RThe Flywheel Tracking mockups →  ·  as of Jul 24, 2026

One machine. Proof is the fuel.

Media puts me in front of Denton County → that attention becomes my own deals → those closings become proof that trains agents to sell more → their results become proof that recruits more agents. Each stage’s output is the evidence that powers the next, and the last one feeds the first.

The one number that rules everything Avg closings / month not yet measured · build the odometer first
working partial missing Each stage runs on a number that proves it works. Build what’s missing, in order.
1

Media

Denton County knows who you are. Attention is the raw input.

what’s in this bucket →
Proof metric Pieces published / month → leads attributed nothing published yet
Next lever

Publish. The drop zone, the machine and the concept queue all work — the blog is empty. Nothing downstream can be credited to media until pieces actually ship.

Jump in →
2

Sell

Your lab. Push your own monthly average up.

what’s in this bucket →
Proof metric Avg closings / month, trending can’t see it yet
  • Capri sales motion documented, 12 stages
  • Personal production view live, but snapshot-only + funnel-integrity bug
  • Proof ledger mints a packet per closing, daily 8am CT — but every story is still only “prompted”
  • Monthly outcome history (odometer) does not exist
  • Experiment log change → lift, untracked
Next lever

Build the odometer: a monthly outcome history + experiment log. Fix the held/signed integrity so the scoreboard is true.

Jump in →
3

Train

Turn what worked for you into agent behavior.

what’s in this bucket →
Proof metric Agent sales lift from the 50OS cohort can’t see it yet
  • The 50OS cycle 2026Q3 live — cohort enrolled, gates evaluating, app in daily use
  • Coach Cockpit daily clear-to-zero worklist — shipped RYR-198
  • Intake baselines captured for every enrolled agent — the “before” exists
  • The 50OS scoreboard no follow-up snapshot to diff the baseline against — lift is uncomputed, not unrecoverable
  • Self-report feeder the50os.self_reports has never received a row
  • Handbook delivery surface (hosts /50os/)
  • Work-product gateway per-agent boards
Next lever

Take the second measurement. The intake baseline exists for every enrolled agent, so lift is one snapshot and one diff away — nothing here is lost. Fix the empty self-report feeder while you are in there.

Jump in →
4

Recruit

Use proven agent results to bring on more.

what’s in this bucket →
Proof metric Recruits / month & their ramp not packaged yet
Next lever

Package proven agent lift into a recruiting proof asset the 50OS and the setters can put in front of recruits.

Jump in →
and back to stage 1 — every closing queues itself as content
Today’s one move

Build the odometer in Stage 2.

Everything downstream inherits from your own numbers. Until your monthly closings are trustworthy and trending, there is nothing real to teach and nothing real to recruit on. Start there: a monthly outcome history + experiment log, and fix the held/signed integrity.

Start Stage 2 →

This is the map, not the meter. As each stage gets instrumented, its proof number goes live here.