CG Common Ground | Jason Scott
What we did and what it producedCompleted

The work, decision by decision

The work, in the order the questions came

  1. The first read, in a day. Known: the deck, the brand identity book and the term sheet. Unknown: whether the founder's story survived contact with the documents. The question was what an investor would find in the first hour, so I built the read the way an investor would: founder and entity verification, the deck taken apart claim by claim, the instrument analyzed, and a written recap of the meeting. It produced a 78-page investor memo inside 24 hours of seeing the collection. What it changed: the founder saw that the deck undersold the product and that the numbers under it would not hold, and he engaged with the substance the same day.
  2. The brand deck, torn down and rebuilt. Known: the deck had typos, stale numbers and no visible mark. Unknown: whether the founder would move on the mark. Our brand team took the deck apart and built a wireframe for the rebuild. The founder kept the markless design, on purpose and by comparison to the quietest houses in the category, and that was his call to make. What it changed: the investor materials stopped competing with the product.
  3. The business plan, changed at the launch. Known: the plan opened with wholesale doors. Unknown: whether a pre-revenue brand should ask a pro shop to commit before a single member had bought a single piece. The question was where proof is cheapest to buy, and the answer was a trunk show: direct retail, no markdown, no receivable, and a room full of the exact customer. The channel architecture moved to trunk shows first, with a credit-not-markdown return policy and a scorecard written before the first show, with gates for sell-through, price integrity, contacts captured and reorders requested. What it changed: the first season became a proof event that opens the raise, instead of a wholesale bet that consumes it. This was the first place the business plan changed, and the rest followed from it.
  4. The model, door by door. Known: the founder's line, his factory terms, his lead times, his channel thesis. Unknown: everything a top-down plan hides. The question at every layer was the same: what does this number rest on. It produced a 29-sheet buy plan (Attachment E, held pending review), and a walkthrough with the founder in June where the new returns policy, a sweater-only contingency for the first two years, and a November launch window were confirmed. Each of those was a change to his business plan, agreed with the model open. What it changed: for the first time, every revenue dollar in the plan traced to a chassis, a colorway, a door and a season.
  5. The Assumptions Log. Known: some inputs were the founder's, some were industry benchmarks, some were nobody's yet. Unknown: whether presenting them with uniform confidence would mislead a careful reader into treating a placeholder as a fact. So every input got a class. Documented meant it came from the brand or its factories, like the deposit split and the lead times. Industry standard meant a benchmark from the category, like the margin targets and the cost stack. Placeholder meant a decision still owed: which doors actually launch, how fast each channel grows, when the personal-shopper team is hired, two prices, and the scope of the first holiday season. The counts are in the assumptions-log chart behind the gate. What it changed: the model's weakest inputs became part of what made it credible instead of something to hide.
  6. The sanity check. Known: a model with this many sheets can drift. Unknown: whether the headline revenue was one formula relabeled four times or four separately built views of the same number. Year-one revenue was rolled up by chassis, by channel, by assortment section times door count, and on the dashboard, from different source tabs, and all four tied to the dollar. The paths are in Attachment A. What it changed: nothing built on top of that revenue could be dismissed as a rounding artifact.
  7. Cash timing, where the gap lives. Known: the assortment implied a buy volume and a wholesale-versus-direct mix. Unknown: how long inventory sits in cash before revenue returns it. The question was the lag between the deposit at the order and the day a wholesale account pays, and what that lag costs in cash the business does not yet have. The next section is the answer.
  8. Delivering the finding. Known: the cash-flow tab contradicted the locked raise. Unknown: how a founder and a capital partner would take a finding that complicated a round in motion. The question was whether to state the gap plainly or soften it to protect momentum. I led with it, in what I sent him and on the call, with the model open so he could argue with any cell. What it changed: the founder and the capital partner could decide, with real information, whether to raise more, phase the plan, or keep refining the model under a monthly embed we proposed.

What the cash-flow tab said

The capital need was built as a bridge, each line addable and auditable on its own, not a single number dressed up. The first line is the lowest point cumulative operating cash reaches before any financing, once revenue collected is netted against inventory bought at landed cost and cash operating expense. Then half a year of first-year operating cost as a reserve, because a launch does not run at break-even cash from day one. Then the deposit the Italian factories require on the next six months of orders, because it falls due before the goods ship and cannot be paid from money that has not arrived. Then a contingency on the sum, standard for an early-stage forecast and not specific to this brand.

The raise as locked did not cover the first line. Before a dollar of reserve, deposit buffer or contingency, the peak burn alone was above it. The model showed the cash line crossing below zero in year two and still below zero at the end of year five, even as operating cash turned positive along the way. A raise that runs dry is best found before it closes, when the company has more to negotiate with than it will later. That is the sentence the memo could not avoid, and it is the reason the finding went first.

The bridge, line by line, and the burn-down against the raise both sit behind the gate, sourced to the buy plan model.

Why the locked number had looked right is worth stating, because a founder who sized a round to his plan's trough was not being careless. On the plan's own targets the trough looked shallower and the raise covered it with cushion. The live model, populated door by door with the operating cost the plan actually required, moved the trough deeper and later. The gap between the two is the difference between a plan and a model, and it is exactly what a bottoms-up build exists to find.

He was not asking for enough money. That was the problem. A number he could argue with was worth more to him than a number I agreed with.

The model's base case also came out far below the founder's own five-year projection. Those two findings pull in opposite directions and they are the same finding. A plan built from the top down had overstated what the business would earn and understated what it would cost to get there, and one build from the unit up corrected both at once. That is what changing his business plan meant in numbers.

What we kept, what we replaced, what we installed

Kept. The founder's line and his instincts about it; his mills and made-to-order factories, and the terms and lead times from his own trips, which were the best-documented inputs in the model; his channel thesis of clubs and resort communities, which the model kept and reordered rather than replaced; his no-logo decision, which was his to make.

Replaced. The business plan. A top-down five-year projection, with a bottoms-up buy plan. A wholesale-first launch, with trunk shows first. A raise sized to a plan target, with a raise sized to the cash-flow tab. Markdowns as the release valve for slow stock, with store credit on returns and a pre-committed markdown trigger by channel.

Installed. The buy plan itself, structured in Attachment A: the Assumptions Log, the cost stack from free-on-board through freight, duty and brokerage to landed cost, the margin targets by channel, the sell-through and markdown logic, the cash-timing model and the lead-time table. The four-path sanity check. The stress table. The trunk-show scorecard with its gates written before the first show. A gated data room the founder could walk an investor through. And the memo architecture in Attachment B, which is how an unwelcome finding has to sit inside an investor memo so the reader reaches it, believes it and keeps reading.

The process touched, and who put it there. The raise sizing was the founder's and the plan team's, and it was reasonable: one round, one dilution event, sized to the trough with cushion, decided in early July on the plan's targets. What was faulty in the logic was upstream. The trough had been read off targets, not off a model populated door by door with the operating cost those doors require, and a trough read off targets is shallow by construction. It had to change then and not later because factory deposits fall due before the first trunk show, and because a raise that is short is cheapest to fix before it closes and most expensive to fix in the year it runs out.

What it cost to hold the line, and what I watch

It cost me the engagement for a while. The finding complicated a round in motion, and putting it first read, to a founder who had already said his number out loud, as a correction of the company rather than of the plan. We proposed a monthly embed to keep hardening the model; the engagement paused in July over price and scope. I would put the finding first again. The alternative was to let him close on a raise the model said would run dry, and that is a worse thing to do to a founder than to lose a few months of his goodwill.

The relationship held. Since then, on the go-to-market side, the Discovery Land communities are hosting trunk shows by my referral, which is the channel the model was built around, and a private aviation company I introduced him to has placed a purchase order, the first order from a client of ours. On the capital side, he raised a smaller first round first, most of it in as of September 2026 and not closed, with a second round planned. The walk-down in Attachment F, behind the gate, shows that the raise as re-sized still sits under the cash-flow tab's need; that gap is the number I would want him to say out loud.

The constraint I did not remove is the anchor itself. A number said out loud to investors is expensive to change, and no message architecture changes when the model arrived relative to when the number was said. What the work did was make the real number checkable, so that when the founder chose to move to it, he could defend every cell.

What I watch is the placeholders. The inputs still owed a decision are the ones that move the trough: which doors actually launch in each tier, the growth multiplier on each channel, when the personal-shopper team is hired, and the scope of the first holiday season. The stress table below is the other thing I watch, because no scenario tested moved the need down, and the two that hurt most are the ordinary ones: sell-through missing badly in year one, and operating cost running over plan. The first trunk shows are the proof event that resolves the largest of these, and the scorecard was written so that a miss triggers a smaller year-one buy rather than a conversation about whether to have one. The stress table itself sits behind the gate.

What it produced

The raise as locked would not have covered the modeled need. He re-sized it into a smaller first round, with a second round planned. Discovery Land communities are hosting trunk shows by my referral, and a private aviation company has placed a purchase order.

A slice of the project list

A few related projects.