Case Study Summary: A vertically integrated real estate and construction company needed an investor-grade financial model for a growth round, and the timeline shrank from three months to about two. The prior year’s books weren’t closed, there was no job costing, inputs had no clear owners, and the scope kept growing. Here is how we built a model that ties to the general ledger anyway, what the client had at the end, and five questions to ask before your next raise.
I spent most of this summer inside one spreadsheet. Somewhere north of eighty meetings, and one Friday that started at ten in the morning and finished at half past one the next morning. I want to write down what that was like while it’s still fresh, partly because it’s a good story and partly because I think a lot of founders are about to walk into the same thing.
The client was a vertically integrated real estate and construction company getting ready for a growth round. They brought us in over the summer to build the model an institutional investor would want to see, on a three-month engagement. Then the investor conversations sped up, and three months became closer to two.
Where the Time Went
Nothing we ran into was unusual. It was the normal state of a company that grew faster than its back office. The hard part was that all of it had to be fixed while the model was being built, with the clock running. The CFO seat also changed hands five weeks in, so the person signing off on the model at the end was not the person we scoped it at the start of the engagement.
Start with the base year. The prior year wasn’t closed when we started. It wasn’t closed a month in either. Large entries were still moving between periods with the auditors, and we didn’t get a “yes, it’s final” until week nine of a twelve-week engagement. You can’t forecast off a number that’s still moving, so we anchored everything on a trailing twelve months, which stays put while year-end entries bounce around, and converted to calendar years the week the base year was confirmed. That conversion touched most of the model and took three days. Fine. But those were three days we’d have had back if the books had been closed before we started.
Then there was no job costing. The CEO wanted to see economics by project. The ledger recorded costs by plant and by department. The business hadn’t needed project accounting before. Now it did. So, we built it: a register of every project mapped to the chart of accounts, so each forecast dollar could be traced both ways. That’s a finance infrastructure project, and it ran inside the model build.
The inputs were what really ate the clock. The pipeline lived in one file, plant costs in another, margin targets mostly in people’s heads. Each had a different owner. At one point we were two versions behind on the project list and didn’t know it, because the list had changed and nobody had sent it. We found out in a working session with six people in the room. The root cause, in their own words, was that nobody owned the list. I’d guess a quarter of our hours went to finding the current number rather than modeling it.
And the scope grew. As a lender was identified, with a new CFO in place, the asks got bigger: a standalone view of the operating business, a standalone view of the real estate side, a combined view that eliminated the gap between the two, per-project capital schedules, bonding and guarantee requirements, three plant capacity scenarios instead of two. Every one of those was a fair ask. We built all of it. We just built it inside a window that was getting shorter, not longer.
The last three weeks we worked seven days. Daily reviews with the incoming CFO and the operating leads. Twenty-three versions saved between a Thursday and the following Tuesday. The model was used for an investor meeting the following week.
What They Had at the End
A model that ties. Actuals reconcile to the general ledger to the cent. That should be table stakes. Most raise models I’ve seen don’t get there. On the way there, we found a multi-million-dollar balance sheet reclassification that the trial balance export had caught halfway through, and fixed it.
Three statements through 2033, five business lines, all driven off the same project register. Three scenarios on one switch, so the CFO can change a single cell and watch revenue, plant cost, capex, and the funding need all move together. A sources and uses page where every line reads a model cell, which means when an investor asks “where does that number come from,” there’s an answer, and it’s a click away.
Checks on nearly every tab, rolled up to one control panel. If somebody breaks something next month, the panel goes red before anyone presents it.
And the paperwork to hand it over. How the model is built, how to refresh it from the accounting system, where the traps are. The incoming CFO and the controller can run it without me.
The model was the loudest part of the engagement, not the whole of it. We were there as the CFO office, so the rest ran alongside it: a 13-week cash forecast with trigger points for the thin months, a capital structure review that re-layered a wall of near-term obligations into something the business could actually service, lender mapping and banking guidance, and an assessment of the finance function before the incoming CFO started five weeks in, so he knew what he was walking into.
Five Questions Before Your Next Raise
- Are the books closed? Properly, with the auditors, before the modeling starts. Everything downstream is provisional until they are, and the investor’s analyst will find that on day one.
- Does every input have a name? Pipeline, unit cost, margins, capital plan. One person, one file, one current version. Nearly every error I’ve chased in this line of work was an input nobody owned, not a formula.
- Are scope changes written down? Not to be difficult. Just so that when the ask list grows, everyone can see what it costs in time and what has to move to make room.
- Did you start early enough? Three months is about right for a model that ties, with controls and documentation. At two, it’s a sprint, and sprints make mistakes.
- Can someone else run it? If you’re the only one who understands the model, you don’t really have one yet. Checks and a short manual take about a week. It’s the week that usually gets cut.
Reframe
Founders tend to think of the model as the deliverable. It isn’t. The model is what you get when the finance function underneath it is in order. If the books are closed, the inputs have owners and the scope is written down, the model is the easy part. If they aren’t, the model becomes the place where every one of those gaps gets found, usually by the person deciding whether to write the check.
What does your finance function look like six months before a raise? Our CFOs2GO team works with growth-stage companies to close those gaps before diligence does.
About the Author

Clifton Leung is a CFOs2GO Partner and Practice Leader for Business Performance Management at 2GO Advisory Group, working as a consulting CFO for growth-stage and mid-market companies. He spent nearly a decade at PricewaterhouseCoopers and Deloitte advising executives on pricing, margins, M&A diligence and long-range planning, then led FP&A for a utility and energy infrastructure portfolio and ran finance and operations for a data consulting firm. Clifton holds an MBA from UC Irvine’s Paul Merage School of Business and is based in Los Angeles.
Email: cleung@cfos2go.com
Phone: 213-973-7914
For your Talent needs in direct hire, full-time or part-time contract staffing, contact Executive Recruiter, Leesa Meintzer at leesa@2gorecruiting.com.

Leesa Meintzer is an executive recruiter with more than 20 years of experience in talent acquisition. She excels in partnering across various business functions and brings a comprehensive perspective to talent acquisition. She works with Engineering, Healthcare, Product, Finance, Accounting, Business Operations, Sales, Legal, Human Resources, Learning & Development, and Talent Acquisition for corporate and high-growth start-ups.