DRAFT FOR REVIEW — not yet announced. Prices, names and wording are proposals.

Example — fictional company

The X-ray: Tidewater Millwork

How your business probably runs, where it's likely leaking, and three to five things worth looking at first.

Prepared for Tidewater Millwork (custom architectural millwork & cabinetry, ~40 people, two shops)
Date September 2026
Inputs 10-minute intake form + public website (tidewater-millwork.example.com)
Reviewed by Pete Kaminski

How to read this

Everything below was inferred, mostly from your website and the handful of answers you gave us on the intake form. We didn't talk to anyone at Tidewater. We didn't watch a job move through your shop. So some of this is going to be wrong, in specific and sometimes obvious ways — wrong about who handles what, wrong about which step actually causes the delay, maybe wrong about whether a thing is a problem at all.

That's fine. Being wrong in useful ways is part of the point. An X-ray is a fast, cheap first pass that gives you something to react to — "no, that's not it, it's actually the finishing schedule" is a genuinely useful sentence, and it's one you can only say once you've seen a guess on paper. If this were perfectly accurate we'd have needed three expert interviews to write it, and then it wouldn't be free or fast. The Map (step 1) is where we go find out what we got wrong.

A word on method, since it affects how much weight to put on any one line below. We read your site the way a prospective client would: what you say you do, who you say you do it for, and the words you use to describe your own process (your "shop drawings," your two-location structure, the language on your quote page). We combined that with whatever you told us directly in the ten-minute form — the parts of your business you already know are sore, in your own words. We did not look at your books, your CRM, or your actual job files, and we didn't talk to anyone who works there. Everything past this point is a hypothesis, stated plainly, so you can agree, disagree, or correct it in about the time it takes to read this page.

What we think your business looks like

Based on your site's description of "custom architectural millwork and cabinetry" for builders, architects, and designers, and a two-shop setup, here's our best guess at how a job moves through Tidewater, start to finish:

StageWho's probably involvedHand-off risk
Inquiry & quoteSales / estimatorVerbal or emailed scope, not always written down the same way twice
Drawings / shop drawingsDesign or drafting, then the client's architect for approvalApproval loop lives in email, easy to lose track of the latest version
ApprovalClient / architect, relayed by salesSign-off often arrives as a reply-all thread, not a clean record
Cut list & materialsEstimator or shop lead, from the approved drawingsRedone by hand from a drawing that may not be the final rev
Production (cutting, assembly)Shop floor, two locationsChange orders made after this point may not reach both shops
FinishingFinishing crew, possibly a separate step/locationWaits on production; schedule slips compound here
InstallInstall crew, on siteScheduling coordinated informally, subject to site readiness
Punch listInstall lead + client walkthroughItems tracked in notes or texts, not a shared list
InvoiceOffice / bookkeepingFinal numbers re-entered from the quote and any changes since

The pattern we'd expect at your size: one or two people carry a lot of this in their heads, and the handoffs between stages are informal — email, text, a conversation on the shop floor — rather than a system that shows the same picture to everyone. That's normal at ~40 people. It's also usually where the money leaks out.

Two shops is the detail that shapes most of what follows. A single-location shop of your size can often get away with informal handoffs, because everyone who needs to know something is within earshot. Split the floor across two buildings and the same informality stops being a quirk and starts being a tax: anything that used to travel by someone walking over and saying it out loud now has to travel some other way, and "some other way" is usually a text, a phone call, or an email that assumes the reader already has context they might not have. We'd bet a fair amount of what's below traces back to that split, more than to any one person's habits.

Where it's probably leaking

1. Change orders don't reliably reach the shop floor

What we saw: your intake answer for "what breaks most often" was "client changes something after drawings are approved and somehow the shop doesn't always get the memo — we've cut the wrong door style twice this year."

Why it likely costs you: two miscuts a year on custom millwork is real money — material, labor, and a delayed install. If each one runs a rework day plus rush materials, that's roughly $6,000–$12,000/yr (estimate, not measured), before counting the client-trust cost of a visible mistake.

What fixing it might look like: a one-page change-order form that, when filled out, automatically posts to whatever the shop already checks daily (a printed sheet, a shared doc, a Slack channel) — no new system to log into, just the existing one updated automatically.

Our guess, worth checking in the Map: this usually isn't a shop-floor attention problem, it's a routing problem. The person handling the client conversation writes the change down somewhere — but "somewhere" is often an email to the client, not anything the shop floor ever opens. If that's true here too, the fix is narrower and cheaper than it looks from the outside.

small AI fixConfidence: high

2. Estimating lives in one person's spreadsheet

What we saw: you told us "our estimator built the pricing spreadsheet years ago and it's kind of a black box at this point — nobody else really knows how it works."

Why it likely costs you: this isn't a weekly cost, it's a risk sitting on the books — if that person is out for two weeks, quoting probably stops or slows sharply. Hard to price as $/yr, but worth naming plainly: it's a bus-factor-of-one on a function your revenue depends on.

What fixing it might look like: not replacing the spreadsheet — documenting the rules of thumb inside it (a written, agent-assisted procedure) so a second person could run it in a pinch, and eventually a lightweight tool that encodes the same logic.

This is the kind of finding an X-ray can flag but not really size. We don't know from your website whether your estimator has been there two years or twenty, whether there's already an informal backup, or how often quoting actually gets backed up in practice. What we do know is that "one person, one spreadsheet, no documentation" is a pattern we see often enough at this size that it's worth a direct question rather than an assumption either way.

procedureConfidence: medium

3. Numbers get re-typed between estimating and your accounting system

What we saw: your intake mentioned "we basically retype the final quote numbers into [our accounting system] once a job is invoiced."

Why it likely costs you: re-keying invites transposition errors and eats real time. At a guess, 2–4 hours/week across estimating and office staff, which is roughly $5,000–$10,000/yr in labor (estimate) — plus whatever a wrong-number invoice costs you in credibility.

What fixing it might look like: a small script that turns your estimating spreadsheet's export into the import format your accounting system expects. This is exactly the kind of thing that can be built and handed to you in step 2, not just described.

script / connectorConfidence: high

4. Install scheduling runs on texts

What we saw: you wrote "install dates get juggled by text between the lead and whoever's free that week — works fine until two jobs collide."

Why it likely costs you: a double-booked crew or a wasted trip to a site that isn't ready probably costs you a half-day of crew time here and there. At a guess, 1–2 half-days/month, or roughly $4,000–$8,000/yr in idle labor and rescheduling (estimate).

What fixing it might look like: a shared install calendar that the office and both shops can see, with a simple rule for who owns a date once it's set. Low-tech, high-leverage — often closer to a procedure than a tool.

procedureConfidence: medium

5. Shop drawing approval loops through email attachments

What we saw: your site references working with architects and designers on custom pieces, and your intake noted "approvals come back as email replies, sometimes to an old version of the drawing."

Why it likely costs you: building to a stale drawing is one of the more expensive mistakes in this business — full rework, not just a patch. Even catching it before the cut, the back-and-forth to confirm "is this the current one?" probably costs 2–3 hours/week of someone's attention, on the order of $4,000–$7,000/yr (estimate).

What fixing it might look like: a single shared link per job for the current drawing revision, with old versions clearly marked superseded — a connector between whatever you draft in and whatever you send out, rather than a new tool to adopt.

connectorConfidence: medium

Three things worth looking at first

  1. Change orders reaching the shop floor. It's the finding with the clearest fix, the clearest cost (miscuts you can point to), and the least organizational friction to change.
  2. Re-typing between estimating and accounting. A script-shaped fix with a fast payback and no process change required from your people.
  3. The estimating bus-factor. Not urgent until it's very urgent. Worth a written procedure now, while the person who knows it is still here to ask.

What we're less sure about

Worth naming the guesses we're least confident in, so they don't quietly harden into fact by the time you finish reading. We don't actually know whether your two shops specialize (say, one for cabinetry, one for architectural millwork) or split work by capacity — that changes which handoffs matter most. We don't know whether "the shop" in your own language means the production floor, the finishing crew, or both, which matters for finding #1 above. And we're inferring the punch-list-by-text pattern almost entirely from base rates at your size and trade, not from anything specific you told us — treat that one as the weakest guess on this page.

What we'd ask in the Map

If you go to step 1, here's roughly what we'd want from your top three or four experts — probably your estimator, a shop lead from each location, and whoever runs install:

What happens next

This X-ray is step 0 — free, and about as deep as ten minutes of your time and a public website can get us. Step 1, the Map, is three one-hour interviews with your own experts over about a week, and it replaces guesses like the ones above with what actually happens on your floor. Step 2, the Action Map, adds a priced, prioritized list of roughly 50 improvements — with the top three already built. Right now, the first few Action Maps are free, in exchange for a frank debrief afterward. See the full ladder for all five steps.