We re-earn the engagement every quarter with the same data we would use to end it. This document contains a pre-filled case for concluding — the argument for firing us, written by us, from our own instruments. It is not a formality and it is not shorter than the case for continuing. Read the decision below, then read page 2 before you make it.
THE DECISION ON THE TABLE — ONE OF THREE, SIGNED TODAY
CONTINUE ☐
Same scope, same fee, same cadence. Recommended only when the forward case is clean.
CONCLUDE ☐
Discharge protocol (Form 247-X) starts inside 60 seconds of this signature. No exit fee, no lite retainer.
CONTINUE — MODIFIED ☒
One quarter, reduced scope, fee stepped down $16.0K → $11.5K/mo, and a written numeric kill rule on page 3.
Q2 did not pay for us on the only basis that measures our work. Same-store incremental collections for the quarter were +$47K against $48K of ADMEN fees, and after netting the incremental spend the quarter is −$10.2K of contribution margin. The forward case is much stronger than the trailing one — June exited at +$45K/month and July is running at +$78K — but a forward case is a promise, and you have already bought eleven months of those. So we are not asking you to renew at the same price. The build is finished; what is left is assembly and judgment, and it does not cost $16.0K a month. We recommend one more quarter at $11.5K/mo against the kill rule on page 3, and we will write the conclude recommendation ourselves if it trips.
THE QUARTER — AGAINST BASELINE, AGAINST THE DIAGNOSIS MODEL
| MEASURE | DAY-0 BASELINE ×3 | Q2 '26 SAME-STORE | Δ SAME-STORE | Q2 '26 ALL-IN | Δ ALL-IN |
|---|---|---|---|---|---|
| New patients, quarter | 639 | 727 | ▲ 88 · +13.8% | 777 | ▲ 138 · +21.6% |
| New patients, exit month (Jun) | 213 | 249 | ▲ 36 · +16.9% | 267 | ▲ 54 · +25.4% |
| Collections, quarter · GL 4000 | $3.090M | $3.137M | ▲ $47K · +1.5% | $3.333M | ▲ $243K · +7.9% |
| Collections, exit month (Jun) | $1.030M | $1.075M | ▲ $45K · +4.4% | $1.147M | ▲ $117K · +11.4% |
| Cost per new patient, loaded | $246 | $219 | ▼ $27 · −11.0% | $253 | ▲ $7 · +2.8% |
| Fully-loaded growth spend, quarter | $164.7K | — | $204.0K | ▲ $39.3K · +23.9% | |
| ADMEN retainer, quarter · GL 6200 agency | $0 | — | $48.0K | — | |
THE ROW THAT MATTERS IS THE HIGHLIGHTED ONE. NEW PATIENTS ROSE 13.8% SAME-STORE AND COLLECTIONS ROSE 1.5%, BECAUSE NEW PATIENTS ARE A MINORITY OF AN ESTABLISHED GROUP'S PRODUCTION — THE SAME RELATIONSHIP THE DIAGNOSIS WARNED ABOUT AND THE BOARD PACK REPRINTS EVERY MONTH. ANY AGENCY SHOWING YOU A COLLECTIONS LINE THAT MOVED AS FAST AS ITS NEW-PATIENT LINE IS NOT MEASURING WHAT IT SAYS IT IS.
THE ROI MATH A CFO WILL RUN ANYWAY — SO WE RAN IT THREE WAYS, INCLUDING THE ONE THAT LOOKS BAD
① ATTRIBUTED — THE FLATTERING ONE
4.5× on our fee
The number most agencies would put on this page. It is not arithmetically wrong; it answers the wrong question. It counts every attributed patient as incremental, including the ones you would have got without us. Do not take this number to your board.
② ALL-IN INCREMENTAL
2.3× on our fee
Honest arithmetic, wrong denominator. $196K of the $243K is Parkway-11 — a location you financed, built, staffed and opened. We routed demand to it. We did not create it, and we will not bill the ramp of your own capital as our result.
③ SAME-STORE INCREMENTAL — THE ONE THAT JUDGES US
NEGATIVE — Q2 cost you money
Eleven locations, same instruments, same definitions, frozen at baseline. Against $48.0K of ADMEN fees the quarter returned $29.1K of contribution and consumed $39.3K of incremental spend. On the only view that isolates our work, Q2 did not clear. April is most of the damage; June is most of the case for staying.
This section is written before the case for continuing, it is longer than the case for continuing, and it is the section we read aloud first in the meeting. If you finish it unconvinced that we argued against ourselves in earnest, the document has failed and you should say so.
The cumulative arithmetic does not clear, and it is not close. Eleven months of engagement have produced +$7K of cumulative same-store incremental collections — about $4K of contribution margin — against $176K of ADMEN fees. Quarter by quarter: −$50K, −$10K, +$20K, +$47K. The line is rising and the integral is still roughly zero. A CFO looking only at money in and money out, on the view that isolates marketing's work from your own capital, is looking at eleven months of spend and no recovered margin. That is a legitimate reason to stop, and no amount of trajectory language makes it not one.
The gain is installed, not rented — you can keep most of it without paying us. Attribution runs on 10 of 12 channels, in accounts held in your name with your admin credentials. The intake rubric, per-desk call logging and the weekly desk huddle (Form 247-K) are already run by your own desk managers, not by us. Overflow and after-hours routing are configured in your phone system. The reminder cadence is live in your PMS. The build is done. What you buy from us now is monthly assembly and judgment — and the assembly is automated. A reasonable person could conclude that you are twelve months past the point where the money bought a capability and are now renting a report.
The binding constraint has moved somewhere we cannot reach. Our own month-11 board pack says it plainly: phone coverage at one pod and chair capacity at two others now cap collections before marketing does. Riverside's permanent desk hire has been open 47 days. North-04 and Central-01 are at 92–94% chair utilisation and need a hygienist and a Saturday, not a media plan. Those are a hire, a schedule and a lease. We do none of the three. Paying a marketing retainer to move a staffing constraint is the single most common way agency spend outlives its usefulness, and we would rather name it than bill through it.
The second-order costs are real, still moving against you, and we do not know where they stop. Since Day 0: case acceptance 43% → 40%, FFS/cash payer mix 36% → 34%, contractual write-downs 13.4% → 14.4%, 90-day production per new patient $612 → $588, payback 2.0 → 2.1 months. Four of those five worsened again in July. We forecast this trade in the diagnosis before we bought a single patient, which makes it honest — it does not make it free. Each additional insured, unestablished patient contributes less than the last, and at some volume the arithmetic inverts. We cannot tell you where that point is, because we have never measured one. If you believe you are near it, concluding is the correct call.
Every headline we send you is flattered by a location you built. All-in new patients are +21.6% and same-store are +13.8%. All-in collections are +7.9% and same-store are +1.5%. All-in incremental contribution is +$111.4K and same-store is −$10.2K. In every pair, the bigger number contains Parkway-11 — your capital, your lease, your staff, your risk. We print both and we tell you which one measures us, but a board reading the all-in column would fund another year of this on the strength of your own construction schedule. If that is the number persuading anyone in this room, conclude.
What concluding actually costs you, stated without softening so this section isn't a bluff: you lose monthly assembly and the judgment layer, you lose the demand/capacity refresh (Form 118-M) that steers spend away from full chairs, and you lose the instrumented record that a banker process would otherwise start from. You keep everything else — the baseline reading, the diagnosis, the instrumentation, the rubric, the frozen definitions, every readout and pack we have ever sent. That was the deal on day one and it does not change on the way out. If you conclude today, the discharge protocol starts before you leave the room, and the last invoice is the one already issued.
The exit rate, not the quarter average. Q2's +$47K is the integral of a rising line: April −$11K, May +$13K, June +$45K, July preliminary +$78K. Twelve months at June's rate is $540K of same-store collections and $335K of contribution — 1.7× the current fee, 2.4× the reduced fee we are recommending. A quarter average is a lagging read of an accelerating series, and it is the wrong instrument to end an engagement with.
Two of the five diagnosed changes are unfinished. Change 4 (reallocate spend by loaded cost per new patient) has landed $219 against a $210 target and the untraceable bottom third is not fully zeroed. Change 5 (broken-appointment repair) has only just earned its clean quarter — 22.4% → 19.4% with definitions now stable — and its movement has never been priced. The diagnosis carried $34K/yr on change 5 alone. Concluding now banks the expensive changes and abandons the cheap ones.
Handoff has a protocol and it has not been run. Discharge (Form 247-X) is a 90-minute transfer with an ownership matrix, and the desk system has been client-run for exactly one quarter. CLAIM WE CANNOT EVIDENCE: we believe the gain decays if the cadence lapses before handoff. We have no decay curve. We have never measured one. Treat this as an assertion, not a finding, and weight it accordingly.
The banker clock. Your stated intent at intake was a process with a banker in roughly two years. The asset in that room is an instrumented history a diligence team can re-run — cohorts, attribution, capacity, definitions frozen at Day 0 — and it compounds only if it keeps being written. CLAIM WE CANNOT EVIDENCE: we cannot show you that a longer instrumented record changes a multiple. We can show you that its absence costs weeks of diligence. Those are different claims and we are only making the second one.
Note what is not in this column: no relationship argument, no sunk-cost argument, no "the market is competitive" argument, and no claim that stopping is risky. If the case for continuing needed any of those, it would not be a case.
| DIAGNOSED CHANGE (FORM 247-D) | TARGET, AS WRITTEN | Q2 '26 ACTUAL · n | VARIANCE | VERDICT |
|---|---|---|---|---|
| № 1 · Instrument every desk on the intake rubric; readout reviewed with the people who took the calls | booking 35% → 43%+ within 90 days | 39% · n = 1,842 scored NP calls (±1.1 pt) | ▼ 4 pt — and 11 months, not 90 days | PARTIAL |
| № 2 · Overflow + after-hours routing; cover the two known lunch gaps | answer rate 78% → 90%+ | 86.7% quarter · 88% exit · n = 11,065 calls | ▼ 3.3 pt quarter · ▼ 2 pt exit | PARTIAL |
| № 3 · Attribution build-out, click-to-chair | 2 → 10+ channels traceable | 10 of 12 channels · census, not sample | on target | MET |
| № 4 · Reallocate spend by loaded cost/NP; zero-spend the untraceable bottom third | loaded cost/NP $246 → ~$210 at flat volume | $219 same-store · n = 727 NP | ▲ $9 — bottom third not fully zeroed | PARTIAL |
| № 5 · Broken-appointment repair: one status definition, then reminder pilot | honest read first; movement priced after one clean quarter | 22.4% → 19.4%; definitions stable 2 quarters | read achieved · movement never priced | MET (read) · UNPRICED (movement) |
| MODELLED OUTCOME — CHANGES 1 + 2 COMBINED | ~50 incremental NP/mo · ~$360K/yr collections | +36 NP/mo at exit · $47K in-quarter · $540K/yr at the June run-rate | NP under model · collections under in-quarter, over at exit | SPLIT |
THREE PARTIALS, ONE UNPRICED, TWO MET. THE PARTIAL ON № 1 IS THE ONE THAT MATTERS: THE DIAGNOSIS PROMISED THE BOOKING-RATE MOVE IN 90 DAYS AND IT IS NOT DONE IN 330. THE CAUSE IS NOT THE RUBRIC — IT IS THAT ONE DESK STILL CANNOT STAFF ITS LUNCH WINDOW, WHICH IS ARGUMENT C3 ON SHEET 2 AND NOT A DEFENCE.
THE KILL RULE — NUMERIC, DATED, AND WE WILL BE THE ONES TO INVOKE IT
If same-store incremental collections do not average at least $60K/month across Q3 '26 (Jul–Sep), we recommend CONCLUDE at the October 15 review. Not "re-scope", not "re-set expectations" — conclude, with the discharge protocol scheduled in the same meeting. July is already at $78K, so this rule is not a low bar dressed up as a high one; it is the June exit rate plus a third, held for a quarter.
Two secondary triggers, either of which alone forces the conversation: (a) Riverside's permanent desk hire still open at the Sep 30 close — because if the constraint cannot be staffed, we are being paid to work around it; (b) 90-day production per new patient below $560 — because that is where the marginal patient's contribution stops covering the marginal acquisition cost on the current mix.
We will write that recommendation ourselves, in this format, and read it aloud first — the same way we read the case for concluding first today.
THE MEETING — 30 MINUTES · OWNER + CFO + DIAGNOSING PARTNER · PRACTICE ADMINISTRATOR OPTIONAL
No slides. This document is the meeting (RULE 7). If a recommitment needed a deck, it was not a decision — it was a pitch. Attendance is three people because a four-person version of this conversation becomes a presentation. The signed sheet goes to the board chair the same day; the board does not attend, because we are not going to argue for our own renewal in front of the people who approve it.
— specimen —
DIAGNOSING PARTNER, ADMEN · RECOMMENDATION AS WRITTEN
— specimen —
CLIENT CFO · ROI BASIS AGREED · Q3 KILL RULE ACCEPTED
— specimen —
OWNER / CEO · DECISION: CONTINUE / CONCLUDE / CONTINUE-MODIFIED · DATE