The model is the firm's core instrument, and it has been refined across engagements for coworking operators, franchise groups, developers, and building owners. What follows is how it is built and why it is built that way. If you are reading a delivered model right now, the companion to this page is what each module covers and what the numbers say.
The problem the method solves
A financial model fails in a particular way. It rarely fails at the thinking. It fails at propagation: a change that was understood perfectly lands on some of the cells it should touch and not all of them. The model still computes, every total still agrees with its own lines, and nothing errors. It is simply describing a slightly different building than the one on the drawing.
That failure mode is invisible to inspection, because the artifact looks correct. So the defense cannot be care alone. It has to be structural: a model built so that a number physically cannot exist in two places, so that counts can only enter through the floor plan, and so that every figure carries its provenance with it. The seven principles below are that structure.
The seven principles
One number lives in exactly one place, and that place computes it
Every figure has a single home. The space allocation holds the counts, the pricing module holds the rates, the lease block holds the terms, and everything downstream references those rather than repeating them. A number typed into two places is a number that will disagree with itself eventually, and nobody will know which copy is right. This is also why changing one input moves everything it should move: the model is wired rather than transcribed.
Counts flow from the space allocation, never from a keystroke
The allocation is the only place raw unit counts exist. The rent roll reads them, the revenue build reads the rent roll, and the operating statement reads that. Nothing in the chain carries a hand-typed count. The consequence is that the model cannot quietly disagree with the floor plan it was built on, which is the most common way a financial model becomes internally perfect and commercially wrong.
The base case uses the documented term, never the target
Rent, abatement, escalation, tenant improvement, and lease term come from the landlord proposal or the executed lease, reconciled line by line against the document rather than read once into a summary. A rate or a concession still under negotiation appears as a clearly labeled scenario so the value of winning it is visible, and nothing is marked negotiated, agreed, or final unless it appears in writing. Free rent is the single highest stakes input this applies to, because it lands where cash is thinnest.
Every input is documented, benchmarked, or named as an assumption
A documented input traces to a source you can open. A benchmarked input comes from calibrated operating data across comparable locations or from surveyed local evidence. An assumption is a judgment made where no source exists, and it is labeled as one at the point it appears. There is no fourth category. A number that fits none of the three does not go into the model, which is why construction cost, dependent on quotes that do not exist yet, is sourced from a contractor rather than assumed in a spreadsheet.
The test fit is the truth root: revenue is drawn on the plan
The office program is established from a measured plan or a confirmed source, never estimated from an image, and the revenue is mapped onto that plan so the path from square footage to monthly income is visible as a picture rather than asserted in a table beside it. Counting what is physical and calibrating what is a membership is the working distinction: an office is a room you can point at, while a coworking membership is a right to show up that no drawing can tell you how many of to sell.
Rates come from market research, never inherited from the last deal
Every rate in the model traces to a rung on a researched pricing ladder for that submarket, with the comparable band, the count behind it, and a confidence rating attached. A rate that happens to match a previous deal is a coincidence to verify rather than a default to accept, and where a market prices by size, the model breaks out by size. Thin evidence is reported as thin rather than smoothed, because a confident number resting on one comparable is worse than an honest range.
A pro forma is a decision instrument, not a prediction
The model answers what a specific building produces under a stated set of inputs. It does not forecast, because results move with lease terms, build cost, financing, staffing, execution, and the market. Its value is that every assumption is explicit enough to be argued with, negotiated against, and tested before capital is committed. A study that ends in a decision not to proceed has done its job as completely as one that ends in a lease.
Structure first, pricing last
Structure and pricing are different kinds of input and they are settled in that order. Structure is facts: rentable area, rent, abatement, term, escalation, unit counts, room sizes, the program. Those are verified once against source documents and propagated. Pricing is judgment, and judgment wants a finished model underneath it, because the office ladder and the desk rate are the only inputs where the consequence of a change is visible in the result.
So the model is stood up on verified structure first, and the rates are tuned against a working result afterward. Carrying a pre-tuning rate is normal and is stated as such. What is never done is holding up the structural work waiting for pricing certainty, which delays the only thing that makes the pricing decision answerable.
Brand calibration versus deal facts
Models for franchise brands run on calibrated operating assumptions specific to that brand: royalty and marketing fund rates, the staffing shape, expense ratios, and the occupancy ramp itself. Those travel with the brand rather than with the property, and they are not adjusted deal by deal without a stated reason.
Deal facts are the opposite. Rentable area, rent, abatement, room counts, and room sizes come from the lease and the drawing for this specific property, and overwriting them is the work. The test that separates the two is simple: would this number exist if the operator started from an empty floor in a different city? If yes, it is calibration. If no, it is a deal fact. Departing from a calibrated standard quietly is not conservatism, it is undeclared drift, so a deviation is surfaced as a finding rather than absorbed.
The spreadsheet is the source, the portal is the reading
Every engagement is delivered two ways: the working Google Sheet in the format operators and lenders expect, and the portal that presents it. The sheet is the single source of truth, and the portal is a projection of it. When a number changes, it changes in the sheet and the portal is resynced everywhere that figure appears, including the headline tiles, the tables, the charts, and the narrative around them. A portal carrying a number the sheet no longer holds is treated as a defect rather than a version.
That arrangement is deliberate. The portal exists so the story is readable at a glance. The sheet exists so every formula can be interrogated. Neither substitutes for the other, and because both are produced from the same engine they cannot drift apart.
Why the model is designed to change
A pro forma that only works in its delivered state is a document. A pro forma built to be re-run is an instrument. Inputs move during a negotiation, a test fit gets revised, a rate is tested against a lower number, and the model is expected to absorb all of that and recompute. That is the reason the wiring discipline matters more than any single figure in the file: the value is not the numbers as delivered, it is that the numbers respond correctly when the deal moves.
See it in practice
A redacted sample portal shows the full format, with the client identity masked and the figures illustrative.
Open the sample Flex Intelligence Portal →
What each module covers, and what the numbers say →
Run the numbers on your floorplate →
Want this run on your building?
Bring the address, the floorplate, and the lease terms. What comes back is a model built on those specifics, with every assumption labeled for what it is.
Book a Strategy Call
