Six reporting mistakes that lead to bad decisions

A manager sees that sales fell by a tenth last week and cuts a server from the next fortnight's rota. Two weeks later, the room is struggling on its busiest nights, and it turns out the fall was a festival that landed in a different week last year. The figure was accurate. The decision was wrong, because the report compared the wrong weeks.

Most bad decisions made from restaurant figures are not caused by wrong numbers. They are caused by reporting mistakes: comparisons that do not match, totals that mix different things, averages that hide what happened and conclusions drawn from too little time. Each one is easy to make and easy to avoid once you know it.

This guide covers six reporting mistakes that lead to bad decisions, how to spot each one and the simple habits that prevent them. Every venue and figure in it is an example.

A report is only as good as the decision it supports. A small mistake in how figures are compared or added up can turn a good week into a bad one on paper, or the other way round, and the decisions that follow, about staff, menus, prices and marketing, carry the mistake into the real business.

The cost is rarely the mistake itself. It is the cut that was not needed, the promotion that seemed to work and did not, the dish removed because it looked unpopular in a week that was unusual. Avoiding the mistakes is cheaper than undoing the decisions.

The most common mistake is comparing periods that do not match. A week compared with the same dates last year may contain a different mix of days. A month compared with last month includes a different season. A week with a festival or a closure compared with one without it shows a swing that has nothing to do with how the venue was run.

The figures in each period are right; the comparison between them is not. And because the difference is presented as a percentage change, it looks precise and convincing.

The fix is to compare like with like: matching weeks, each starting on the same day of the week, the same day of the week within them, and festival weeks with festival weeks. Leave out weeks that were closed in either year, and note anything unusual, such as weather or a local event.

A fair year on year comparison takes a little more care than comparing dates, and it changes the conclusions surprisingly often. Our guide to year on year comparison that is actually fair covers the method in full.

The second mistake is mixing figures that include different things. Sales can be counted with or without tax, with or without service charge, before or after discounts, including or excluding voids and comps. A report that adds sales with tax to sales without, or compares a gross figure this year with a net figure last year, is comparing different things.

The difference can be large. Tax alone can add a fifth to the price in some places, and a change from one definition to another can look like a sudden rise or fall in trade.

Choose one definition of sales and keep it everywhere: in the daily report, the weekly review, the budget and the accounts. Most venues use sales before tax and service charge, after any discounts, because that is the money the business actually earns.

Write the definition down once, at the top of the reports, and check that every figure follows it. When a figure comes from a new source, such as a delivery platform or a new till, check what it includes before adding it to the rest.

The third mistake is averaging figures across services that behave differently. A day's average spend per head blends a quick lunch with a long dinner; a week's average ticket time blends a quiet Tuesday with a packed Saturday. The average describes neither.

Averages can even move in the opposite direction to every part of them. If both lunch and dinner spend per head rise, but more of the week's guests come at lunch, the overall average can fall. This is a form of Simpson's paradox, and it catches out experienced managers.

The fix is to split figures by service before averaging: lunch and dinner, weekday and weekend. Read each service's figures on their own, and look at the mix of services separately.

A report that shows spend per head for lunch and for dinner, with the number of covers in each, tells a manager what is happening. A report that shows only the blended average can tell them the opposite of the truth.

The fourth mistake is drawing conclusions from too little time. A single week's figures are full of noise: a rainy weekend, a big booking, a local event, a sick chef. A change that appears in one week may vanish in the next.

Decisions made on one week's figures are often decisions made on noise. A dish removed after one bad week, a price changed after one quiet weekend, a member of staff cut after one slow fortnight: each may be reacting to something that was never really there.

The fix is to wait for a pattern before acting. Look at four weeks or more, compared with the same weeks last year, before deciding that something has changed. Use rolling figures, such as the last four weeks against the same four weeks a year ago, to smooth out single-week swings.

Some things need acting on at once: a dish that makes guests ill, a supplier delivering short, a member of staff upsetting guests. But the figures in a report are rarely that urgent, and waiting for a pattern usually leads to better decisions.

The fifth mistake is measuring what is easy rather than what matters. Covers are easy; margin is harder. Likes are easy; bookings are harder. Ticket times are easy; whether guests felt the food came at the right pace is harder.

Reports drift towards the easy figures, because they are always available and they always move. A report full of easy figures can look thorough while saying little about whether the venue is doing well.

The fix is to start from the decision rather than the data. What do you need to decide, and what figure would tell you? A decision about a dish needs its margin as well as its sales. A decision about marketing needs bookings and covers, not likes.

If the figure that matters is hard to get, an estimate is often better than a precise figure that answers the wrong question. Mark it as an estimate, and improve it over time.

The sixth mistake is reporting figures without defining them. What is a cover: every guest, or only those who ate? Does a no-show include late cancellations? Is utilisation measured against opening hours or against the hours guests can be seated? Different people answer differently, and their figures do not match.

Without definitions, two managers can report different figures for the same night, and both be right by their own measure. The argument that follows is about the definitions, not the business, and it wastes everyone's time.

The fix is to define each figure once, in writing, and use that definition everywhere. A short page of definitions, kept with the reports, covers most of it: what counts as a cover, how sales are measured, what period a week covers, how a no-show is recorded.

Share it with everyone who prepares or reads the reports. New managers learn the venue's figures faster, and comparisons between months and between venues become meaningful.

A percentage change on a small number can look dramatic and mean nothing. A dish that sold four portions last week and eight this week has doubled, but four extra portions is not a trend. A no-show rate that jumps from two per cent to six may be three tables.

Always show the numbers behind a percentage when they are small. "No-shows rose from two tables to six" is far less alarming, and far more accurate, than "no-shows tripled".

Charts can mislead as easily as lists of figures. A chart whose scale starts well above zero makes a small change look enormous; one that mixes periods of different lengths makes a short week look like a collapse.

Start bar charts at zero, keep the periods the same length and label the scale clearly. A chart should show the shape of what happened, not exaggerate it.

Restaurant benchmarks, industry averages for food cost, labour cost, spend per head and the like, are a common source of bad decisions. They come from other venues, in other places, measured in other ways, and the definitions behind them are rarely clear.

Use them as background, not as targets. A venue whose labour cost is above an industry average may be well run for its style of service; one below it may be understaffed. Compare yourself with your own past first, and with benchmarks only when you know how they were measured.

Consolidated reporting, adding several venues' figures together, brings its own versions of these mistakes. Venues with different definitions, different currencies, different opening dates and different trades are added up as if they were the same.

The group's total can then move for reasons no single venue would recognise: a new venue opening, a venue closing for refurbishment, a currency rate moving. Our guide to consolidated reporting across four venues covers how to add venues together fairly.

These mistakes can creep into any report, from the daily note to the annual accounts. The best defence is a small set of reports, each with a clear purpose, each built the same way every time.

Our guide to restaurant reporting: six reports, no more covers the reports a venue needs and what each one is for. Fewer reports, kept consistent, leave less room for mistakes than many reports built in different ways.

A report built before the day's figures are complete is wrong from the start. Bills left open at close, payments recorded the next morning, bookings marked as no-shows only after a delay: each can change yesterday's figures after the report was written.

Build reports at the same time each day, once the figures for the period are complete, and note any figure that may still change. A daily report prepared in the morning, after the night's bills are settled, is more reliable than one built at midnight.

Having too many reports is a reporting mistake of its own. When a manager receives a dozen reports a week, the important figure is lost among the rest, different reports show slightly different versions of the same number and nobody is sure which one to trust.

Fewer reports, each with a clear purpose and a clear owner, leave less room for contradiction. If two reports show the same figure, keep the one that is read and retire the other.

Before acting on any surprising figure, ask three questions. Is the comparison like for like? Does this figure include the same things as the one I am comparing it with? Is this a pattern, or a single week?

If the answer to any of them is no, or not sure, check before deciding. Most surprising figures turn out to have a simple explanation, and most of those explanations are one of the six mistakes.

When a figure looks wrong, check the source records rather than the report. A bill left open, a payment entered twice, a booking recorded on the wrong day or a void entered as a sale can each distort a report.

Fix the mistake at the source, so that every report built from it is right. A figure corrected only in one report will be wrong again in the next.

Build a short review into the reporting routine. Once a month, look back at the decisions made from last month's figures and ask whether the figures turned out to be right. Over time, this shows which reports can be trusted and which mislead.

It also builds a culture where questioning a figure is normal. A team that asks "is that comparison fair?" before acting makes fewer bad decisions than one that takes every figure at face value.

A second pair of eyes catches most reporting mistakes. A manager who prepares the report and an owner or assistant who reads it with a few simple questions will spot the mismatched weeks and the mixed definitions that the person who built it no longer sees.

Make the check a habit rather than an accusation. The question "is this comparison like for like?" should be as routine as checking the cash.

When a mistake is found in a report that has already been shared, correct it openly. Send the corrected figure with a sentence explaining what was wrong and whether any decision should change.

Quietly fixing the figure, or hoping nobody notices, damages trust in every report that follows. A team that sees mistakes corrected openly learns that the reports are meant to be right, not to look good.

Sometimes the report says one thing and the team's experience says another: the figures show a quiet week, and everyone remembers being run off their feet. Take the disagreement seriously. Either the figures are missing something, such as walk-ins that were not recorded, or the team's memory is shaped by one busy night.

Look at the detail: covers by day and hour, the bookings, the busy nights. The answer usually lies in one of them, and finding it makes both the report and the team's judgement better.

An invented brasserie's weekly report shows spend per head down eight per cent on the same week last year, and the owner considers raising prices. Before acting, the manager checks. Last year's figure included service charge, and this year's does not. Last year's week had one more weekend night. And the brasserie's new weekday lunch has brought in many guests at a lower spend.

Matched by week, with the same definition of sales and split by service, dinner spend per head is up three per cent and lunch is new trade. The owner leaves prices alone, writes the definition of sales at the top of the report and splits every figure by service from then on.

servearound settles several of these definitions in how its reports work. Every report runs on the venue's own clock, so a day is the venue's day, and where a report compares, each headline figure is set against the period of the same length just before it. There is no comparison with last year, which is run by hand as a custom range. Utilisation is the time each table had a party against the service time its shifts offer, capped at 100%, and the kitchen's on-time rate is the share of tickets ready within fifteen minutes of being sent, one bar for every station.

In a group, money from venues trading in different currencies is added up in the network's own currency at the rate for each transaction's own date, so last month's total does not move with today's rate. servearound holds no industry benchmarks, and its reports are for one venue at a time, with the network dashboard for the group's totals. See Analytics.

Comparing periods that do not match, such as the same dates rather than matching weeks.

Adding or comparing figures that include different things, such as tax or service charge.

Because an average across two services can hide what happened in each, and can even move the opposite way.

Usually four or more, compared with the same weeks last year.

As background, not as targets, and only when you know how they were measured.

So everyone measures the same thing and figures can be compared.

Check the source records, and fix the mistake there.

Review last month's decisions against what the figures turned out to show.

Six reporting mistakes cause most bad decisions made from restaurant figures: comparing the wrong weeks, mixing gross and net, averaging two services, trusting one week of data, measuring what is easy and leaving figures undefined. Each one turns accurate numbers into a misleading picture, and each has a simple fix.

Compare like with like, keep one definition of sales, split figures by service, wait for a pattern before acting, start from the decision rather than the data and write each definition down once. Treat restaurant benchmarks as background, add venues together with care in consolidated reporting and keep restaurant reporting to a few consistent reports, so a fair year on year comparison is the habit rather than the exception.