The work · demonstration build
Margin X-Ray
Find out which jobs you actually made money on — your quotes, your timesheets, and your invoices finally agreeing with each other.
For contractors: the post-job autopsy that shows where the estimate and the build parted ways — flip the quiet "For contractors" switch inside the demo and the same engine reads a remodeler's year: estimates, time cards, draws. Demo data is a realistic fabricated sample — no client's real numbers on this site.
The challenge
The quote lives in one spreadsheet. The hours live somewhere else. The invoices live in
accounting. Nobody connects them — so nobody knows which of last quarter's jobs lost money.
The three files don't even call the job the same thing: the quote says
ACME BRKT-R2, the timesheet says Acme bracket job, and the invoice
just says Inv 4471.
The demo
It opens pre-loaded on sample data and never asks you for anything. If you ever do want to point it at your own exports, that happens entirely on your computer — your files never leave your machine, because the page has no server to send them to. And it's optional, always.
How it earns the numbers
Matching three files that share no common ID is the hard part, so the demo shows that work instead of hiding it. A "how we matched these" view lists every pairing it made — same job, three different names — and the two or three it wasn't sure about, waiting for a human to confirm or reassign. Flip one and the totals recompute in front of you. It also counts honestly: "41 of 47 matched automatically" means exactly that.
What it would cost
A Margin X-Ray for your shop — typically $2,500–$7,500; your exact price is fixed in writing after the look-under-the-hood session, and once it's in writing, it doesn't move.
That range holds when your data arrives in formats we've agreed on first: QuickBooks (Online or Desktop) invoice exports; quote logs in CSV or Excel we can map in one sitting; timesheets from QuickBooks Time, ClockShark, or an Excel time log. Something older or stranger — the 2009 Access database, paper time cards — gets a separately priced look-and-map step ($500–$1,500, flat) before the fixed quote, so the price never pretends your data is cleaner than it is.
How it's built
For the people who read this far on purpose: everything runs client-side in the browser — no server, no accounts, nothing leaves your machine. The matching is a few hundred lines of deterministic code: token normalization, string-similarity scoring, and date/amount corroboration, unit-tested, with every decision explainable on screen. No black boxes — a match either has a visible reason or it goes in the review pile. CSV parsing happens in-browser; money math is integer cents. The sample shop comes from the same seeded data generator as the other demos, and the generator is portfolio-grade code in its own right.
Working with something like this? Call or text me: [FILL: phone]