How to use your own numbers without hiring an analyst

The problem is not data. It is attention.

Almost every restaurant already produces the data needed to answer its three biggest questions: what should I cook more of, when should I staff, and what is quietly losing me money.

What is missing is not collection. It is a fixed twenty minutes a week and a short enough list that the twenty minutes is enough.

The failure mode is predictable. An owner installs something with a dashboard, looks at it enthusiastically for a fortnight, finds forty metrics of which thirty-six are not actionable, and stops opening it. The dashboard was not wrong. It was too long.

This guide covers the short list, how to read each number, and what to do about it.

The short list

Weekly — four numbers, twenty minutes:

Monthly — add eight:

Food cost %, labour %, table turnover, top and bottom sellers, orders by hour, repeat guest rate, payment mix, net margin.

That is twelve numbers in total. The full definitions and calculation methods are in the twelve restaurant KPIs worth tracking; this guide is about what to do with them.

Reading your best sellers properly

The single most common analytics mistake in restaurants is looking at one best-seller list.

There are two, and they usually disagree:

• By quantity — how many plates left the kitchen. This tells you what to prep.
• By revenue — how many rupees each dish generated. This tells you what to protect.

A worked example from a mid-sized casual dining menu:

Masala papad is the top seller by a wide margin and seventh by revenue. Mutton biryani is eighth by volume and fourth by revenue.

Reading only the quantity list, you would prep more papad and consider cutting the biryani. That would be a mistake — the biryani is generating half again what the papad does from a quarter of the plates, and it is almost certainly a reason some guests chose you.

The rule: prep to the quantity list, price and protect to the revenue list, and only cut a dish that is low on both.

A third dimension makes this better still: contribution margin, which is revenue minus food cost per dish. A high-revenue dish with a terrible margin is doing less for you than it appears. Once you have food cost per dish — the method is in food cost percentage — you can rank by contribution rather than revenue, which is the version that actually decides menu changes.

Finding your real peak hour

Ask any restaurant team when the rush is and you will get a confident answer. Measure it and the answer is usually wrong by an hour.

Count orders placed, by hour, by day of week, over at least four weeks. Not covers, not revenue — orders placed, because that is what loads the kitchen and the floor. Four weeks because one week is distorted by weather, a festival or a single large booking.

What owners typically find:

• The peak is 45–90 minutes later than assumed.
• It holds longer than assumed, with a long tail rather than a sharp drop.
• Weekday and weekend peaks are different enough to need different rosters.
• There is a genuine lull that nobody had noticed, which is where prep or breaks should go.

Then compare that curve against your actual roster. The mismatch is usually visible immediately, and fixing it costs nothing — it is the same number of staff hours, moved.

This is the highest-return analysis on this page. It requires no new spending and it directly affects both service quality and labour cost. The operational side of acting on it is covered in how to run a service that holds together at 8pm.

Average order value, and why it beats covers

Most owners instinctively chase covers. AOV is usually the better lever, because a cover consumes table capacity and AOV does not.

A worked example. A restaurant doing 1,200 orders a month at ₹850 average:

• Current monthly sales: ₹10,20,000

Route A — 10% more covers. Requires 120 more orders a month, which needs either more table capacity, faster turnover, or more demand. New sales: ₹11,22,000.

Route B — 10% higher AOV, same covers. Requires ₹85 more per order — one starter, one dessert, or one upgraded main. New sales: ₹11,22,000.

Identical revenue. But Route A consumed 120 additional table-turns you may not have had, while Route B used none. And because your rent, utilities and most of your labour did not change in either case, both increases land largely in profit — Route B just did not need capacity you would have had to find.

AOV moves through three routes, in order of how easy they are:

1. Additions to an existing order — a second round, a dessert, a side. Cheapest, because the guest is already seated and already spending.
2. Trading up — the same guest choosing a higher-value version of what they wanted anyway.
3. Price — the fastest and most direct, covered in restaurant profit margins.

Route 1 is where most restaurants leave money. Second rounds do not happen if nobody comes back to ask, which makes it an operations problem as much as an analytics one.

Common mistakes

Comparing day against day. Tuesday against Monday tells you it is Tuesday. Compare a Saturday to Saturdays, a month to the same month last year.

Reacting to a single day. One quiet night is weather. Four is a pattern. Acting on one produces whiplash — changing the menu, the roster and the prices in the same fortnight, then not knowing which change did what.

Tracking numbers you will not act on. If you would take no action at any value of a metric, do not track it. It costs attention and buys nothing.

Cutting a dish on quantity alone. Covered above. It is how restaurants lose the item that made someone choose them.

Letting historical prices move. If your system recalculates last quarter's revenue when you change a price today, your trend lines are fiction. This is worth checking directly — the question is in the restaurant management software guide.

Confusing a dashboard with a decision. The value is not in looking. It is in the one thing you change afterwards.

Making it a habit

The habit matters more than the tooling. A restaurant reviewing four numbers on a fixed day beats one with a beautiful dashboard nobody opens.

1. Pick a day and time. Tuesday morning is common — the weekend is complete and there is a week ahead to act on it.
2. Four numbers, twenty minutes. Sales, prime cost, AOV, covers. Compare to the same period, not to yesterday.
3. Write down one action. Not five. One thing you will change this week.
4. Monthly, add the other eight and do a proper menu and roster conversation.
5. Check the previous action first. Did last week's change do anything? This is the step everyone skips, and it is the one that turns reviewing into learning.

The short version

• Four numbers weekly, twelve monthly. Longer lists get abandoned, and an abandoned dashboard is worse than a notebook.
• Best sellers by quantity and by revenue are different lists. Prep to one, protect to the other, cut only what is low on both.
• Measure orders by hour over four weeks. Your peak is probably later and longer than you think, and fixing the roster costs nothing.
• Average order value beats chasing covers, because it does not consume table capacity you may not have.
• Compare like with like, and never act on a single day.
• The review is worthless without an action. One change a week, checked the following week.