Why Elgin restaurant owners lose Sunday to the weekly numbers

It is Sunday morning and the dining room is dark. The owner is at the two top by the window with a laptop, because the office behind the walk in has no window and the signal is better out here. He is putting last week together.

Sales come out of the point of sale by day, and he splits lunch and dinner by hand because the report will not break it the way he wants. Labor comes out of the scheduling app in hours, so he multiplies. Two servers swapped a Friday and nobody moved it in the schedule, so he fixes that from memory. Then the delivery platforms, three of them, three logins, each with its own idea of where a week starts and what counts as a fee. Food cost is a folder of invoices from the produce guy, the meat guy, and the fish guy, and Thursday's has not been entered.

By eleven thirty he has labor as a percentage of sales. Thirty one point something. He is fairly sure. Last week's tab in the same workbook says twenty nine, and he cannot remember whether that one had payroll taxes in it. Then Tuesday starts and nobody opens the file again.

What the Sunday rebuild costs

Losing the morning is the cheap part. The expensive part is that the number arrives after the decision it was supposed to inform.

Labor is the clearest case. What an owner actually steers with is next week's schedule, and that schedule usually goes out before the last week has been counted. Friday ran three servers heavy. That was a fixable problem the following Wednesday. Read nine days later it describes two Fridays that are already spent, and the schedule for the third went out Thursday night.

Reporting also quits exactly when it matters. The weeks worth studying are the ones with the walk in cooler down, two call offs, and a Saturday that never landed, and those are the Sundays nobody sits down with a laptop. The definitions drift too. Do comps come off sales or sit below them. Does labor include the owner's own sixty hours. Assembled by hand, the answer changes with whoever is tired, and September stops being comparable to March.

Then there is what nobody else sees. A chef who gets food cost by category every week, next to what he actually ordered, holds it differently than one who hears about it in a tense conversation at month end. A general manager who knows on Thursday that labor is running hot can still change Saturday. Both are guessing, because the numbers live in one workbook on the laptop of the person with the least time to circulate anything. The quieter leaks never surface at all: voids by server, discounts nobody set, a delivery refund pulled out of a payout two weeks after the guest complained.

What it looks like when the week assembles itself

The hard part is agreeing on what you want to see. After that it is wiring, and the wiring has one job: put those numbers in front of you without anybody logging into anything.

• We pick a short list first, usually six to ten. Sales by day and daypart. Labor as a percentage of sales, with payroll taxes in or out, decided once and printed on the page. Cost of sales by category. Comps and voids with who and why. Delivery gross against what actually cleared the bank.
• Data gets pulled where the platform allows it. Square publishes open developer documentation, including a Labor API that carries timecards and scheduled shifts. Toast gates API credentials behind an integration partnership application and a signed agreement, so there the fast path is usually a scheduled export. Everywhere else, a recurring report email into a mailbox a system reads each morning is an unglamorous pipe, and it has never once gone down on me.
• It runs nightly instead of weekly, so the week gets counted while it is still happening. Monday morning you get one page: last week, the week before it, and what moved.
• A Thursday note says where labor sits against sales so far, while Friday and Saturday can still be changed.
• You set the lines that get to interrupt you. A cost of sales category crossing a number you pick, or comps running high two weeks running, sends a note that week.
• The chef gets the food and ordering slice, the general manager gets labor, voids, and comps, both on the same schedule and sent straight to them.
• History keeps itself, so this September sits next to last September and nobody maintains a workbook.

I have built this shape of thing in other industries. A mortgage professional I work with gets scored, early intent leads delivered on a schedule instead of digging through sources himself, and a pool service company's leads are tracked from the first call through the finished job. Same mechanics: take the data out of wherever it already sits, assemble it the same way every time, and send it to whoever has to make a call.

What stays human

Reading the number. A thirty four percent labor week with a cooler down and a new server in training is a different animal from a thirty four percent week when everything ran clean, and no system knows which one it is looking at. Choosing what to track stays yours too, because a report full of things you will never act on costs more attention than it returns.

Then every decision under it. Whether to cut the Tuesday dinner server or fix the reason Tuesday is slow. Whether the server who comps the most needs training or just needs somebody to ask him why. That is the job, and no page of numbers does it for you.

When this is not worth building

One room, one owner who works every service, a point of sale with a decent dashboard, and he can already tell you Friday's labor from memory. That owner does not need this, and I will say so in the first ten minutes. Neither does a place that has not settled what it wants to track. If nobody in the building can name the six numbers that matter, the first job is an hour and a legal pad, because sending out numbers nobody agreed on just puts the argument in your inbox every Monday.

When it is worth building, builds start at $500 one time, then a flat monthly from $99 after an included run in period, cancellable. More on the restaurant page and on what I build for businesses around Elgin.

Where to start

The first step is a free 20 minute process audit. We open your last four Sundays, walk every place a number came from, and count the logins. If assembling that week on its own would give you the morning back and show you Friday in time to change it, I will tell you what it costs to build. If it would not, you keep the map. Start here.