The decisions waiting on you

The chain is: our audit takes days to a couple of weeks · the insurer approves in 2–3 weeks if the paperwork is right, 6–8 if it isn't · then they pay in 60–120 days. So money we find in week two lands in their bank somewhere between month three and month six. If we only get paid when they get paid, our first real revenue from a client is three to six months after we start working for free. That is the cash-flow risk the plan already names, now with a number on it.

On this one lever — work done and never billed — restoration has the weakest proof of any industry I looked at. Every published restoration number fell over when I checked it. Meanwhile 77% of specialty trade contractors have already written off work they couldn't collect, in a real survey, which means they already know they have the problem — that removes the hardest part of the sale. Trucking is next ($15.1bn a year, less than half of those invoices ever get paid). This is one input, not the whole decision — restoration still wins on job size, urgency, marketing fit and your own history.

🔴 REWRITTEN 5 August after verification — the problem is bigger than I first told you, and it's about HOW WE'RE PAID, not what we do. Florida's law lets a contractor discuss an estimate with the insurer as long as he's being paid his usual fee for the work. The moment the fee becomes a share of the insurance money, that protection disappears — and the restoration industry's own association tells its members the same thing: avoid contracts that pay you a percentage of insurance proceeds, because courts read that as public adjusting. Texas is harder: a contractor may not adjust claims on property he's also working on, even if he holds the licence. So getting licensed doesn't fix Texas. (Drop Oregon — I named it earlier and the evidence doesn't support it.) This rules out our most natural pricing model in the two biggest restoration states in the country.

This closes the single biggest unknown in the entire plan. The "22 to 28% more per job" figure has now failed verification twice, and two numbers that surfaced this week turned out not to exist in the articles they were credited to. More searching won't fix it. One friendly owner's real jobs settles it permanently, in our own data, in a form we can actually show a prospect.

Your note says hundreds of pieces a week, lots of handles on every platform, the Gary Vee approach. The go-to-market section of the same plan says the opposite — content as proof, not reach: one genuinely specific teardown of a real company's numbers beats a year of posting. Both are real strategies and both can work. They just can't both be true under one name.

The reason it matters more here than it would for most businesses: the thing we're selling is precision. The most persuasive move we have is showing the three savings we refused to count. A company selling operational excellence, publishing 300 pieces a week of filler, has argued against itself in public. Gary Vee can run pure volume because volume is his brand — the brand in this plan is the opposite one.

There's also a hard finding behind this. When people watch testimonial-shaped video — someone vouching for results — an AI presenter scored 43% lower on trust, and viewers couldn't say why it felt off. Our proof content is testimonial-shaped by construction. So whichever way you go, the proof stays your real face and your real voice, permanently — the question is only what sits alongside it.

"Lots of handles on every platform" is a real strategy but it's also a real cost in attention, and it's the one part of the setup I can't start until you say where. Spanish is worth calling out separately — you live in Mexico, and Spanish versions are the one place an AI avatar genuinely earns its place, because nobody's trust is on the line in a translated explainer. That's a lane nobody else in this market is running.

Restoration has your story and real credibility. But the insurance company sets the price, cash takes 60–120 days, and margins are 10–20%. Foundation repair is customer-paid, so nobody controls their price, margins are 20–28%, cash is fast — and their number one problem is cost per lead, which is your decade, not something we'd learn.

Paul Davis is a franchisee. Franchisees are contractually forced onto their franchisor's software — Paul Davis mandates Xactimate licences, Microsoft 365 at set prices, Symbility and FAST estimating. Servpro mandates DryBook, WorkCenter and QuickBooks Online, described as not optional. They also run 8–14% lower profit because of royalties they aren't allowed to cut. A franchise pilot deletes part of our offer before we start — and if it comes back weak, we won't know whether the model was weak or the sample was.

Your recording says under $3M pays around $7,500, up to $10M pays twice that, and $10–50M "pays twice that" again. That reads either as 2× the base or 4× the base depending on what "that" points at. I won't write a price list off a maybe.

Things we hand over (switched payment processor, killed subscription) run whether we're there or not — only a clock keeps us paid. Things we operate (claims recovery, referral engine, AR chasing) stop the month we stop — they're already permanent and already cancel-anytime. On a restoration company the handed-over stuff is about 3% of the value.

You floated 1–5 years or forever, at least 6 months. Six months is too short — it doesn't even cover the restoration cash cycle, where one claim takes 60–120 days to pay. And there's a real reason not to go long: a permanent revenue share is something a buyer finds in due diligence and knocks the price down for — while we're simultaneously pitching "we make you worth more when you sell." Those two can't both be in the deck.

You've been working off 30%. Published outcome-based fees in this world run 10–25%, with a common rule that the total cost shouldn't exceed half the first year's savings. 30% is defensible because we carry the whole build and the whole risk — but it should be a decision, not a default.

We front everything — the audit, the connection, the build, the operator — across every client at once. The 90-day commitment softens this a lot because that cash lands on day zero. I won't guess this one; it comes from a live cash pull, never a file.

We place Sidekicks inside engagements and take a percentage of the savings. So we get paid more when we place our own people. Your words: "even when they come in at $8 an hour instead of $12, we're still saving them that money, we're still taking a percentage." The arithmetic works. The optics don't unless we say it first — and a client who works it out alone never trusts another number we give them.

Several of our clients' biggest controllable cost is us. This will come up on a call if we haven't settled it first.

You asked whether $1–3M is reasonable. Honestly: that's where the model stops working, not where it starts. A $2M company yields maybe $12–30K a year to us, while the audit, the connection, the build and the operator cost almost exactly what they cost at $15M. Break-even at best, against 5–10× the profit at $15M for the same effort. A $2M shop usually doesn't have a back office to fix or a marketing budget to move.

Your own note on Sunday, and your own doubt with it: many small operators may be the ones losing share to the businesses we serve, so it could be built on a shrinking base. I haven't resolved that and I'm not going to pretend to.

You said we prove it on current clients. I'd rather you name them than have me pick — some of these relationships are load-bearing and I can't see the politics from here.

Three finalists from the session on 3 August. Aurora was your favourite — dark and luminous, a real fluid simulation you can stir with your mouse. Meridian is the opposite: warm paper, ink and oxblood, a drawing machine that builds the artwork line by line and never the same way twice. Klein has no pictures at all — type, lines and colour, cream and black with one electric blue. Boldest, and cheapest to build out.

Open since July. Your personal brand can't be scoped until this is answered — "Nick Deck, rebuilding his own agency in public" and "Nick Deck, founder of a new company" are different people with different audiences.

The founder story we drafted opens with it — "I didn't come to AI to scale, I came to it because I couldn't work." It's the reason the whole story works, and it's yours to fence. There's a safer version that leads with the money leak and lets the collapse arrive as the reason you were looking.

Flagged in July as the single biggest structural risk and unanswered since. You at 10–15 hours a week plus one builder is a project, not a company — and value pricing in a brand-new category is founder-led selling, so revenue caps at your calendar unless someone else can sell it.

Every hours-based number on the Value Visualizer scales off a $12 placeholder I refused to invent. Under a model where we take a percentage of these numbers, that placeholder has to become a real figure.

Every H&S figure on the Visualizer is quoted from June 2026 files. The standing rule is that dollar figures come from a live pull, never a file. Say go and I'll refresh all of them.