Restaurant Software Features: the 7 That Control Costs

Why this order and not the one on the sales sheet?

The order of this list follows the DEPENDENCY between features: each one needs the previous one to deliver a reliable number, which is why the recipe comes first and the accounting export closes the list.

Sales sheets rank them the other way around, because they sell what looks good in a demo, and the demo almost always opens on the point-of-sale screen. Nearly half of operators (48 %) prioritize investing in point-of-sale systems, according to the National Restaurant Association (2024), a reasonable decision that leaves the cost question untouched.

At Masterestaurant we apply this criterion when we review a client's software: if a feature at the top fails, everything hanging from it inherits the error, and a margin alert calculated on stale recipes only produces well-formatted noise. There are seven features, set in the order in which they hold each other up, and none of them gets skipped.

Recipes with live costing: the feature everything hangs from

A recipe with live costing is a spec sheet that recalculates the cost of the dish every time a new supplier invoice comes in, without anyone opening a spreadsheet to update it. Without that link the system reports two-month-old margins with the confidence of today's data, and you set prices on an old snapshot.

For example, if beef goes from 10 to 12 dollars a kilo and the dish uses 200 grams, the portion rises from 2 to 2.40 dollars, and on a dish with 8 dollars of net sales the food cost jumps from 25 % to 30 % without a single report flagging it. Two more increases like that and the dish is already past the 32 % ceiling the method sets as a MAXIMUM, never as a target.

It pays off from the first location with a fixed menu; a caterer quoting event by event needs it far less.

Theoretical versus actual inventory: where the money the register never sees goes

Theoretical versus actual inventory compares what should have been used according to sales and recipes with what the physical count shows, and the gap between them is waste, generous portions, badly written recipes or theft. No point of sale shows that gap on its own, because the ticket records the dish sold, not the grams that left the walk-in.

Some 30 % of surveyed restaurants already use AI to manage inventory, according to FSR Magazine (2026), and that AI only helps if the count exists and happens with weekly discipline. For example, if the week's theoretical usage says 40 kilos of chicken and the count shows 46, those 6 kilos without a ticket are product you paid for and never charged.

The useful version of this feature splits the gap by ingredient and by shift, so the manager knows which station to check before accusing anyone.

Daily prime cost: the number the board should read before sales

Prime cost adds food and beverage cost to total labor cost, and reading it daily, instead of at month-end close, lets you correct the week while it is still underway. I got this wrong for years: I asked clients for a monthly prime cost and always arrived late, with the damage already on the books.

The labor side is usually handled in another system, because FSR Magazine reports that 54 % of those who run restaurants automate payroll with software; what remains is crossing that payroll with the day's theoretical usage into a single figure. Two things that often get confused need separating.

Payroll goes into prime cost, but it is not loaded onto the dish's spec sheet, because wages, rent and utilities belong to the break-even point, and mixing them inflates unit cost until it pushes price increases nobody needed to make.

Margin per dish: menu engineering with this week's numbers

Margin per dish shows how much each item leaves after its direct cost, and crossed with popularity it becomes the basis of menu engineering: what gets pushed, what gets redesigned and what leaves the menu, or gets repriced. A sales report ranks dishes by what they bill, and that list misleads, because the best seller can be the one contributing the least contribution margin per unit.

For example, a dish that goes out 300 times a month with 3 dollars of margin leaves 900, while another with 120 orders and 9 dollars of margin leaves 1,080, and it is almost always the second one that servers recommend least.

The feature worth paying for calculates that margin with the recipe's live cost, never with a cost typed in by hand the day the restaurant opened, and puts it in front of the manager without anyone having to ask.

Cost by channel: the dish that wins in the dining room and loses in the app

Cost by channel separates the profitability of the same dish depending on whether it sells in the dining room, as takeout or through a delivery platform, because commission, packaging and the beefed-up portion change the margin radically. The urgency comes from adoption: 68 % of operators automate online ordering with their software (FSR Magazine, 2026), and most of them measure those orders as sales, not as margin.

Think about what would happen if a three-location group pushes its signature burger on the app with a promotion: volume rises, the kitchen chokes at peak hour, the dining room waits longer and, at close, commission and packaging have eaten what the dish used to leave at the table. The group worked harder to earn less, and nobody noticed because the dashboard was celebrating orders. Software without cost by channel hides exactly that math from you.

Alerts and accounting export: the machine warns and the accountant stops re-keying

Alerts warn you when an indicator crosses a threshold, such as an ingredient getting more expensive or a dish breaking the food cost ceiling, and the accounting export sends those numbers to the accountant's system without re-keying them by hand. They come last because they are the easiest to sell and the ones worth least without a foundation underneath. The FSR Magazine report shows 52 % already automate invoicing, which covers half the road toward bookkeeping without double entry.

Artificial intelligence fits right here, and the National Restaurant Association puts at 28 % the operators planning to invest in technologies such as AI during 2025; an agent that reads prime cost every morning and flags the deviation saves the manager hours. But if the recipes lie, the alert repeats the lie faster.

Which feature should you prioritize if you can only tackle one?

If you can only tackle one feature, start with recipes with live costing, because without them theoretical inventory, margin per dish and alerts all calculate on prices that no longer exist.

It is the recommendation Diego F. Parra holds to in software reviews under the Masterestaurant method, and there is a practical reason: a costed recipe can be built in a week, with the menu in hand and the latest invoices, while inventory needs several weeks of counts before it stabilizes.

The paradox is that owners usually want the AI dashboard first, which is the last thing that works, and the way out is simple: the dashboard comes almost free once the recipe tells the truth. Open your three best-selling dishes today, compare the spec-sheet cost with the latest invoice for each ingredient and write down the difference in money, dish by dish.

The 7 features in dependency order: who needs each one and who can wait

1. Live recipe costing. Everything else hangs on this, because a standard recipe that never updates when the supplier raises prices turns every later report into old news. For example, if oil goes up on Tuesday's invoice and the recipe card keeps the price from two months ago, the system will keep swearing your dish sits under the 32% food cost ceiling the method treats as a MAXIMUM, not a target.

Who needs it: any restaurant with a fixed menu, from day one. Who can skip it: a caterer who prices every event from scratch. 2. Theoretical vs actual inventory. Here is the money nobody sees at the register: theoretical is what should have been used according to sales and recipes, actual is what the count says, and the gap is waste, heavy portioning or leakage.

The 7 features in dependency order: who needs each one and who can wait — in practice

FSR Magazine (2026) reports AI-driven inventory management at 30% of surveyed US restaurants, and I would add a condition before copying them, since without disciplined counts the algorithm learns from dirty numbers. It pays off for teams that count proteins and liquor weekly; it gets in the way where recipes are not yet costed. 3. Daily prime cost. Food and beverage cost plus labor, read every morning rather than at month end, when nothing is left to fix.

The most repeated conceptual error deserves a name: payroll belongs in period prime cost and break-even, and it is never spread into the cost of a plate. Operations with variable shifts need it daily, because a slow Tuesday with a full crew eats the whole week; an owner-operator with two fixed employees can review it weekly. 4. Dish-level margin.

Before cutting a dish with a high percentage, look at the dollars it leaves, and that tension between percent and contribution is settled by menu engineering, not instinct.

The 7 features in dependency order: who needs each one and who can wait — key points

For example, a $10 dish costing $2.40 leaves $7.60, while a $16 dish costing $4.80 leaves $11.20: the second has the worse percentage (30%) and still pays more rent, and both respect the ceiling. Built for long menus of twenty-plus items; a five-dish menu can live in a spreadsheet. 5. Cost by channel.

The same dish does not cost the same in the dining room and on delivery, since packaging, platform commission and promotions change the math before the food leaves. Digital ordering is already routine, with 68% of US operators automating online orders through software (FSR Magazine, 2026). If you sell through apps several nights a week, you need it now; if takeout is marginal, park it and protect the dining room margin. 6. Alerts.

A dashboard you have to open is a dashboard you forget, which is why alerts change a manager's behavior more than any polished report.

The 7 features in dependency order: who needs each one and who can wait — examples and figures

This is where financial intelligence with AI comes in: a cost assistant that flags an ingredient priced above the agreed rate or an inventory gap that is widening, and explains why in plain words. The National Restaurant Association (2025) put the share planning to invest in technologies like AI at 28%; everyone else should wait until features one to three are clean, because alerts on bad data train the team to ignore them. 7. Accounting export.

It goes last because it closes the loop, not first, as it is usually sold. When purchases, sales and inventory reach accounting without re-keying, the monthly P&L matches what the kitchen lived through, and the board debates decisions instead of debating which number is right. Essential for groups of two or more units or partners demanding a monthly close; optional for a single unit whose accountant works fine from the invoices.