Predicting next week

Every week you make two forecasts whether you admit it or not: how much to buy, and how many people to roster. Most restaurants make both on a hunch, and pay for the misses twice — food thrown away and hours that were not needed on the quiet weeks, ingredients that ran out and a room that suffered on the busy ones. A forecast is not a corporate exercise. It is a written-down version of the guess you were making anyway, and the only real difference is that a written one gets better every week.

🔢 A forecast in three lines

You do not need statistics for this, and the sophisticated methods buy a restaurant almost nothing. The whole thing is three inputs, on paper, for each day of next week.

An example: last year the same Friday did 380 covers. You are running about five percent above last year, which gives 399 — call it 400. There is nothing special happening, so 400 is the number. Write it down. On Saturday morning, write next to it what actually happened.

That last step is the entire method. A forecast you do not check against reality never improves and is genuinely worthless. A forecast you check every week teaches you your own restaurant faster than anything else in this course: within a couple of months you will know that your Fridays are predictable within a few covers and your Sundays are not, and that alone changes how you buy for each of them.

🛒 What it is actually for

The forecast earns its keep in two decisions, and both of them cost real money when they are wrong.

Ordering is the first. Buying to a number rather than to a feeling is what stops you throwing away produce on the weeks you over-bought and running out of the star dish on the weeks you under-bought — and running out is the more expensive of the two, because it sends away a customer who came for a specific thing. The detail of how to convert a covers forecast into an order, with par levels and shelf life, belongs to the cost control course; what this course provides is the number that goes into it.

Rostering is the second, and it is where the biggest money moves. Labour is roughly a third of your sales, as the financial reporting course showed, and it is scheduled days in advance against demand nobody wrote down. A forecast in covers converts directly into the hours a shift needs, which is exactly the input the employee scheduling and labour cost courses assume you have. Those two books tell you how to build the roster; this one tells you what to build it for.

There is a third use worth mentioning, though it is more occasional: deciding whether to take something on. A large catering job, a private hire, an extra service — the answer depends on what that day was going to do anyway, and the forecast is the only place that lives.

⚠️ Where forecasts lie

A forecast is useful and it is not a promise, and knowing where it breaks keeps you from trusting it at the wrong moments.

• Averages hide the extremes. Two Fridays of 300 and 500 average to 400, which is a number that never happens. Look at the spread, not just the middle.
• One-off weeks poison the history. A week distorted by an event or a closure should be marked as such, or next year you will forecast from it.
• Something changed. A new competitor, roadworks, a price rise, a chef leaving — after any of those, last year stops being a good guide until you have a few weeks of the new normal.
• Growth is not a straight line. A restaurant that has grown five percent for six months will not necessarily grow five percent forever, and forecasts that assume it do damage in the quiet season.
• Weather beats everything for a terrace. If that is you, the forecast is a starting point on Monday and gets revised on Thursday.
• Do not forecast further than you can act. Next week is useful because you can still change the order and the roster. Next quarter is the budgeting course's job, and a different exercise entirely.

The honest way to hold it is as a starting number that you adjust with what you know. The forecast is there so that your judgement starts from evidence instead of from the last busy night you remember — not to replace the judgement.

🧭 What you have built

Put the six topics together and the reports you were already paying for have become a working instrument. You found the four things every check records, and made sure the terminal was recording them honestly (topic 1). You mapped the real shape of your week and stopped confusing busy with profitable, and you made a deliberate decision about the day that does not pay (topic 2). You crossed sales against contribution and found that your best-seller and your best dish are different plates — the most profitable afternoon's work in this library (topic 3). You compared your channels with the same arithmetic and learned to judge each by the capacity it fills rather than by its share of sales (topic 4). You learned what a discount really costs and how to tell a promotion that made money from one that only made noise (topic 5). And you write down next week's number, then check it.

Where this sits in the finance run is worth restating. The previous course reads the statement and tells you the result got thinner; this one goes underneath and tells you it was the mix, or the channel, or the Tuesday. What comes next is the plan: turning all of this into a budget for the year ahead and comparing it month by month against what really happened. And after that, the other half of the equation — bringing the costs themselves down without damaging the thing people came for.