Mariana Tek's studios are often the multi-site ones — a brand with three rooms across a city, one account, one member record that books at any of them. That is a real operating advantage, and it produces a reporting problem that arrives quietly: the company total is up, and one site is down, and the total is the reason nobody noticed. A flagship on a good quarter covers a second site's soft one for months. By the time the second site's numbers get their own look, the fade has been running since spring.
The fix is a per-site scoreboard — the same five questions, one row per location, one company row — built on three rules that most multi-location reports break. Here are the rules, the trap inside each, and the scoreboard they produce.
Rule 1: every member has one home location, by a written rule
A member who books across sites is the multi-location advantage and the reporting hazard in one person. Count her at every site she visits and the company has more members than it has people; count her nowhere and the sites sum to less than the company. She needs one home location, by a rule you decide once and write into the definitions alongside what counts as a member.
Two rules are defensible. First-visit location — the site that actually met her — is what intro cohorts and conversion rates need, because the studio that earned the membership is the one that hosted the first class. Most-visited location over a window is what capacity and staffing need, because that is where she actually takes up a spot. Either is fine. Switching between them, or letting each report pick its own, is what makes two managers show up to the same meeting with two member counts for the same site.
Rule 2: recompute rates from counts — never average the sites
This one is arithmetic, and it is broken on most portfolio dashboards we have seen. A rate — churn, intro conversion, fill — is a fraction with a denominator, and the denominators differ by site. Average the sites' rates and the small site counts as much as the flagship.
The company churn rate is the company's cancellations divided by the company's members — sum the counts, then divide. The same rule applies to intro conversion (joiners over starters, summed), fill (attendance over capacity, summed), and cost per member (spend over joiners, summed). It sounds obvious written down. It is not what a spreadsheet with a row per site and an AVERAGE at the bottom does.
Rule 3: show the rows that carry no location
Here is the one that makes owners think a board is broken. Sum the sites and the total does not reach the company number — and it is not supposed to, because some of the data carries no home location. Profiles created before a location was set. Online-store revenue that belongs to no room. Accounts that book everywhere and satisfy no attribution rule. In one multi-site studio we work with, roughly one member in eight carried no location at all; the sites summed to about 88% of the company, and both numbers were right.
A scoreboard that shows a "no location recorded" row and a company row is telling the truth. One that silently drops the unattributed data so the sites add up is inventing a number. The disclosure also does useful work: a large unattributed row is a data-hygiene task with a name, and it shrinks as the desk sets home locations.
The scoreboard
With the three rules in place, the scoreboard is the five questions per site, judged against each site's own history and its own break-even — never against the flagship. A site with a smaller room and a newer team is not underperforming because its fill is lower than Uptown's; it is underperforming if its fill is lower than its own last quarter's.
Notice what the company row would have said on its own: revenue at 99% of pace, churn at 3.7%. Fine. Park's row says churn has doubled against its own history and intro conversion has fallen by a third — the flagship was covering it, and the recomputed company rate was telling the truth about the whole and nothing about the part. Switch on the morning email's location list for Park's manager, and Park's numbers land in her inbox each weekday, so the site's fades have an owner and nobody argues about whose members are whose.
Which location actually made money?
You cannot tell from the company total, and usually not from revenue. The scoreboard above answers the member questions per site. The money question needs one more join: each site's revenue against its own payroll and rent. Five numbers do it, and the fifth is the one that turns a finding into a decision.
Read the rows in order. Revenue says Uptown is the smaller site, which you knew. Payroll as a share of revenue says it is staffed like the larger one. Rent says its room costs more. What is left after both says one location is carrying the other, and the estimated break-even says why: a class at Uptown has to draw eleven people before it pays for itself, against seven downtown, so the same half-full schedule that works at one site loses money at the other. That is not a verdict on Uptown. It is the list of what to fix: the classes running under eleven, and whether the rent was set for a membership the site has not reached yet.
The new site: out of the averages until it is live
A location has three stages — planned, in presale, live — and only the third belongs in the company averages. A presale site has a founding cohort, a launch goal and no fill history; average it in and every company rate drops while the new site's actual progress hides inside the blend. Give it its own row with its own goals — founding members against the presale target, then fill against its own break-even — and add it to the company rates on the day the doors open. The second-location piece covers the readiness tests before the lease; this is what the board does after it.
| Rule | The trap it closes | How to apply it |
|---|---|---|
| 1 One home location | a member counted at every site, or none | first-visit or most-visited, written down, never switched |
| 2 Recompute, never average | a 100-member site weighing as much as a 900-member one | sum the counts, then divide, for every rate |
| 3 Show the unattributed row | sites that don't sum to the company, read as a broken board | a "no location recorded" row, and the hygiene task it names |
| + Own history, own break-even | the smaller site judged against the flagship | each site vs its trailing months and its own line |
| + Stage before averaging | a presale site dragging every company rate down | planned · presale · live; averages from live only |
Where this lives
Mariana Tek holds the raw material for all of it: one account, every location, one member record that books anywhere, with the location on every visit and sale. The scoreboard is a reading of that record under three rules — and the rules are the part a report cannot supply, because the attribution, the recomputation and the disclosure are decisions, not data.
That is the shape Xyzios gives a multi-location Mariana Tek studio. It connects on top — Mariana Tek stays your system of record across every site — reads the studio's own history as changes arrive, and builds the scoreboard under one set of definitions: attribution by your rule, rates recomputed from counts, the unattributed row shown, and a new site kept out of the averages until it is live. The fades and the follow-up are drafted per site for your approval. Nothing reaches a member without it. Nothing is written to Mariana Tek except what you tap or approve — a check-in at the front desk (and its undo), or a message you approved.
Wherever you build the scoreboard, write the three rules down first. The company total will keep saying everything is fine; the site rows are where the truth lives. See how the board runs a portfolio, or see how Xyzios sits on top of Mariana Tek.