Studio 1 — fewer leads, a capped ad budget, and where else the money is

At 75% of today's leads, city by city

Bookings and revenue are scaled by each city's own lead flow at that city's own booking rate, show rate and price — nothing is averaged across cities. The chairs column is what each studio would have to close to hold its current fill; leave the chairs open and the revenue is the same, the fill just falls.

† Florida and Orlando share one lead pool. The FL codes shoot at both studios, so the pair's 15,808 leads a month are split here by each studio's share of chairs sold. Revenue and the diary are studio-true and need no split. Never quote a Florida-only or Orlando-only lead cost.

What it does to the year

The caveat that matters most — the first month is cushioned

Everything above is a steady state. September 2026 is the only time we have actually run a cut this size, and it did not behave like the arithmetic: leads fell 24.8% (20,251 → 15,227 over days 1–11) and the phone room's bookings went UP 13% (739 → 654 the other way — 739 in September against 654 in August) on the same headcount, because it worked a backlog of leads that had not been called. Maturity-matched at three days it was 225 against 246, only −8.5%, well inside the May–Aug range of 218–252. So the first month or two of a lead cut is absorbed by the stock of unworked leads, and the full 25% only shows up once that stock is gone. Do not read a good first month as proof the cut was free. Measured 12 Sep 2026 on /month-gap.

Five days a week, or fewer chairs every day?

A 25% lead cut means 941 of the 3,766 chairs a month have to come out if fill is to stay at 79.3%. The question is which 941. The last column is the one that decides it: what each option costs in revenue above or below the $886,597 a month the lead cut already forces on you.

The answer, and what it is worth

Close Tuesday and Wednesday, not Saturday and Sunday. Saturday and Sunday are the two fullest days we trade (82.2% and 81.4% fill) and they carry 48.5% and 50.6% kids — the segment that pays more and no-shows less. Closing them takes out 35.0% of every person who arrives to save 34.2% of the chairs: that is $356,166 a month worse than the lead cut alone.

Tuesday and Wednesday are already the thinnest days in the week — 13.9 and 15.1 chairs a studio-day against 19.2 on Saturday — and we already do not open every studio on them (29.3 and 28.0 studio-days a month against 36.0 on Sunday). Closing them removes 830 chairs — 88.2% of what you need — for only 21.9% of arrivals, which is less than the 25% the leads already cost you. It is also the only option that removes studio-days, and studio-days are where the fixed cost lives: the building, the shift, the photographer.

The remaining 111 chairs should come off Monday morning — the weakest real block in the week at 69.2% fill and 43.5 arrivals per 100 chairs offered.

The better question: five days in the studios that aren't full anyway

Every option above assumes the days you keep stay the same length. They don't have to. If the remaining five days hold more appointments, the chairs come back and no booking is lost at all — the only thing saved is the fixed cost of opening the building twice a week. That turns the decision from "which revenue do we give up" into "can the bookings fit into five days", which is measurable.

It is already maxed where it matters and untouched where it doesn't. 09:30 runs two appointments 74.0% of the time — so double-running is a proven practice, not a theory. But by part of the day:

So: 16.4 chairs a studio-day now, 24.7 if every band doubled up as often as 09:30 already does — 50.7% more, not 70%. The five-day move needs 19–21. Comfortably covered, and that is the answer to "do we open longer": no — you run two at once in the afternoon.

And the headroom is almost all in the afternoon, which is the good news. Mornings already double up 28.2% of the time; the afternoon only 5.6% — yet the afternoon is the fullest part of the day (87.2% against the morning's 71.5%) and the least kids (25.6% against 52.4%). So the spare capacity sits exactly where demand is strongest and it does not touch the morning kids engine — the thing September broke by cutting mornings.

The second appointment also sells better, not worse: doubled slots fill at 88.2% against 76.0% for single slots, with the same show rate (71.1% vs 69.6%). Read that carefully though — it is SELECTED. The studio puts a second person in when it already expects demand, so 88.2% is not proof a second appointment anywhere would fill that well. It is proof customers are not the thing stopping it.

What is NOT in the data, and it decides the cost. Why the afternoon is single-run is not recorded anywhere — it could be crew (the photographers are on the morning kids rush), a deliberate call, or just habit. And two simultaneous appointments need a second photographer and probably a second viewer in that half-hour. So "+8.3 chairs a day" is a capacity statement, not a free one, and the crew cost is the number to go and get. Note also that New York is already the most double-run studio in the network, which is consistent with it being the one that cannot fit into five days.

Two honest caps in the test below. Day length is each city's own p95 chairs-per-day, not its best day ever — every city has one to three freak days at 32–42 chairs against a 15–18 mean, and planning off an outlier says Nashville could do a whole month in ten days, which is nonsense. Fill is capped at 90.0%, which is the top of what has actually been observed (the best cells are the 88–91% Friday/weekend afternoons), not a comfortable target.

Your instinct is exactly right, and the data draws the line for you: only 3 of the 8 studios can do it today, and they are the under-filled ones — Nashville, Houston, Chicago. Moving just those three to five days frees 25.1 studio-days a month (10.9% of the network's 230) and loses nothing — no chairs, no bookings, no revenue. It is the only move on this page that is free.

The other 5 cannot, and it is worth seeing why: New York needs 5.2, Boston needs 5.1, Dallas needs 5.6, Orlando needs 5.4, Florida needs 5.0 days a week. Dallas and Orlando are simply too full to lose two days — which is the same thing as saying they are the studios doing their job. Boston is the odd one out: it already trades only 5.4 days a week, so there is barely a second day left in it to close.

And here is where it joins up with the lead question. At 75% of the leads, all 8 studios fit into five days — every one of them with 3.4–5.9 days a month to spare — freeing 65.7 studio-days a month, 28.6% of the network, with no bookings lost beyond the ones the lead cut already took. So the five-day move and the lead cut are complements, not alternatives. The lead cut costs you $886,597 of revenue whatever you do; the five-day move is how you take the cost base down to match it instead of running a half-empty seven-day estate.

This supersedes the three options above as the recommendation. Closing Tue+Wed at constant day length costs 21.9% of arrivals. Closing two days and lengthening the other five costs none. The difference is about $777,297 a month, and it is entirely a scheduling decision.

Three things that have to be true, and one is not measured. (1) The demand has to move days — a Tuesday customer has to accept a Friday. Nothing in these feeds says whether they will, and it is the single assumption the whole thing rests on. The weekend is already the fullest part of the week (81–82%), so the displaced volume has to land on Mon/Thu/Fri, not Sat/Sun. (2) A p95 day has to become the normal day — those are currently one day in twenty, and running them routinely needs more crew per day, which eats into the saving. (3) Cost per studio-day is unknown, so this is priced in days, not dollars — but days are what a lease and a rota are written in.

Option 3 is the trap — it loses the same revenue and saves no cost

Taking 25% off every day (16.4 → 12.3 chairs a studio-day) loses exactly 25% of arrivals by construction, and frees no studio-days at all — every studio still opens 29 days a month, still staffed, still lit. It is the option that feels least disruptive and is the only one that saves nothing.

By day of week — May 26, Jun 26, Jul 26, Aug 26, settled

Settled months only, so every booking's outcome is known. A forward diary reads 9–20% full midweek purely because it has not been sold yet — that is a lead-time artefact, not a midweek weakness, and it is why this table is built on closed months.

And by time of day — this is where September went wrong

Mornings look like the obvious cut and they are not. Morning fill runs 69.2%–74.6% against 84.1%–90.8% in the afternoon, so per chair the morning is plainly the weaker slot. But the morning is where the kids are — 63.6% of Saturday morning bookings and 67.1% of Sunday morning bookings are children, against 36.7% in the Saturday afternoon. A kids shoot cannot move to 2pm; an adult can. Cut the morning and the kids booking simply does not happen.

This is not a theory — we ran it. In September 2026 seats on offer were cut 8am −33%, 9am −18%, 10am −21%, 11am −22% against −3% to −10% in the afternoon. Settled adult appointments held flat (539 → 546); kids fell 42% (385 → 225), and kids × price accounted for about 85% of the entire revenue gap. Fill did not move at all — 76.1% both months — so the chairs were not the problem, the shape of the cut was. Measured 12 Sep 2026 on /month-gap.

One 2-day kids studio, built entirely from measured rates

A 2-day studio trades Saturday and Sunday — 8.7 days a month. Today's weekend studio-day puts 18.4 chairs on offer, so one new site is 160 chairs a month. Kids show at 70.4% against adults' 62.1% and bring $1,990 a head against $1,707 — so a kids chair is worth $1,400 booked against an adult chair's $1,060, 32.1% more.

The plan survives the kids lead price doubling, and that is the single most important thing on this tab. Ad spend is only 12.1% of a 2-day studio's revenue at $11.55 a lead and 20.7% at $19.68. A kids lead would have to reach $102.29 — 9× today's price — before it stopped covering its own ad cost. The risk in this plan is not the lead price. It is the site cost and whether the leads exist at all.

How many do you need?

Two different questions, and they give two different numbers. Both net off the adult revenue the re-pointed lead spend was already producing — a kids lead bought is usually an adult lead not bought, and ignoring that overstates every answer.

Your 12–15 gets you most of the way, not all of it. Holding revenue flat on 75% of the leads needs 2 new sites — 10 in total, the top of your range. Fully replacing the $886,597 the cut costs needs 20, so 28 sites. At 12 sites (4 new) you recover $180,958 of the $886,597 — about 20.4%. The gap between the two answers is real and worth understanding: holding revenue flat is easier than replacing contribution, because the kids mix shift does some of the work for free.

The constraint is kids leads, not chairs — and you called the studio count, not the buy

We buy 14,391 kids leads a month today and convert them into 1,115 kids bookings at 6.81 per 100. Seven new 2-day studios need 13,010 more — roughly 1.9× the kids volume we buy now. Every one of them comes from a single supplier: Blue Rooms is the only kids source in the business. That is the thing that decides whether this plan works, and it is a supplier question, not a property question. Ask Blue Rooms for 27,401 kids leads a month before signing a lease, and get the price in writing at that volume — a supplier whose CPL already went $5.58 → $16.53 in four months will not hold $11.55 while you triple the order.

The good news on the mix shift: it is accretive even before the new sites. A kids lead returns $102.29 against an adult lead's $69.06 — 1.5× — at a cost of $11.55 against $9.67, which is cheaper. So swapping one adult lead for one kids lead adds about $33 of revenue and costs nothing extra. The ceiling on doing that without new sites is the weekend grid — today's weekend day runs 18.4 chairs against a measured physical ceiling of 22–24, so there is roughly 25% of headroom on the weekends you already trade. Fill that first: it needs no lease.

What would have to be true — and what is not established

There is already a live sizing tool for this with sliders — https://reports.iconicbyai.com/popup-planner — and the measurement behind it at https://reports.iconicbyai.com/weekend-effect. This tab is the network-level version of the same question and uses the same measured rates.

Which cities are actually under price pressure

None of them is close to losing money on ad spend — the worst city still clears its ad cost more than four times over. So price pressure here is relative, not existential. Spend is the Apr–Jun 2026 monthly average, the last countable window, with Blue Rooms at invoiced cost; revenue and volume are May–Aug. Read the FRESH column, not the all-leads one — the all-leads column is kept only so the size of the old error is visible.

Chicago is the one genuine problem, and it is a double squeeze. It pays $11.51 a lead — near the top — and earns the least per lead of any city at $54.51, which is why it ends up spending 21.1% of revenue on ads. It also runs the thinnest kids mix at 20.6%. Compare New York at 11.6% on $7.25 a lead — it buys leads for 1.6× less than Chicago does.

High cost per lead is not the same as a problem. Boston pays $17.37 for a fresh lead — the dearest in the network — but earns $67.05 from one, also the best, so its contribution per lead is fine. Judge a city on contribution per lead, never on cost per lead. Ranked that way the order changes and the cities worth acting on change with it.

What is still missing from this, and it is not small. Alan's own cost sheet stops at June 2026, so July and August carry invoiced Blue Rooms but no Alan at all — those two months are still not a total and are not used in this table. And the Blue Rooms feed is not lapsed: its API answered last-synced 16 Sep 2026 09:00, the same morning. The standing note that the cost feeds died in July is about our puller, not the source. Writing a puller against that API is the single highest-value fix on this page, and it is one GET request.

Adult and kids, separately — because they are not the same business

Kids cost is Blue Rooms' own kids campaigns over May + Jun 2026, the two months the USA Models daily file covers completely; adult cost is everything else over the same two months. Revenue and conversion are May–Aug, split by the diary's own age flag — who actually sat in the chair — not by the marketing code.

The finding that flips the obvious plan: the cities with the BEST kids economics are the ones where we buy the FEWEST kids leads. Dallas earns $149.49 of contribution per kids lead — the best in the network, on 11.69 bookings per 100 kids leads — and yet kids are only 16.8% of what we buy there, the thinnest mix of the seven. Meanwhile New York has the heaviest kids mix at 35.8% and only middling kids contribution. The kids mix is close to inversely related to how well kids actually perform. That is the cheapest thing on this page to fix: it is a buying decision, not a building decision.

Independent check on that conversion column: the 11.69 figure for Dallas is computed here from the diary's age flag against code-attributed leads. The /roi cohort measured KIDSX separately in Aug 2026 and put Dallas top at 12.16 per 100 with Chicago and New York bottom at 7.58 and 7.56. Different feed, different basis, same ordering.

Kids cost per lead is the number that has moved, and it is not in this table

The kids CPL above is May + Jun 2026 — $11.55 network. Blue Rooms' own weekly feed was measured again in August and read $19.68 for kids against $18.52 for adults over 26 Jun – 14 Aug. That is roughly double the May–Jun kids figure, and it matches the supplier trend already on record — Blue Rooms went $5.58 a lead in April to $10.98 in June to $16.53 in July. So treat $11.55 as the cheap end of a rising range, not as today's price. Everything on the 2-day studio tab is stress-tested against the higher figure.

Fresh adults against recycled — the booking rate you asked for

Your ~3% is right. Over May 26, Jun 26, Jul 26, Aug 26 the whole recycled pile books at 2.90 per 100. A fresh adult lead books at 4.90 per 100 — so a fresh adult is 1.69× a recycled lead, and recycled would need to be 68.8% more efficient to match one booking for booking.

Ordering trap worth knowing, because getting it wrong breaks two answers at once: KDX ends in "DX" and KPX ends in "PX", so the kids families have to be matched before the Google ones or Google silently swallows the Google-kids codes. The recycled R is stripped first, so a recycled #DALGXR stays with Google rather than falling into the R bucket. Fresh adults on this page therefore means fresh adults EXCLUDING Google.

Google — GX, PX and DX — split out, as asked

You are right about them, and DX is worse than either of the other two. Together GX + PX + DX are 3,562 leads a month, 6.3% of intake, booking at 2.06 per 100 against a fresh adult's 4.90 and returning $20.93 of revenue per lead against $69.06 — a fresh non-Google adult lead is worth 3.3× a Google one.

The cost side, and one correction. A note from August recorded that neither GX nor PX had any cost data. That is now half wrong: GX does. All eight city GX campaigns carry daily spend in the USA Models feed, so GX can be priced properly. PX and DX still have no cost data in any feed — not in the CRM per-code column, not in Boost's daily costs, not in USA Models — so their $16.72 and $5.30 per lead are revenue-side only and their true margin is unknown. Given DX returns $5.30 a lead against a fresh cost of $16.53, it would have to be nearly free to break even.

Two things that stop this being an open-and-shut case. PMax was stopped once already, so some of this volume may be residue rather than live buying — check before cancelling something that is already cancelled. And PMax has historically had no settled cohort (it carried almost no leads before ~Jun 2026), so its rate is measured on thin and recent volume. The GX and Google-kids numbers are the solid ones; DX at 330 leads a month is small enough that its rate is noisy even if its direction is not in doubt.

Basis, and it matters. These are same-window rates — bookings made in a month over leads that arrived in it — so a lead that came in on the 30th had no time to book and every family reads low. The /roi cohort measured the same families properly in Aug 2026 and got fresh adults 5.48, CMN 10.00, 5AB 3.67, 6AB 3.21, N1N 3.20, R-suffix 4.28. Use the cohort numbers for absolutes and these for the ratios — the same-window bias hits every family the same way, which is why the ordering and the multiples below survive it.

Why the booking rate is the wrong way to judge a recycled lead

A recycled lead costs no new ad money. That is the whole asymmetry. A fresh adult lead costs $16.53 — adult ad spend divided by fresh adult leads only, which is the right denominator because the recycled ones were free — so a booking bought through fresh adults costs $337 of ad spend. A booking that comes off the recycled pile costs nothing. So recycled at 2.90 per 100 is not "worse" than fresh at 4.90 — it is free, and the only things it consumes are a booker's time and a chair.

Cross-check on that $16.53: /cpl-history computes the network fresh cost per lead independently, over Apr–Jun 2026, and gets $14.42. Different route, same number to within a few cents.

So: how much more efficient, and what is each point worth?

Every percentage point you add to the recycled booking rate lets you stop buying 2,841 fresh adult leads a month — $46,951 a month, $563,413 a year. Getting recycled to fresh-adult parity needs 68.8% more efficiency and is worth $1,124,695 a year of ad spend you would simply stop spending.

Or keep buying the fresh leads and bank the bookings instead: parity would add 278 bookings a month, worth about $294,298 of revenue at the adult rate of $1,060 a booking — and there are 781 unsold chairs a month to put them in, so the capacity already exists. Those are alternatives, not additions: take the saving or take the bookings, not both.

The sharpest version: the gap is inside the recycled pile, not between recycled and fresh

CMN books at 5.87 per 100 and N1N at 1.79 — a 3.3× spread between two families of our own recycled leads. On the cohort basis the spread is wider still: CMN 10.00 against N1N 3.20. So the 68.8% of efficiency the table above asks for is not hypothetical — one of our own recycled families already beats a fresh adult lead, and it does it on leads that cost nothing.

N1N alone is the prize. It is 4,219 leads a month at 1.79 per 100 — the worst rate of any family in the business. Lift N1N to CMN's rate and nothing else, and it is 172 extra bookings a month, 3,511 fresh adult leads you stop buying, and $58,029 a month — $696,351 a year. Start there.

What this does not tell you, and it is the part that decides feasibility. CMN and N1N are different stages of the same recycling process, not different suppliers, so the gap is about which people get re-sent and when — and that is exactly what these feeds cannot see. CMN has already been established as the one recycled family that supersedes whatever code a lead arrived on, so a CMN lead may simply be a better person rather than a better-worked one. Before committing to a lift target, check whether CMN's advantage is the treatment or the selection — compare lead age, prior contact count and prior booking history between the two piles. If it is selection, the 68.8% is not available by working N1N harder.

Dropping the far areas and paying more for close ones

The gradient is real and it is worth about $1.98M a year — but the number that decides the plan is not the gradient, it is how much more a close lead costs. 226,664 fresh in-range leads, 2026-04-01 to 2026-09-16, off /distance-age's own cache so the geocoding, the drive table and the show-rate denominator are that report's and not re-derived here.

The finding that answers "who is bordering on not making a profit"

On the cash that actually arrives on the day, the far bands do not cover the cost of the lead. A fresh lead costs $14.42. Inside 25 miles it brings $23.49 of same-day cash — comfortable. Beyond 200 miles it brings $12.43, which is $1.98 LESS than the lead cost. Cash-negative bands: 200 mi +, No zip / not geocoded. On drive time the same thing happens past three hours, and the prefix-not-on-the-table band is $5.20 under water.

They only look profitable on signed revenue — and signed revenue is mostly a finance plan. Crediting the financed balance at the 40.0% that actually gets collected, every band clears its cost, the worst at $19.91 against $14.42. So both statements are true and they answer different questions: nothing is loss-making over the life of the agreement, and everything past 100 miles is loss-making on today's cash. Which one matters depends on whether the constraint is profit or cashflow.

By distance

Index is on collected value per lead — cash plus 40.0% of the financed balance — against the nearest band. Book %, show % and close % are that report's own, on the roster capped at today so a booking made for next month is not scored as a show.

By drive time — and the first hour is FREE

Do not discount inside an hour. ≤ 30 min returns $22.67 of same-day cash and 30–60 min returns $23.42 — slightly MORE. The value only starts falling past an hour. Paying a premium for a 20-minute lead over a 50-minute one would be spending money for nothing, and it is the mistake this table exists to stop.

Why the loss is upstream, and what that changes

Deal size and close rate get BETTER with distance, not worse. Close rate runs 70.3% inside 25 miles and 81.2% at 100–200 miles — someone who drove two hours buys. The entire loss is in booking % (5.53% → 3.04%) and show rate (65.1% → 54.6%).

That matters because it means there are two ways to fix a far lead and only one of them is a media decision. Stop buying it, or get it to turn up. Being confirmed has already been measured as beating distance by about 5× as a predictor of showing up, and 12.6% of no-shows were never contacted at all. Fixing confirmation is free and it works on the whole book, not just the far end.

And age is a bigger lever than distance. The distance spread on collected value is 1.84× from nearest to furthest. The AGE spread measured on the same data is wider — an under-18 lead was worth $27.33 of cash against $8.71 for a 55+ lead, 3.1× (measured 20 Aug 2026 on /distance-age). If you are re-pointing the buy, tightening the age target pays more than tightening the radius, and the two are additive.

One concrete thing to drop today

Florida ZIP 33101 — 1,619 leads, 2 bookings, 0.1%. The biggest single zip in the whole dataset and the worst-converting. 33101 is the generic downtown-Miami / PO-box zip and it is the exact zip the supplier target list names for Miami, so it reads like a default being stamped on leads with no real address. Full list at https://reports.iconicbyai.com/zip-conversion, which ranks the worst zips by leads burned rather than by rate — more than 20 zips sit at exactly 0.0% and a rate sort pushes the biggest problem off the page.

What this cannot tell you, and it is the crux. Ad spend is held per SUPPLIER, never per distance band, so a true cost per lead by band is not derivable — these are value indices, not margins. The 56.4% premium is therefore a ceiling on what you may pay, not evidence that close leads are available at that price. Tighter geo-targeting shrinks the audience and normally raises the price; whether it raises it by more or less than 56.4% is a question for the supplier, and it is the one thing that decides this. Source: https://reports.iconicbyai.com/distance-age.

How many bookers would each city carry?

A booker can be allocated to a city because they behave like one: the median booker takes 97.4% of their bookings in a single city, so their top city is their home city rather than an accounting convenience. 77 distinct names took a booking across May 26, Jun 26, Jul 26, Aug 26, and the median day had 61 on shift.

Why that last column is a question and not an answer

It holds leads per booker constant at 735 a month and scales the room to the work. That is the right arithmetic and it is almost certainly the wrong answer, because we already know the room has slack: in September leads fell 24.8%, headcount was held, and the phone room made more bookings, not fewer. A booker working 735 leads a month converts 4.49% of them — so the binding question is whether 735 leads a month is a ceiling or a floor for one person, and nothing in these feeds answers that. If it is a floor, 25% fewer leads needs materially fewer than 58 bookers. If it is a ceiling, cutting the room at all costs bookings.

What would settle it: the leads-per-booker figure at which a booker's own conversion starts to fall. That is a per-booker, per-day cut of the bookings feed against the leads actually distributed to them, and the distribution side is not in any feed on this page. Until it is measured, treat 58 as the most the network needs at 75% of leads, not the right number.

Two payroll facts that sit next to this decision

Quoted from /booker-levels, measured 14 Sep 2026 — not recomputed here. Over 12 months $35,735,338 booked against $4,446,464 paid, 12.44%; the Paycom runs give $4,528,604 to the 96 bookers (≈12.67% of turnover) and $5,555,882 across 143 people company-wide. Booker pay is a percentage of turnover, so it falls with the leads on its own — the headcount question is about the base salaries and the seats, not the commission. Separately, 25 bookers sit above their own ladder rate, $294k a year, and the median at every level is exactly the ladder rate — so that $294k is 25 individual decisions, not a structural overpay.

A $600,000 a month cap — what it buys

The "average 3 month lead price" is taken over Apr 26, May 26, Jun 26 — the last three months the cost feeds cover well enough to count. Every figure is Σspend ÷ Σleads over those three months, never an average of three monthly averages: volume moves too much for those to agree.

Running lead cost at 18% of revenue

Two answers, and they disagree in a way that is the whole point.

The ratio cannot be improved by cutting leads. At 75% of leads, spend falls to $447,832 and revenue to $2.66M — 16.8%, the same number. The only things that move it are revenue per lead ($63 today) and the price of a lead ($10.80). Every lever on this page that actually moves money — the no-shows, the unsold chairs, the leads we buy and throw away — works on the first of those.

Ad spend and cost per lead, month by month

Two sources and then a wall. The CRM's own per-code column is complete to Jan 2026 and nobody has entered spend since. From Feb 2026 the figure is rebuilt from the supplier feeds — Alan's invoices, Boost's daily ad costs, Blue Rooms via USA Models, Lead Pronto at its real blended rate — each with a per-day coverage test. A month whose coverage is short is shown and never counted, and no estimated dollar is ever added to a spend figure.

Jul and Aug 2026 cannot be answered and are not estimated. July is 96.3% covered (Alan missing 8 days) and August 0% (Alan all month, Boost all month, Blue Rooms 11 days). The cause is the lapsed ad-cost scraper logins — only Neil can fix that, by running node scripts/reports/login_costs_profile.js and logging back into USA Models and Alan's leadcodes. Until then every cost figure on this page ends at June 2026.

Adult cost per lead, by city, the 12 months to Aug 2026

Say this first: 10 of the 12 months can be answered and 2 cannot. July 2026 is 96.3% covered and August 2026 has no cost data at all — both are printed dimmed and neither is counted into any average. Nothing is estimated to fill them.

Blended over the 10 countable months — Sep 25 to Jun 26

Σ adult spend ÷ Σ adult leads, cheapest city first. Never an average of monthly averages — the volume moves too much for those to agree.

Three things that will make you misread this table

1. Kids did not exist as a code family until about April 2026. Kids were 0% of intake through May 2025, under 2% to January 2026, then 11.3% in April, 23.6% in May, 27.4% in June and 28.5% in July. So on the five CRM months the "adult" cost is effectively the whole cost, and the months where adult and kids genuinely diverge are exactly the months whose spend data is thinnest. An adult-vs-kids comparison across the full year is not available and should not be constructed.

2. The basis changes in February 2026, and that alone moves the number. Sep 2025–Jan 2026 is the CRM's own per-code column; Feb–Jun 2026 is rebuilt from supplier invoices. The step from $8.30 in Jan to $11.71 in Feb is a source change as well as a price change and the two cannot be separated here. Treat the two blocks as two series that meet, not one trend.

3. † Florida and Orlando must stay joined. The FL codes shoot at both studios and Alan's cost sheet has no Orlando line at all, so Orlando carries $0 of attributed spend before July 2026. A Florida-only or Orlando-only adult lead cost would be wrong in both directions at once.

You named four aspects. You are missing one, and it is the biggest.

Your four were fixed costs (back-office wages, studio costs), ad spend, staff wages, and money collected in the back end. Those are four real buckets. What is not on the list:

Your $0.00M–$0.00M a month does not reconcile from what I can measure, and the gap is the interesting part. Ad spend is $597,110 and the Paycom runs put $462,990 a month across 143 people — about $1,060,100 together. That leaves roughly $639,900–$839,900 a month in studio rent, utilities, the product itself, software, card fees and any back-office payroll not on that sheet — none of which is in any feed I can read. If you want the cost side modelled rather than assumed, that is the missing input and it has to come from the bookkeeping, not from here.

Where reductions could come from, largest measured first

The back-end book — flat for 20 months, and shifting from people to the robot

This is the aspect you named fourth, measured. It collects $795,760 to $876,645 every single month — a 10% band over 20 months with no trend in it at all. Note the failed column: roughly as many attempts fail as succeed, every month.

Nothing here retries a failed payment. There is no automated retry anywhere in this business — every second attempt is a debt collector on the phone taking a card, or a customer paying a reminder link the team sent. That is why the "failed" column is not a retry queue, and why any conversion rate measured on a second attempt is selected and cannot be read as a lever: a collector only puts a card through once the customer has agreed to pay. The safe numbers are the ones counting things that did not happen.

The cash number nobody should budget against yet

Three measurements of cash that do not reconcile, and I am not going to pick one. The CRM records about $0.00M a month collected up front on new sales (37% of signed). The finance book collects $828,103 a month. Those sum to roughly $0.00M. But the clean Square extraction over 1–14 Sep 2026 found 3,157 COMPLETED payments worth $746,724 in fourteen days — about $0.00M a month across everything — with another $654,252 attempted and never collected in the same fortnight. $0.00M and $0.00M are not the same business. One clean per-location pull, filtered on payment status, is needed before any cash-in figure goes into a plan — and that is a day's work, not a guess.

Every basis on this page

What this page cannot answer