play is measured against a metric chosen in advance.
Win rate tuning: consented records with identifying details stripped tune future recommendations.
Open to use: the project is not a checkout funnel.
Scope, stated plainly.
WHAT IT IS
An intelligence layer beside HubSpot, PostHog, spreadsheets, and product events.
A deterministic staged pipeline. AI drafts, but it does not decide.
Generated plays approved before anything sends.
Explicit confidence and missing inputs.
WHAT IT REFUSES TO BE
xA chatbot.
xA dashboard.
xAn autopilot spending money.
xA product being sold. Nothing to buy yet.
BUILD PROGRESS, WITHOUT THE THEATRE
Where the engine is now.
LAST UPDATED JULY 14, 2026
DONE
Research corpus, 31 assets
DONE
Methodology pack, 33 plays and 60 signals
DONE
Pack validation and learning overlay
CURRENT
Core pipeline through recommendations
NEXT
Growth map and plain language output
LATER
Golden case review, then founder beta
94 test cases run across four packages. Progress is stages, not a number.
A NOTE FROM THE BUILDER
Journoa is my attempt to encode how a careful growth marketer actually thinks in a B2B context.
I used AI coding tools like Claude Code and Codex to write the code. Scientific research, frozen specs, typed contracts, and tests decide what gets called done.
- Jai, building in the open. Build notes will be sent to the waitlist first :)