Restaurant Reports: Six Numbers, Each Changing One Decision

011. The hourly heat map: it redraws your shifts

The heat map shows revenue by weekday and hour. One glance tells you which cell is dark and which is empty. The answer is often surprising: Thursday evening beats Saturday, while Monday lunch is nearly dead.

That number changes two decisions. First, staffing: an empty hour does not need three waiters, and a peak hour is not covered by two. Second, automatic discounts: the rule goes exactly into the lightest cell (say, Monday–Thursday 15:00–17:00, 20% off coffee).

022. Average check — but sliced

A blanket average check is nearly useless: it blends the business lunch with the evening banquet. To be useful it has to be sliced — by channel (hall, QR, site, aggregator), by hour and by waiter.

The channel slice is revealing: a QR order is usually larger than one taken by a waiter. The reason is simple — photos, modifiers and an unhurried choice. If yours is the other way round, the menu has no photos or the modifiers aren't set up.

033. Table turnover: your hall's real capacity

Table turnover is how many times a table takes new guests in an evening. With 40 tables at 1.4, an evening sells 56 seatings; at 2.1 it sells 84. Same room, same rent, 50% difference.

The cheapest way to raise turnover is to cut waiting. Guests wait in three places: for the menu, for the waiter after calling, and for the bill. A QR removes all three: the menu opens instantly, the order goes straight to the till, and the guest closes the bill themselves.

The service-time report breaks it into stages: from opening the order to sending it to the kitchen, to ready, to served. The long stage is where the problem lives. More often than not it is not the kitchen but an order stuck in "not sent".

044. Menu engineering and 5. stop-list history

Menu engineering is the profit-versus-popularity matrix. It tells you which 5–7 dishes to remove and which to move up. For it to work, the dish card must carry a food cost.

Stop-list history answers a different question: what runs out and when. If plov ran out at 14:30 on 22 days in a month, that is a production question, not a menu one. Add a second kazan or revisit the portion size.

The two reports are most useful read together. A dish is popular and profitable but sells out every afternoon — that means you lose money daily. It is the easiest kind of loss to fix.

056. Order source: where your marketing actually works

Every order records where it came from: the table QR, the website, the iOS app, Android, Telegram, an operator or an aggregator. That number becomes the basis for splitting the marketing budget — an exact share instead of "I think Instagram is working".

Paired with promo codes the precision improves further: one code for Instagram, another for Telegram, a third for a blogger. Within a week you can see who actually brings orders.

The cancellation report belongs here too. If cancellations spike in one channel, the problem is in that channel: addresses arriving wrong, a lagging stop-list or too long a wait.

• Once a month: menu engineering plus stop-list history (to clean the menu)
• Once a week: the heat map and table turnover (for shifts and bookings)
• Daily: shift close-out, cancellations, call response time