Everything measured about chairs, fill, shows and what a capacity change costs. Written 16 September 2026. Every figure names the report and date it came from — nothing on this page is computed at build time, so re-measuring produces a new number with a new date rather than silently editing this one.
The chair is the binding constraint, not the lead. Only 1,945 of 3,766 chairs a month — 51.6% — have a person sitting in one. The other 1,821 split into 781 never sold, 833 booked-and-no-showed and 207 with no recorded outcome. At $1,823 an arrival the no-show pile alone is $1.52M/month of revenue whose ad spend is already paid, and 12.6% of those bookings were made to someone we never actually reached.
So the first capacity question is not “how many chairs should we open?” It is “why is half of what we already open empty?”
May–Aug 2026, four settled months, monthly averages. September 2026 is excluded on purpose — it cut chairs 17% and is the decision being modelled, not a baseline to model against. It appears once, in the Options tab, as evidence.
| What | Number | Source |
|---|---|---|
| Leads bought | 56,576 / month | /capacity-scenario, 16 Sep 2026 |
| Bookings made | 2,540 / month (4.49% of leads) | /capacity-scenario, 16 Sep 2026 |
| Chairs on offer | 3,766 / month | /capacity-scenario via lib/diary.js, 16 Sep 2026 |
| Chairs sold | 2,985 / month — 79.3% fill | /capacity-scenario, 16 Sep 2026 |
| Chairs arrived | 1,945 / month — 70.0% show, 51.6% utilisation | /capacity-scenario, 16 Sep 2026 |
| Sales | 1,747 / month | /capacity-scenario, 16 Sep 2026 |
| Revenue | $3,546,389 / month | /capacity-scenario, 16 Sep 2026 |
| Revenue per lead | $62.68 | /capacity-scenario, 16 Sep 2026 |
| Revenue per chair sold | $1,188 | /capacity-scenario, 16 Sep 2026 |
| Revenue per arrival | $1,823 — the price of an empty chair | /capacity-scenario, 16 Sep 2026 |
| Studio-days traded | 230 / month across 8 studios | /capacity-scenario, 16 Sep 2026 |
| Chairs per studio-day | 16.4 | /capacity-scenario, 16 Sep 2026 |
| Bookers | 77 on the roster, 61 on a median shift | /capacity-scenario, 16 Sep 2026 |
| Cross-check the page prints | $14,185,554 vs the /spend-vs-cash engine’s $14,185,654 — $100 apart on $14.19M | /capacity-scenario, 16 Sep 2026 |
One month, top to bottom. Each step names what falls out of it. This is the table to put in front of anyone who thinks the answer is more chairs.
| Stage | Chairs / month | What it means | Source |
|---|---|---|---|
| Raw diary grid | 3,036 seats | Meaningless as capacity — ~half closed because the studio is shut | /month-gap, 12 Sep 2026 |
| Chairs on offer | 3,766 | A + N + I + L. The only honest capacity number | lib/diary.js |
| — never sold | 781 | Nobody ever booked them. 20.7% of everything we opened | /capacity-scenario, 16 Sep 2026 |
| Chairs sold | 2,985 | 79.3% fill | /capacity-scenario, 16 Sep 2026 |
| — booked then no-showed | 833 | Sold, paid for in ad spend, and empty on the day | /capacity-scenario, 16 Sep 2026 |
| — no recorded outcome | 207 | Neither arrived nor marked no-show. A data gap, not a result | /capacity-scenario, 16 Sep 2026 |
| Chairs with a person in them | 1,945 | 51.6% of what we opened | /capacity-scenario, 16 Sep 2026 |
At $1,823 per arrival: the 833 no-shows are $1.52M/month and the 781 never-sold chairs are $1.42M/month of revenue the building was staffed and lit for. Both are already paid for in ad spend and rent. Neither needs a single extra lead.
All four are sized against the same requirement: a 25% lead cut needs 941 of 3,766 chairs closed to hold fill. What separates them is how many arrivals go with the chairs, and whether any fixed cost actually comes off.
| Option | What it costs | Detail | Source |
|---|---|---|---|
| Close two days, LENGTHEN the other five | Frees 25.1 studio-days/month today (10.9% of the network) — losing no bookings at all | Only 3 of 8 studios fit today, and they are exactly the under-filled ones: Nashville (72.6% fill), Houston (74.5%), Chicago (80.1%). Dallas needs 5.6 days/wk, Orlando 5.4, New York 5.2, Boston 5.1, Florida 5.0 — too full to lose two days, which means they are doing their job. At 75% of leads all eight fit, freeing 65.7 days = 28.6% of the network. | /capacity-scenario round 3, 16 Sep 2026 |
| Close Tue + Wed (same day length) | Removes 830 chairs (88% of a 25% target) for 426 arrivals — 21.9% | The best of the fixed-length options and the only one that frees studio-days, where fixed cost lives. We already do not open every studio then (Tue 117 of ~144 possible studio-days, Wed 112). SUPERSEDED by the option above, which costs no arrivals at all. | /capacity-scenario round 1, 16 Sep 2026 |
| Run 7 days thinner | 16.4 → 12.3 chairs/studio-day. Loses exactly 25% by construction | The option that feels least disruptive and saves no fixed cost whatsoever — the building still opens seven days. Loses the revenue without banking the saving. | /capacity-scenario round 1, 16 Sep 2026 |
| Close Sat + Sun | Removes 1,289 chairs but 681 of 1,945 arrivals — 35% | Do not do this. Ten points worse than the leads require, ≈$354k/month worse than necessary. They are the two fullest days (82.2% / 81.4%) and carry 48.5% / 50.6% kids — the highest-value segment in the building. | /capacity-scenario round 1, 16 Sep 2026 |
Every earlier option assumed the kept days stay the same length. They do not have to. Lengthen the remaining five days and the chairs come back, so no booking is lost — all that is saved is the fixed cost of opening the building twice a week. This supersedes the Tue+Wed recommendation, which cost 21.9% of arrivals (≈$777k/month).
Method trap: day length must be each city’s own p95 chairs/day, never its max. Every city has 1–3 freak days at 32–42 chairs against a 15–18 mean; planning off the outlier said Nashville could do a whole month in ten days. Fill is capped at 90% — the top of what has ever been observed, not a comfortable target.
The 5-day move and a lead cut are complements, not alternatives. The cut costs $886,597/month whatever you do; going to five days is how the cost base comes down to match, instead of running a half-empty seven-day estate.
Days 1–11 against August days 1–11: $962,591 vs $1,264,868, −23.9% (−20.7% calendar-adjusted). Decomposed multiplicatively:
| Factor | Move | Share of the gap |
|---|---|---|
| Chairs put on offer | −17.0% (1,308 → 1,086) | 68.1% of the entire revenue gap |
| Fill rate | −0.0% (76.1% both) | 0% — every chair we opened sold at the usual rate |
| Show rate | −3.7% | 13.8% |
| Price per sale | −3.4% | 12.6% |
| Sales per arrival | −2.0% | 7.4% |
Chairs were 68% of it and fill did not move at all. Every chair opened sold at the usual rate and the phone room made more bookings (739 vs 654) on the same headcount. But flat fill is also consistent with demand falling in step — if we had cut the diary with demand still there, fill would have risen. It did not. So chairs and demand came down together and this data cannot say which moved first. What it can say is that there is no pile of unmet demand pressing on the reduced diary.
And the cut fell on mornings, so the whole loss was kids. Seats by hour: 8am −33%, 9am −18%, 10am −21%, 11am −22% against afternoons −10% to −3%. Settled appointments: adults 539 → 546 (flat), kids 385 → 225 (−42%). Kids × price is ~85% of the entire revenue gap. The shape of the cut chose the casualty.
Settled months only. Retro-close tested and passed: March still holds 883 unsold empty seats and share-of-grid-on-offer is stable at 40–45% every month, so old months were not bulk-closed and these percentages are sound.
| Month | Fill (sold ÷ offered) | Utilisation (arrived ÷ offered) | No-show |
|---|---|---|---|
| March 2026 | 75.4% | 46.6% | 34.7% |
| April 2026 | 72.2% | 44.4% | 34.9% |
| May 2026 | 80.1% | 51.8% | 30.7% |
| June 2026 | 79.4% | 51.1% | 30.4% |
| July 2026 | 79.5% | 52.5% | 28.7% |
| August 2026 | 78.1% | 51.1% | 30.2% |
The whole spring-to-summer improvement is no-shows, not bookings. Fill moved +2.7 points (75.4 → 78.1) while utilisation moved +4.5 points (46.6 → 51.1), because no-show fell 34.7% → 30.2%. We did not sell more diary. We got more of the sold diary to turn up. Weekly range across 26 settled weeks: 70.7% to 84.5%.
| Day | Fill | Utilisation |
|---|---|---|
| Saturday | 81.5% | 52.2% |
| Sunday | 79.8% | 50.3% |
| Friday | 78.4% | 49.9% |
| Tuesday | 75.7% | 49.6% |
| Wednesday | 75.4% | 49.3% |
| Monday | 75.1% | 46.3% |
| Thursday | 74.6% | 49.3% |
A 7-point spread, Saturday best and Thursday worst. This is the settled picture and it matters: a forward snapshot shows midweek days at 9–20%, which looks like a structural midweek collapse and is not — it is lead time. Weekends still matter disproportionately because they carry the kids (48.5% / 50.6%), not because the fill gap is large.
Nothing is opened beyond about 17 days out. In a 28-day forward window the last week has under 1% of the grid on offer — there is literally nothing there to sell. Fill by distance out: 0–6 days 68.7% · 7–13 days 46.6% · 14–20 days 34.2% · 21–28 days 25.0% (on 20 seats). Whether the horizon is a deliberate policy or just how far ahead the team works is an ops question nobody has answered, and it caps how far in advance any plan can load.
Today’s book is live at https://reports.iconicbyai.com/diary-fill, rebuilt daily.
Ordered by how much of it there is. The top two are contact problems, not demand problems — which is why they are the cheapest things on this page to fix.
| Loss | Size | What it means | Source |
|---|---|---|---|
| No-show — last 3 complete months | 29.73% | Falling: Dec 2025 41.8% → Jul 2026 28.6%. Never quote the 12-month 34.22% blend as current. | /no-shows, 10 Sep 2026 |
| Sold to someone we never reached | 12.6% of bookings, 80.7% no-show | 3,879 chairs a year sold to a person the team rang and never got hold of; 3,129 of them sit empty. A 55-point gap against fully-confirmed bookings, and not circular — 19.3% of them still arrived. | /no-shows, 10 Sep 2026 |
| Confirmed → no-show | Y 25.6% · H 35.0% · N 47.1% · A 80.7% | Confirmation is ~5× bigger than distance or time of day, and both of those act through it. Fix contact, not geography. | /no-shows, 10 Sep 2026 |
| Never sold at all | 781 chairs / month | 20.7% of everything opened. Nobody booked them. | /capacity-scenario, 16 Sep 2026 |
| Cancellations | ~706 / month | 98.3% ARE taken off the diary and the seat goes back on sale — the ghost-slot theory was tested and does not hold. | /diary-fill, 7 Sep 2026 |
| Cancelled with time to resell | ~80% give 2+ days notice | August freed 694 seats through cancellation and ended the month with 803 unsold. The cancellation flow is the same size as the entire hole in the diary — the seat is freed and then nobody sells it. | /diary-fill, 7 Sep 2026 |
| Distance | 31.2% at ≤10mi → 39.5% at 100–200mi | Holds inside adults, inside kids and inside all eight cities. But it acts through confirmation. | /no-shows, 10 Sep 2026 |
| Dormant studios | ~986 seats each, 100% empty | Apex A, Apex B and Icon Studio each sit ~986 seats open and completely unused over 28 days. Maps B and Soul Studios are 100% closed. Whether any is re-openable is an ops question, not a data one. | /diary-fill, 7 Sep 2026 |
3,879 chairs a year are sold to someone the team rang and never got hold of, and 3,129 of them sit empty. That is an 80.7% no-show rate against 25.6% for a fully-confirmed booking — a 55-point gap, and confirmation is about five times bigger than distance or time of day, both of which act through it.
And ~80% of cancellations give us two or more days’ notice. August freed 694 seats that way and finished the month with 803 unsold. The diary does release the seat — 98.3% of the time — and then nobody sells it.
Neither of these needs one extra lead, one extra chair or one extra pound of ad spend.
This is the split that decides every capacity question, because kids sit in the mornings and at weekends — exactly the blocks a capacity cut tends to reach for.
| Measure | Kids | Adults / detail | Source |
|---|---|---|---|
| Show rate | Kids 70.4% | Adults 62.1% | /capacity-scenario, 16 Sep 2026 |
| Revenue per arrival | Kids $1,990 | Adults $1,707 | /capacity-scenario, 16 Sep 2026 |
| Revenue per chair booked | Kids $1,400 | Adults $1,062 — a kids chair is worth 32% more | /capacity-scenario, 16 Sep 2026 |
| Where they sit | Mornings 55% kids | Afternoons ~25% — so any morning cut is a kids cut | /month-gap, 12 Sep 2026 |
| Weekend concentration | Sat 48.5% / Sun 50.6% kids | 55.1 kids per weekend day vs 28.0 per weekday | /capacity-scenario · /weekend-effect |
| A 2-day kids studio | 160 chairs/mo → 89 shoots → $177,094 | Contribution before site cost $161,670/mo; $1,816 per shoot at any fill (it is linear) | /capacity-scenario round 2, 16 Sep 2026 |
| Free headroom before any lease | 18.4 chairs per weekend day vs a measured ceiling of 22–24 | ~25% more weekend volume needs no new site at all. Take this before signing anything. | /weekend-popup sizing, 18 Aug 2026 |
| The binding constraint | Kids LEADS, not chairs | 7 new sites need ~2.0× today’s kids volume (14,391/mo), all from one supplier whose CPL went $5.58 (Apr) → $16.53 (Jul). Get a written price at that volume before signing a lease. | /capacity-scenario round 2, 16 Sep 2026 |
The cities with the best kids economics buy the fewest kids leads. Kids contribution per lead: Dallas $154.01 (11.69 bookings per 100 kids leads — best in the business) on the thinnest mix at 16.8% of its buy. New York has the heaviest mix (35.8%) at $84.35. Chicago is worst at $71.64. Independently corroborated — /roi’s August cohort put Dallas top at 12.16 per 100, Chicago 7.58, NY 7.56: different feed, different basis, same ordering.
Re-pointing the kids buy toward Dallas is the cheapest capacity move available and needs no new site, no new chair and no lease.
Before any new site: today’s weekend day runs 18.4 chairs against a measured physical ceiling of 22–24, so ~25% more weekend volume needs no lease at all. Take the free headroom first. And note Austin/San Antonio would move volume, not add it.
Put this in front of anyone asking for a capacity decision. The first two carry the whole subject, and neither is expensive to answer.
| Open question | Where it stands | What it blocks |
|---|---|---|
| Cost per studio-day | Not in any feed here. The single most valuable missing input — every "close two days" number on this page is an arrivals-and-revenue figure with no cost side, so the saving cannot be stated. | Blocks option pricing |
| Whether demand will MOVE DAYS | The five-day plan needs a Tuesday customer to accept a Friday, and no feed says whether they will. The weekend is already the fullest part of the week, so displaced volume has to land on Mon/Thu/Fri. A p95 day also has to become the normal day, which needs more crew per day and eats the saving. | Carries the whole 5-day case |
| Is 735 leads/booker/month a ceiling or a floor? | Unknown. Needs a per-booker cut against leads actually distributed. 58 bookers is the MOST needed at 75% of leads, not the right number. | Blocks headcount planning |
| The September cushion | Month one of a cut is absorbed by unworked stock — bookings MADE went up 13% on 24.8% fewer leads because the room worked a backlog. Maturity-matched at 3 days it was only −8.5%. The full effect arrives once that stock is gone, and we have not seen month two. | Makes any one-month read optimistic |
| Why those studios closed in September | Studio-days traded fell 83 → 77 and the days traded were shorter (15.8 → 14.1 seats). Why is an ops question the data cannot answer. | Blocks root-cause |
| Repeat no-showers | The booking carries a noshowcount per person and it has never been looked at. Cheapest unexplored thing in this subject. | Untouched |
| The booker spread | No-show runs 18.5% to 63.1% across the top bookers — far bigger than distance or time of day, and not yet explained. | Bigger than anything measured |
| ~$640–840k/month of cost | Ad $597k + Paycom $463k ≈ $1.06M against a stated $1.7–1.9M/month. Rent, utilities, product, software, card fees and off-sheet payroll are in no feed in this repo. | Blocks any full P&L |
1. Cost per studio-day. Every option on this page is priced in arrivals and revenue with no cost side. Until someone puts a number on what it costs to open a building for a day, “close two days” cannot be compared against “lose some arrivals”. It is one number and it unlocks the entire Options tab.
2. Will the demand move days? The five-day plan rests entirely on a Tuesday customer accepting a Friday. Nothing in any feed says whether they will, and it would take one deliberate test in one under-filled studio — Nashville, Houston or Chicago, all of which already fit today — to find out.
The seven rules that govern every conclusion above. Most of them exist because breaking one produced a wrong answer that had to be walked back.
The booking calendar is a 24-hour grid, so roughly half of every studio’s seats are marked closed simply because the studio is shut then. Counting those as capacity makes a studio that closes its quiet mornings look empty rather than disciplined. The only honest denominator is seats on offer = A + N + I + L (booked + empty + incentive + locked), which is what scripts/reports/lib/diary.js computes and what every figure on this page uses. The raw grid is 3,036 seats a month; chairs on offer are 3,766 — different populations, and the grid number is never the answer to anything.
Fill is sold ÷ offered (79.3%). Utilisation is arrived ÷ offered (51.6%). The 27.7-point gap between them is the no-show rate, and it is the single largest number in this whole subject. A plan built on 79% capacity that actually delivers 52% will be wrong by a third. Where this page projects arrivals it multiplies by a measured show rate and says so.
The diary sells in the last fortnight. A day 13 days out reading 20% is normal, not lost. Six months of settled history land at 72–80% however thin the book looks a fortnight ahead. Judge the forward book on the days-ahead curve, never on one blended number, because the blend hides which end is short. This is the mistake that makes a healthy diary look like a collapse.
At a 4.49% booking rate, leads buy chairs faster than chairs get filled: 781 chairs a month are never sold at all. So a lead cut of 25% does not hit a wall of unmet demand — there is no pile of demand pressing on the diary (September proved it: chairs fell 17% and fill did not move). The one place this reverses is a kids studio, where the binding constraint is kids lead supply, not chairs: 7 new sites would need ~2.0× today’s kids volume from a single supplier.
September cut chairs 17% and is the decision being modelled, not something to model against. Every baseline on this page is May–Aug 2026, four settled months. September appears once, in the Options tab, as the only ~25% cut we have actually run.
Kids show more (70.4% vs 62.1%), pay more ($1,990 vs $1,707 per arrival), and a kids chair is worth 32% more than an adult one ($1,400 vs $1,062 per chair booked). They also sit in the mornings (mornings run 55% kids, afternoons ~25%). Any capacity change that touches mornings is a kids decision whether or not anyone intended it — September cut mornings, adults held flat, and kids fell 42%, about 85% of the entire revenue gap.
The diary carries an age on every booking; the marketing code carries what we bought. Mixing them — dividing diary-age revenue by code-attributed leads — overstated the kids per-lead figure by ~18% ($108.46 published, $102.29 correct). The two bases agree 99.1% but they answer different questions. Say which one a number sits on.
Capacity has no single feed. It is assembled from four:
| Source | What it gives | Watch out for |
|---|---|---|
diary/service/events.phpvia lib/diary.js | The diary itself — one row per seat per 30-min slot, typed booked / empty / closed / locked / incentive, with age and arrival outcome. | The CRM has no diary-fullness report. Booked Client Data returns neither date nor studio, so it cannot answer a capacity question at all. |
csv_noshowcancelledlib.php | No-shows and cancellations in one file. | Split on whether Date Cancelled is filled: blank = no-show, filled = cancellation. Reconciles exactly with the diary (807 = 807 for Aug 2026, studio by studio). |
The sales feed, joined on Model Name | Revenue per chair, per age band, per day. | The sales feed has no date column — revenue is dated by joining back to the diary seat. 99.8% match, 99.87% of revenue. Use the join for distribution only; totals come off the sales feed. |
lib/finance_cities.js | Studio → city. | Studio 1 is FLORIDA, not New York, and Maps A is New York. The two older maps in this repo both get it wrong. |
https://reports.iconicbyai.com/Studio1_Capacity_Scenario.html — the model and the scenarios
https://reports.iconicbyai.com/diary-fill — today’s book, rebuilt daily
https://reports.iconicbyai.com/no-shows — the confirmation finding
https://reports.iconicbyai.com/month-gap — current month vs the previous four
https://reports.iconicbyai.com/Studio1_Figures_Handoff.html — every figure and the basis it sits on