The $100M model nobody wanted to build
Twelve days before a capital allocation meeting, the executive asked me for three slides. The full value-chain model was a month away. The number on the one-pager would have been wrong.
Two weeks into the iron-ore haulage engagement, the executive asked me for a summary. Three slides. Something he could take into the next capital allocation meeting. The full value-chain model, covering spare parts, maintenance cycles, haul routes, driver utilisation, and three fuel contracts with three different indexation clauses, was a month off. He had decisions to make in twelve days.
I nearly agreed. The summary version would have looked credible. It would have cited the parts of the business that were easy to read and quietly ignored the ones that weren't. The executive would have walked out of the room with a number. The number would have been wrong by about forty percent.
Why models are boring is why they matter
Business cases fail in the parts everyone skips over. In mining, that is almost always the haul. The haul is where the interactions compound. A 2% change in truck cycle time reshapes maintenance schedules, driver shift rosters, port loading slots, and contractor margin all at once. Three of those interact non-linearly. Model one variable without the others and you get a confidently wrong answer. Model all three and the spreadsheet gets ugly.
The uglier the model, the harder it is to sell upward. The harder it is to sell upward, the more executives push for the summary. The summary is where $100M in savings goes to die.
- 01
Build the ugly model
Six tabs, three thousand cells, every interaction visible. The spreadsheet should be ugly — if it is pretty, you have simplified too early.
- 02
Extract the one-sheet dashboard
Surface only the three variables that actually move the P&L. The dashboard sits on top of the ugly model and cannot be shown without it.
- 03
Write the three-slide summary
Bet, evidence, decision. Slides are a read-out of the dashboard, which is a read-out of the model. In that order.
- 04
Deliver the decision, not the deck
The executive walks out with a signed-off bet. The model stays referenceable — three years later, someone will still need it.
Executives want the summary. The summary is only defensible when there is a bad-looking spreadsheet underneath it that nobody ever needs to see.
What shipped
I wrote the summary he asked for, but only as a read-out of the model he hadn't. Six tabs. Three thousand cells. A one-sheet dashboard that surfaced the three interactions that actually moved the P&L: truck cycle time, port slot utilisation, and a weighted fuel index nobody on the commercial team had yet stitched together. When he took it into capital allocation, what got approved wasn't the summary. It was the bet the summary represented.
The model is still referenced in their review cycle three years later. Not because I wrote it well. Because bothering to build it at all surfaced the parts of the business they had been flying blind on.
The operator takeaway
When a model feels too complex to build, that is usually the sign it is worth building. Executives who ask for a summary are not being lazy. They are being rational about their time. Your job is to collapse the complexity honestly, not remove it. A good one-pager always rides on a bad-looking spreadsheet underneath.
If the model is small enough to hold in your head, the business is too small for the model to matter. The question is never whether to build the ugly one. It is whether to show it.
This post came out of a fractional engagement I ran in 2023-25. The case study is on the Work page.
Perspectives, by email
Get the next essay in your inbox.
One a fortnight, sometimes less. No nurture funnel. Unsubscribe in one click whenever it stops being useful.
Found this useful?
Thirty minutes. Free. No prep needed.
If the diagnosis is clear without me, you go do it. If not, we talk about the sprint. Either way, the first call takes 30 minutes and costs nothing.
Book the callKeep reading
All posts
Most of your product is table stakes. The rest is the business.
A founder spent four months shipping features his prospects already expected. The demo was smooth. The feedback was devastating: 'nothing differentiates you.' Points of parity vs points of difference, and the research that tells them apart.
Some users forgive broken things. Others leave.
A founder shipped a rough beta to technical ops leaders and got 15 bug reports in a week. Another shipped the same category of product to regional facility managers and got silence. Your audience's tolerance for half-built is an industry variable, not a B2B/B2C one.
The runway story your investor update never tells
A founder I worked with sent a great-month update at $11k MRR. The number she did not put in the email was that her runway had dropped from thirteen months to nine. The growth was real. The runway story was the more important one.