Prepared 2026-07-22. Figures verified as of this date unless marked otherwise.
Machine-readable version: data/kpi_scorecard_template.csv (feeds the workbook's KPI Dashboard tab). All target ranges are tunable assumptions — recalibrate at month 3 with real data. Vanity metrics (follower counts, impressions, likes) are deliberately excluded.
Design rule: leading indicators are managed weekly; lagging indicators are read monthly. A red leading indicator triggers a behavior change now; a red lagging indicator in months 1–4 mostly means "the lag hasn't elapsed yet" — check the leading side before panicking.
| # | KPI | Definition | Source of truth | Target (assumption) | Green | Yellow | Red |
|---|---|---|---|---|---|---|---|
| L1 | Product drills completed | Structured scenario drills (T3 template) with a scored debrief | Drill log in repo | 5/wk | ≥5 | 3–4 | ≤2 |
| L2 | Personalized agent briefs created | New one-page research briefs on pilot-list agents (T2) | Briefs folder | 3/wk until list covered, then refresh | ≥3 | 1–2 | 0 |
| L3 | Meaningful agent conversations | Two-way, business-relevant exchange with a next step (call, in-person, or substantive email thread — not voicemails) | CRM/contact log | 5/wk | ≥5 | 3–4 | ≤2 |
| L4 | Agent meetings | Scheduled sit-downs, coffee, or office visits | Calendar | 2/wk | ≥2 | 1 | 0 |
| L5 | Scenario requests received | An agent sends a real deal scenario for a Desk brief | Scenario log | month 1: 3–4 total (≈1/wk from wk 2); mo 3+: 1/wk sustained | mo 1: ≥3 cum; mo 3+: ≥1/wk | mo 1: 2 cum; mo 3+: 4-wk avg ≥0.5 | mo 1: ≤1 cum; mo 3+: 4-wk avg <0.5 |
| L6 | Consumer game-plan calls | Structured first-time-buyer planning calls held | Calendar + CRM | 2/wk (mo 4+) | ≥2 | 1 | 0 |
| L7 | Applications started | New applications in the LOS | LOS | per scenario model | ≥ base-month plan | ≥ cons-month plan | below cons |
| L8 | Preapprovals issued | Preapprovals delivered to borrower/agent | LOS | per scenario model | ≥ base | ≥ cons | below cons |
| L9 | Scenario response time | Ack within 1 business hour; written, guideline-checked brief within 4 business hours (Mon–Fri) — the published v0 SLA (file 09 §2.5) | Scenario log timestamps | 100% SLA met | 100% | 1 miss/wk | ≥2 misses/wk |
| L10 | Follow-up completion | Scheduled follow-ups executed on the scheduled day | CRM task list | ≥90% | ≥90% | 75–89% | <75% |
| L11 | Workshop registrations | First-Home Lab workshop signups | Registration list | 8/event (mo 4+) | ≥8 | 4–7 | <4 |
| L12 | Referral introductions | A person (agent or past contact) proactively introduces Mitchell to someone new | CRM, tagged | 1/wk (mo 4+) | ≥1 | 4-wk avg ≥0.5 | 4-wk avg <0.5 |
| L13 | Overdue next-actions | Pipeline/relationship next-actions past their date at Friday review | CRM task list | 0 | 0 | 1–3 | ≥4 |
| L14 | Energy at close (red-team M9) | Self-rated 1–10 at each daily debrief; weekly average logged | Daily debrief (file 14 template) | ≥6 avg | ≥6 | 5 | <5 |
L3/L5 note: weekly targets assume the ramp — weeks 1–2 (setup + half-dose waves) expect L3 of 1–3/wk, not 5; grade weeks 1–2 against the file-12 ramp, not the steady-state target. L14 is a first-class input to the Version-B (35–40 hr) calendar switch and to rule 10 — a red L14 with green activity means the dose is wrong, not the person.
| # | KPI | Definition | Source of truth | Target (assumption) | Green | Yellow | Red |
|---|---|---|---|---|---|---|---|
| G1 | Purchase contracts | Preapproved borrowers going under contract | LOS/agent confirmations | per scenario model | ≥ base | ≥ cons | below cons |
| G2 | Funded loans | Loans funded in the month | LOS/funding reports | per re-based scenario ramp (file 22 §6: cons 1–3 / base 5–6 / agg 14 yr-1) | ≥ base | ≥ cons | below cons |
| G3 | Volume ($) | Sum of funded loan amounts | LOS | funded × AVG_LOAN_AMOUNT |
≥ base | ≥ cons | below cons |
| G4 | Revenue ($) | volume × [BPS_COMP]/10,000 |
Comp statements (source of record — not the model) | formula | ≥ base | ≥ cons | below cons |
| G5 | Pull-through | Funded ÷ applications (rolling 90-day) | LOS | ~27% base assumption | ≥27% | 20–26% | <20% |
| G6 | Repeat/referral share | % of new borrowers from referral or repeat sources | CRM source tags | rises over time; no yr-1 hard target | trending up | flat | falling |
| G7 | Cost per application | Month's cash expenses ÷ applications | Expense log ÷ LOS | informational yr 1 | — | — | — |
| G8 | Cost per funded loan | Month's cash expenses ÷ funded loans | Expense log ÷ LOS | informational yr 1 | — | — | — |
Source-of-truth rule: the LOS and comp statements are the record for anything financial. The scorecard copies from them, never the reverse, and AI never generates a number into this scorecard — it only summarizes numbers Mitchell entered.
These convert red streaks into decisions instead of anxiety. "Weeks" = consecutive.
When: Friday, 4:30–5:00 pm, 30 minutes, non-negotiable calendar block. (If Friday collapses, Saturday 9:00 am — never skipped, never >15 min late into the following week.)
Process (30 min): 1. (5 min) Fill the scorecard row for the week from CRM, calendar, logs, LOS. Actuals only — no vibes. 2. (10 min) Run the T7 prompt (file 19) with the anonymized numbers: what moved, what's red, which stop/continue rules triggered. 3. (10 min) Answer four questions in writing, one line each: - What actually created this week's best conversation or scenario request? - Which red is a behavior problem vs. a target-calibration problem? - What am I avoiding? (Usually the answer is a specific call.) - What are the top 3 outcomes for next week, and which calendar blocks own them? 4. (5 min) Update next-actions in the CRM so L13 = 0 at review close; queue Monday's sprint list.
Monthly (last Friday): extend to 60 min, read the lagging table, update the workbook actuals, and run the file-22 recalibration check.