Desk ↔ Media Loop

SEP '26 CYCLE · GROUP · 12 LOCATIONS · LOOP HELD TUE SEP 1, 8:00A · 45 MIN · ADMEN MEDIA LEAD + CLIENT OPS LEAD · PARTNER ON EXCEPTION

CONFIDENTIAL · RX 03FORM 247-L

What the desk hears changes what the ads say — on a schedule, with a ship date and a measured read. A theme that gets collected and never reaches media is a leak, not an insight. Every entry below is a paraphrased composite of what callers ask — never a caller's words, never a name.

CADENCE — THE HUDDLE IS WEEKLY AND IT IS YOURS. THE MEDIA LOOP IS MONTHLY AND IT IS OURS. ADMEN DOES NOT SIT IN ON 52 HUDDLES A MONTH; WE SUPPLY THE INSTRUMENT AND YOUR DESK MANAGER RUNS IT. FULL ROUTING, P.3.

SAMPLE FRAME — AUGUST INTAKE POOL · WHAT THE LEDGER IS BUILT FROM, AND WHAT IT ISN'T

NP-OPPORTUNITY CALLS LOGGED

728

12 desks · 14.0 per desk-week

SCORED TO RUBRIC

513

70% of logged · 10 desks · 51 avg

UNANSWERED — NO RECORDING

73

10% · = the 90% answer rate on 247-R

RECORDING FAILED

31

4% · under 20 sec or capture error

LOGGED, NEVER SCORED

111

South 08 + Parkway 11 ran no huddle — the two desks we route the most demand to. See p.3.

INTAKE → ACTION LEDGER · WHAT WAS HEARD, WHAT CHANGED, WHAT IT DID · ALL THEMES PARAPHRASED COMPOSITES — NEVER A TRANSCRIPT QUOTE
ITEM · CYCLE STATUS	THEME HEARD AT THE DESK	n / POOL DESKS	WHAT CHANGED IN MEDIA / SITE	SHIPPED	MEASURED EFFECT
L-06-05 JUN CLOSED	Caller asks whether the doctor they read about is the one who will actually do the work, and can't tell from the page who practises where.	50 / 494 10% 8 desks	Named provider + photo above the fold on all 12 location pages; provider name into paid-search headlines, per market. No spend change.	JUL 7	Location-page consult requests 2.60% → 2.90% of sessions (201/7,730 → 231/7,965), 35 days vs matched prior.
L-07-11 JUL CLOSED	Caller asks the fee, and the question underneath it is whether it can be paid over time. The fee answer is accurate and the call ends there anyway. Arrived as the 247-K TO MEDIA line, WK 29.	105 / 502 21% 9 desks	Financing block above the fee block on 12 location pages; financing headline + sitelink in paid search, all markets; the desk's price answer now travels with a financing line and a date.	AUG 11	Consult requests 2.90% → 3.60% (135/4,660 → 170/4,720), 20 days vs matched prior. +33 requests on flat traffic. Traced in full, p.2.
L-08-03 AUG IN FLIGHT	Caller asks whether their plan is taken; the desk isn't certain, offers a call-back, and the call-back loses the caller. Some are plans the group accepts. (247-K ②)	68 / 513 13% 8 desks	Accepted-plan list published per location page + a plan field on the consult form, so the desk has the answer before the call. Same list on one card at the desk — an ops fix media only publishes.	SEP 15 IN BUILD	— READ AT THE OCT 6 LOOP · 30-DAY MINIMUM
L-08-07 JUL IN FLIGHT	Callers to Cedar Park 10 can't find the entrance and arrive late, or call a second time from the car park. The facade has been under scaffolding since spring; the desk says it eats the first five minutes of a first visit.	24 / 41 59% of Cedar Pk 10	Wayfinding block — "rear lot, entrance B, scaffolding on the front until October" — on the Cedar Park 10 page, and the same line in the confirmation SMS. No arrival photo: the facade can't be shot until the scaffolding drops (302-S). Gate B — one page, no spend.	AUG 5	First-visit late arrivals: 11 of 53 in the 90 days before (21%) → 2 of 16 in the 27 days since. n = 16 — too small to state a rate, so we don't. A 3-chair pod can't make an arrival sample in one cycle. Held to Nov 3, when the scaffolding drops.
L-05-02 MAY ROLLED BACK	Caller asks for a Saturday or an after-six appointment because they can't leave work, and takes a weekday slot reluctantly or not at all.	42 / 459 9% 6 desks	"Saturday appointments" headline + extended-hours copy on 6 location pages and in paid search. Reverted Jul 14 — it now runs only at the 2 locations that actually have Saturday chairs.	JUN 9 BACK JUL 14	61 Saturday-preferred bookings: 23 filled, 38 re-slotted to a weekday, 12 of those never came and never rebooked (32%). We advertised capacity the schedule didn't have.
L-06-09 JUN REJECTED	Caller opens by asking for the cheapest way to do it, and compares the fee against a number quoted elsewhere before any clinical question is asked.	38 / 494 8% 7 desks	Nothing. A "starting at" price offer was requested at the loop and declined by the partner. It cleared Gate A — barely, 38 against a threshold of 25 — and at $267 loaded cost per new patient (610-M p.3) paid search has no acquisition problem this offer solves.	—	Case acceptance is 40% against a 43% Day-0 reading (247-B / 247-R WK 29), and the fall traces to new-patient mix. A price offer buys more of exactly that mix. Revisit only if acceptance holds above 43% two months running.
L-09-01 AUG HELD	Caller asks whether a provider speaks their first language. Heard at one desk, three times, in one month.	3 / 513 <1% 1 desk	Nothing yet. Below Gate A (n ≥ 25 and ≥ 2 desks) and below Gate B (n ≥ 15 at one desk). It stays on the sheet, dated and uncounted, until it clears one.	—	Three calls is not a trend. We do not move media on n = 3, and we do not coach a human being on n = 3 either.

STATUS KEY — CLOSED: SHIPPED, MEASURED, EFFECT STATED · IN FLIGHT: APPROVED, NOT YET READABLE · ROLLED BACK: SHIPPED, MEASURED, MADE THINGS WORSE, REVERTED WITH A DATE · REJECTED: CLEARED THE GATE, DELIBERATELY NOT ACTED ON, REASON ON THE RECORD · HELD: BELOW THE N-GATE. THE n COLUMN DOES NOT SUM TO THE POOL — ONE CALL ROUTINELY CARRIES TWO THEMES. NO PUBLISHED BENCHMARK EXISTS FOR ANY RATE ON THIS SHEET; EVERY COMPARISON IS TO THIS GROUP'S OWN MATCHED PRIOR WINDOW.

One Theme, Traced — and What It Takes to Move Media at All SEP '26 · GROUP · 12 LOCATIONS · FORM 247-L · PAGE 2 OF 3 · SPECIMEN

THE CIRCUIT, TRACED — ONE THEME FROM THE PHONE TO THE AD AND BACK · ITEM L-07-11 an open loop is a collection habit; this one closes

① HEARD

THU JUL 16 · 8:15A

Riverside 09 weekly huddle, run by your desk manager. 6 of 15 scored calls carried a price question. The TO MEDIA line on FORM 247-K: callers ask about financing before price.

② POOLED

JUL 1—31 CYCLE

Same theme, 105 of 502 scored calls (21%), at 9 of 10 scoring desks. Clears Gate A. One desk is an anecdote; nine desks is an instruction.

③ DECIDED

TUE AUG 4 · 45 MIN

Monthly ADMEN media loop. Approved, owner named, ship date set. Estimated cost to ship: 3.0 hrs. No spend increase requested.

④ SHIPPED

TUE AUG 11 · 7 DAYS

Three surfaces: financing block above the fee block on 12 location pages · financing headline + sitelink in paid search, all markets · the desk's price answer travels with the financing line and a date.

⑤ MEASURED

AUG 11—31 · 20 DAYS

Location-page consult requests 2.90% → 3.60% of sessions (135/4,660 → 170/4,720). +33 requests above the prior rate on the same traffic.

⑥ CLOSED

TUE SEP 1 · THIS LOOP

Closed and made standing: financing stays above the fee block. Re-read at the December recommitment (FORM 610-Q). Goes back to the desk as a KEEP line on 247-K.

Three surfaces changed on the same day. We cannot split the lift between the page, the sitelink and the script, and we are not going to pretend we can — the honest claim is that the theme was worth acting on, not that the page did it. A single-surface ship would have been cleaner and would have cost two extra cycles; we chose the cycles, and we are telling you which we chose.

THE GATES — WHAT IT TAKES BEFORE A CALLER'S QUESTION IS ALLOWED TO MOVE MEDIA the instrument that stops us acting on noise

GATE A · MEDIA CHANGE

n ≥ 25 AND ≥ 2 desks

Anything that moves spend or changes copy across the network. One desk is a local condition; two is a pattern. 25 of a ~513-call monthly pool — 5% of everything you hear.

GATE B · ONE PAGE OR ONE OPS FIX

n ≥ 15 at that desk

Cheap, local, reversible in an hour, no spend attached — so it can move on a single desk's evidence. L-08-07 came through here.

READ GATE · BEFORE WE CLOSE IT

30 days AND a matched prior window

Nothing closes on a short window or an unmatched comparison. L-08-07 is shipped, looks good, and stays open at n = 16 because 16 isn't a read.

Why the gates exist. The WK 29 desk readout scored 15 calls and showed close-when-offered down 7 points. At n = 15, seven points is one call. Coaching a person on one call destroys the instrument; buying media on one call destroys the budget.

What a weekly loop would clear at your size — nothing. Your monthly pool is 513 scored calls, so a week is 118. Run this quarter's six themes against a weekly pool and they score 24.2, 15.7, 11.5, 9.7, 8.8 and 5.5. Not one reaches Gate A — the financing theme, the strongest thing in this ledger, misses by eight-tenths of a call — and the single-desk item doesn't reach Gate B either. A weekly loop here would ship nothing, or ship on anecdote. That is the argument for monthly, and it is about statistics before it is about money.

Where this loop stops working — the number to watch. Gate A clears one or two themes a cycle here, and the margin is thinner than it looks. The item we rejected (L-06-09) sat at 38 against a threshold of 25; a theme at that prevalence needs a pool of at least 325 scored calls to clear at all — and you are at 513 with two desks dark. Roughly four more and the whole 8%-prevalence band goes silent: we would still hold the loop, and stop hearing anything that wasn't already obvious. The scoring rate, not the location count, keeps this instrument alive.

WHAT THIS LOOP IS NOT PERMITTED TO DO four rules that keep it an instrument instead of a suggestion box

It never quotes a caller. Every theme is a paraphrased composite written by your desk manager against a fixed taxonomy. No transcript text, no caller detail and no name reaches media, or this document, ever. Recordings stay in your system under your retention policy.
It never moves on one call — however memorable, and however senior the person repeating it. The gates are the whole point, and the partner is the one who gets told no.
It never advertises capacity the schedule doesn't have. Learned the expensive way in June (L-05-02); the ops sign-off in the escalation table on p.3 is the scar.
It does not close its own items. A read is a read. If the n isn't there the item stays open and looks unfinished on page 1 for another month, which is correct.

WHERE THIS SITS — FORM 247-K FEEDS IT · FORM 610-M P.3 CARRIES ITS SPEND CONSEQUENCES · FORM 610-Q RE-DECIDES WHETHER IT STILL EARNS ITS HOURS · NOTHING IN THIS LEDGER ORIGINATES WITH US.

Cadence, Routing & Escalation — Who Owns Each Turn of the Loop SEP '26 · GROUP · 12 LOCATIONS · FORM 247-L · PAGE 3 OF 3 · SPECIMEN

THE CADENCE, PLAINLY — AND WHY IT ISN'T WHAT YOU WERE PROBABLY SOLD stated once, up front, so nobody discovers it in month four

ADMEN does not sit in on your weekly huddles. At twelve locations that is 52 huddles a month — 39 analyst hours before a single change ships. We would be selling you our attendance instead of your changes, and you would be paying for both.

So the loop is split, and the split is deliberate: the huddle is weekly and it is yours — the media loop is monthly and it is ours. We supply the instrument (FORM 247-K), your desk manager runs the 20-minute huddle with the people who took the calls, and the sheet comes to us. We sit in on one huddle a month, and on any week one goes wrong. Changes are decided at a single 45-minute loop on the first Tuesday, they carry a ship date, and they are read at the next one.

This is not a downgrade. It is the discharge posture (Rx 05) applied at the start rather than at the end: a system your team runs without us. A loop that only turns when we are in the room is a loop you are renting. At your call volume it is also the only version that works — a weekly pool of 118 scored calls clears no gate in this document (p.2). Weekly costs more than five times as much to hear less.

THE ROUTING — SIX TURNS, EACH WITH ONE OWNER AND ONE CADENCE · NO STEP IS OWNED BY "THE TEAM"
TURN	WHO RUNS IT	WHO IS IN THE ROOM	CADENCE	INSTRUMENT	OUTPUT
1 · HEAR — score the calls, hold the huddle	Your desk manager	The people who took the calls	Weekly · 20 min	FORM 247-K	A TO MEDIA line
2 · POOL — tag themes, count n, apply the gates	ADMEN analyst	Nobody — assembled, not met	Continuous · closes monthly	247-L intake	Gate-cleared candidate list
3 · DECIDE — ship / reject / hold, with reasons	ADMEN media lead	+ your ops lead · partner on exception	Monthly · first Tuesday · 45 min	FORM 247-L	This ledger, signed
4 · SHIP — make the change, log the date	ADMEN dev + media	—	Within 10 days of the loop	Change log	A dated change on a named surface
5 · READ — measure against a matched prior window	ADMEN analyst	—	At the next loop · 30-day minimum	247-L read	CLOSED, ROLLED BACK, or still open
6 · RE-DECIDE — does the loop still earn its hours	ADMEN partner	+ owner / CFO	Quarterly	FORM 610-Q	Continue / conclude

FEEDS: FORM 247-K (DESK READOUT) → 247-L → FORM 610-M (MONTHLY BOARD PACK, P.3 CHANNEL ECONOMICS) AND FORM 610-Q (QUARTERLY RECOMMITMENT). NOTHING ENTERS THIS LEDGER THAT DIDN'T COME OFF A DESK.

ESCALATION — WHAT BREAKS THE MONTHLY RHYTHM, AND WHO HEARS ABOUT IT FIRST
TRIGGER	WHO IS TOLD	WITHIN	WHAT HAPPENS
A theme touches safety, licensure or a payer claim	ADMEN partner	Same day	Out of cycle. Nothing waits for a Tuesday.
A shipped change reads negative at 30 days	Media lead + your ops lead	At the read	Rolled back by default. It survives only with a written reason in this ledger. L-05-02 is what that looks like.
A theme asks media to promise capacity — hours, same-day, a new service	Your ops lead — sign-off required	Before ship	No capacity claim ships without the schedule behind it. This gate exists because of L-05-02.
Scored calls fall below 325 in a cycle	Ops lead, then owner	At the loop	The 8%-prevalence band stops clearing Gate A and the loop goes half-blind. A threshold, not a soft target — see p.2.
WATCH fires — answer-rate collapse, booking flatline, tracking outage	Owner / ops lead	Same day	One-paragraph interrupt note: what broke, what it costs per day, what we are doing about it.
A desk logs calls but runs no huddle, two cycles running	Ops lead → owner at the quarterly	At the loop	The desk is named on page 1 and drops out of the ledger denominator. We do not average it away.

THE PART OF THIS LOOP THAT ISN'T WORKING — AGREED AT THE SEP 1 LOOP named here, not buried in an appendix

Two of your twelve desks ran no huddle in August — South 08, two cycles dark, and Parkway 11, which has never run one. That is 111 calls, 15% of the month's new-patient demand, that this ledger cannot hear. Worse: those are the two desks we are routing the most new demand toward, and the July board action put $6K/mo of paid search behind them. We are buying demand for the two desks we can't hear. We are not going to claim the loop covers 12 desks — it covers 10, and four more dark desks and this instrument stops finding anything you didn't already know. Both are the first item on the Oct 6 agenda with your ops lead, and if they are still dark at the December recommitment it comes to you as a staffing question, not a marketing one.