Measuring promotions

How to tell if a bar promotion actually worked

You ran the special. The room filled up. That is not yet evidence of anything, and this page is about why, and what to do instead.

A busy night is not proof

You ran two-for-one on Thursday. The room filled up by ten. The promotion worked, right?

Maybe. The trouble is you only got to run that Thursday once. The comparison you actually want is the version of that same night where you did nothing, and that version does not exist. You cannot go and look at it.

Fridays and Saturdays make this worse, because they were going to be busy anyway. Run a promotion on your strongest night and you will nearly always look like a genius. Run the same promotion on a quiet Tuesday and it may have done far more work for far less credit.

Then there is everything else that moves a door count and does not care about your special: the weather, a game on TV, a bad night at the place down the street, a festival in town, the first warm week of spring, a bank holiday, road works on your block. Any of those can swamp the thing you changed.

Without something to compare against, you are reading noise and calling it a result. That is how a special that never made money ends up on the calendar every week for two years.

What a baseline actually is

A baseline is a boring sentence: this is what this room normally does on this weekday, at this hour.

Not what the venue averages. Not what last month looked like. Thursday at 11pm is a different business from Thursday at 8pm, and your patio is a different business from your main floor. If the comparison is not specific to the weekday and the hour, it will hide the exact thing you are trying to see.

It also has to be several weeks deep. One normal Thursday is not a baseline, it is one data point, and one data point cannot tell you what normal looks like or how far normal bounces around on its own. You need enough ordinary Thursdays to see the spread. If your Thursdays already swing widely week to week, a promotion has to move the number a long way before you can honestly say it moved anything at all.

Here is the part everyone skips. You have to record the ordinary nights too. If you only measure the nights you did something special, then every night in your history is a promotion night and you have nothing to compare them against. The boring Tuesdays are not filler, they are the measuring stick. They are also the first thing that gets dropped when the room gets busy, which is why so few venues have one.

The costs that quietly eat the win

A promotion is rarely as cheap as the line item it shows up as. Four costs, roughly in order of how easy they are to forget.

The margin you gave away

Two-for-one is not a one-off marketing spend. It is a discount on every unit that went out the door under the offer, including all the ones that would have sold at full price. Multiply the discount by the units sold, not by the new faces you think it brought.

The extra staff

Busier means another bartender, another barback, maybe another person on the door. That cost is real, it is loaded with payroll tax, and it almost never sits anywhere near the promotion in your books.

The people who were coming anyway

Your regulars did not need the offer. They walked in, took the discount, and cost you margin on a visit you already had. That is not a failure, it is just a cost, and it belongs in the arithmetic like any other.

What you paid for the draw

The DJ fee, the flyer, the boosted post, the promoter cut. This is the one operators do remember, and on most nights it is the smallest number on this list.

What "it worked" should actually mean

Most people measure a promotion in bodies. Bodies are easy to see and easy to count, which is exactly why they are the wrong finish line. A packed room of people nursing one discounted drink each is a worse night than a half-full room buying at full price, and the packed room feels much better while it is happening.

The number that pays rent is what an hour of your room earns while you have it open. Take what the night sold, and put it against the capacity-hours you actually had running: how much room you had open, for how long. That is sales per capacity-hour, and it is the honest scoreboard, because it charges you for the space you opened whether or not you filled it.

So the working definition is this. A promotion worked if, compared with a normal night of the same weekday and hour, it beat that night on sales per capacity-hour by more than the extra labor, the extra product and the draw fee cost you. Headcount is a useful clue on the way to that answer. It is not the answer.

The arithmetic, written out

You can run this on paper tonight. None of it needs RollCall. RollCall does the tedious half, which is keeping the history.

  1. Baseline crowd. Average headcount for that weekday, at that hour, across several ordinary weeks of the same weekday.
  2. Extra people. Headcount on the promotion night minus the baseline. If that difference sits inside the normal week-to-week swing, stop here. You have not measured anything yet.
  3. Baseline sales per capacity-hour. What a normal night of that weekday earns per hour of open room.
  4. Promotion sales per capacity-hour. The same measure, on the promotion night.
  5. Revenue lift. Step 4 minus step 3, multiplied by the capacity-hours you were open that night. This figure already has the discount inside it, because the discounted drinks rang up at the discounted price.
  6. The costs the till does not show. The extra labor you scheduled, the cost of the extra product you poured, and whatever you paid for the draw.
  7. The verdict. Step 5 minus step 6. Positive, and the promotion paid for itself. Negative, and the room was busy and you paid for the privilege.

If you would rather not do steps 5 through 7 by hand, we keep a free promotion break-even calculator that works out how many extra people a discount has to pull before it pays for itself. It runs in your browser, needs no account, and nothing you type into it is sent anywhere.

One thing to watch. If you are working from headcount alone, with no sales figures, then the discount is not inside your number and you have to subtract it yourself. That is the version most operators end up doing, and it is exactly the version where the margin you gave away is easiest to leave out.

And one caution worth being blunt about: none of this proves cause. It tells you the promotion night beat a comparable normal night by more than it cost. That is the most an operator can honestly claim without being able to run the same night twice, and it is enough to decide whether to run it again.

Where RollCall fits, and where it does not

The hard part of the method above is not the maths. It is having a record of the ordinary nights months before you need it. Here is exactly which piece does what, and what plan it is on.

Deals Pro

Post the offer: cover, drinks, food, an event or a perk, live now or scheduled. Happy hour can repeat weekly by itself, so the thing you are measuring goes out the same way every week instead of whenever someone remembers.

Crowd history Pro

Your crowd, built from the scans your staff post, sits alongside tonight and the live room: the last two weeks at a glance, and the full history over any range, which is what a real baseline is built from.

Busy heatmap Pro

Average crowd by night and hour, in your local time. This is the weekday-and-hour view the comparison actually needs, rather than a monthly average that hides everything.

Deal performance Pro

Every deal ranked against a normal night, with the best hour to post, learned from your own history rather than from someone else's venue.

Capacity and utilization Pro

Scheduled capacity-hours, how much of them you actually used, and where the room you are paying for sat empty. This is the denominator in the arithmetic above.

Sales per capacity-hour Pro

What an hour of your room earns when it is open, so you can tell a busy night from a good one. Needs your POS connected read-only on Pro.

The honest limits

Your first promotion has nothing to compare against. That is not something we can design around, it is arithmetic. A baseline is made of ordinary nights, and on day one you have not recorded any. So the first few weeks of RollCall will not tell you whether your happy hour works. They will tell you what a normal Thursday looks like, and that is the asset. The promotion answers come after it exists.

The history is only as good as your scanning. It is built from scans your staff post. If nobody scans on the quiet nights, your record is made entirely of the nights you cared about, and you are back to having no baseline. Scanning the boring nights is the actual work.

RollCall cannot see your costs. It can see how full the room was, and with your POS connected (read-only) on Pro it can see what the night sold. It has no idea what your pour cost is, what you pay your bar staff, or what the DJ charged. Those numbers you bring to the arithmetic yourself.

RollCall also publishes original operating stories for people running hospitality businesses.

Questions operators actually ask

Straight answers, including the ones that are not flattering.

How do I fill slow nights at my bar?

Start from the number the night has to beat, not from the idea for the promotion. Find your quietest tracked slot, run one offer there for several weeks running, and judge it on what an hour of open room earned against a normal night of that weekday, minus the discount, the extra staff, and whatever you paid for the draw. A packed room holding one discounted drink each can earn less than a half-full room at full price. Our free promotion break-even calculator works out how many extra people an offer must pull before it pays for itself, and the dashboard keeps the crowd history that makes the comparison honest.

How do I know if my happy hour is actually making money?

Compare it against what that weekday and hour normally does for you, not against an empty room. Work out what an hour of your open room earned during happy hour, do the same for the same weekday and hour across several ordinary weeks, and take the difference. Then subtract the extra labor you scheduled and the cost of the extra product you poured. If what is left is positive, it is paying for itself.

How many weeks of history do I need before a comparison means anything?

Enough ordinary weeks of the same weekday that you can see how much that night swings on its own. One normal Thursday is a single data point, not a baseline. If your Thursdays vary a lot week to week, you need more of them before a promotion night can stand out from the noise.

Did the DJ bring people in, or was it just a Friday?

You cannot know from one Friday. Compare that Friday against your other Fridays at the same hours, and check whether the difference is bigger than the normal spread between your Fridays. If it is not, the honest answer is that you do not know yet. Booking the same act on a slower night is usually the faster way to find out.

Is headcount a good way to measure whether a promotion worked?

It is the easiest thing to see and the easiest thing to be fooled by. A full room of people holding one discounted drink each can earn less than a half-full room paying full price. Headcount tells you the promotion pulled bodies. Sales measured against the hours you were open tell you whether the bodies were worth what they cost.

Can RollCall measure my first promotion?

No, and we would rather say so. A baseline is made of ordinary nights, so on day one there is nothing to compare against. Start scanning your normal nights first. Once you have several weeks of the same weekday recorded, a promotion night has something honest to be measured against.

What does RollCall need from me to measure a deal?

Scans posted by your staff on ordinary nights as well as big ones, the deal posted in RollCall so its timing is recorded, and for the money side, your POS connected read-only. Posting deals, staff scanning, crowd history, deal performance and sales per capacity-hour are all included with every RollCall plan.

Can I do this without any software?

Yes. Write down your headcount at the same two or three times every night for a few weeks, keep your nightly sales next to it, and you have a rough baseline. It is tedious and it is the first thing that gets dropped on a busy shift, which is the actual reason most venues do not have one.

You cannot measure a promotion
you have no baseline for.

RollCall Pro covers one venue. Start scanning the ordinary nights now, so the next promotion has something honest to be compared against.

Start now Questions? Contact us