A Nibble For a Pint of Coke: The Exchange That Never Shows Up on Your Stock Report
A waiter gets fed by the kitchen. What the chefs get back doesn't show up as a wage, a tip, or a till transaction. It shows up in your post-mix deficiency instead.
Parts One and Two of this series covered how a variance report gets built and what a normal yield actually looks like, category by category. Spirits and post-mix are different from everything covered so far. These are the two areas where the numbers get genuinely sensitive, and where the first question worth asking often isn't "how much" but "where is it actually going?"
Spirits: You Shouldn't Really See Much Variance At All. But You Probably Will.
Spirits are one of the few categories where significant variance is genuinely unusual. Unlike draught beer, there's no head loss, and unlike food, there's no preparation waste. If measures are being used correctly, spirits should be one of the most accurate areas of the entire stocktake.
Over four weeks, a loss of 0.1 of a bottle (2.8 x 25ml shots on a standard 70cl bottle) isn't worth losing sleep over. That's noise, not a signal. It could be a little over-measuring at the bar, a bounceback from the previous count, a slight undercount this time, or the same cross-keying problem from Part One.
Counting is normally done to the nearest 0.05 of a bottle, so being out by that much on its own only represents about 1.4 shots, well within what a careful count can be out by. It's worth understanding bouncebacks. A counting error in one period often reappears as an opposite variance in the next. Undercount a bottle today and you'll create an apparent shortage. Correctly count it next time and you'll create an apparent surplus. Nothing changed in the bottle; the same error simply showed up twice.
A variance swinging positive one period and negative the next usually isn't two separate events. It's one misread figure showing up twice. The real test isn't the single figure, it's whether it repeats. A one-off 0.1 is exactly the kind of noise two slightly different reads produce, and isn't worth chasing. The same product showing a variance in the same direction, period after period, is what turns noise into a pattern worth investigating. Anything over 0.2 of a bottle (5.6 x 25ml shots) needs investigating. That's too much to explain away as counting noise.
When the products showing variance are the cocktail liqueurs, that's exactly where you need to focus your attention... Vanilla vodka, Kahlúa, Passoã, Cointreau and similar.
First, check the recipe specifications the variance was calculated against. If those are wrong, the variance itself may not be real. If the specs are right and the variance is still there, the issue is very likely with how cocktails are actually being made. Are the specs being followed at the bar, or are drinks being given away? That's where the investigation goes next.
Server reports can point you straight at it. If Passoã's down half a bottle, check who's sold the most of whatever uses it. That's often enough to narrow down where to look first.
Worth knowing one more mechanism specifically: a double poured but a single rung through the till. Different from cross-keying, the product's being poured correctly, the quantity hasn't been registered through your till. Different from a dash issue too, where a tall glass may be topped up and a regular sized dash is punched into the till. These show as real deficits, however there's nothing on the till pointing as to why.
None of this is complicated, and none of it is a trick. It's all about keeping your nose to the coalface and grinding out the same message, over and over:
Follow the specs, use the measures, don't free-pour. Picking up on wastage at dispense, on every product, every time. There's no shortcut past that.
The targets above only hold if the discipline behind them never lets up.
Post-mix: An Area That Can Throw All Sorts of Results Out
These products are genuinely sensitive. Coca-Cola dispenses at 5:1, one litre of syrup generates six litres of dispensed drink. That litre of syrup yields around 10.56 pints. Scaled up to a full 7-litre box, that's roughly 73.9 pints.
If ice is going in the glass, and the recipe calls for a pint to the rim, you should actually be seeing a surplus, not a loss. Ice takes up space in the glass and displaces the liquid, so a "pint to the rim" with ice in it uses less actual liquid than a full pint of the drink itself. The same mechanism works in reverse. Too little ice in the glass means more liquid is required to fill it to the same level, which erodes that expected surplus or turns it into a loss outright. It's worth looking into ice measures alongside dispense if a product that shouldn't be generating a loss is.
There's another mechanism specific to post-mix worth knowing: a vodka and Coke rung through as just the vodka, the splash never hit as its own line on the till. That shows up as a shortage on your post-mix even though the spirit sale looks perfectly legitimate.
If you're seeing a loss on post-mix, in my experience I would always look at staff consumption first. Are staff drinking it? Are they supposed to be drinking it?
This brings us back to the waiter and the kitchen. Nobody involved necessarily thinks of it as stealing, it's just how things have always been done on that site, but it's exactly the kind of staff consumption that quietly eats into post-mix figures without ever looking like an obvious theft.
If you have an arrangement like this, and it's all above board, get it posted into the waste report rather than left to sit as an unexplained deficiency. It won't undo the cost loss, but it turns a mystery into something accounted for, something that can actually be looked at and addressed on its own terms.
Keep on top of calibration, too. These products are genuinely sensitive to it. If a site that's normally been generating surpluses suddenly starts showing deficiencies, that shift in pattern is worth checking. A taste test will quickly show whether the mix is actually out, the fast, free first check before anything else. If it tastes wrong, then it's worth calling an engineer out to check the mix-to-water ratio properly, rather than assuming the cause is staff behaviour.
Dashes may also account for some loss, if the spirit trade is heavy enough. If a dash recipe is set at 140ml and it's being charged as a single dash, but what's actually going into a tall glass is a much bigger top-up of mixer for that same one-dash charge, that's a real loss sitting in the post-mix figures. It will usually take a genuinely high spirit trade before this makes a noticeable dent in the results, though, it's a real area of potential loss which isn't usually the first place to look.
How Big Does a Post-mix Loss Have To Be Before It Stops Being Any of The Above?
Take a product mixed at 5:1. One litre of syrup produces 6 litres of dispensed drink, which works out to 21.13 half-pint glasses at £3.00 a glass, £63.38 in retail value for every litre used.
If Coke is deficient by £300 at selling price that's roughly 99 glasses unaccounted for over the month. Broken down, that's a little over 3 glasses a day, every day.
Translated back into what's actually being counted, that's 4.68 litres of syrup, over half a box. No plausible counting error gets anywhere near that kind of variance. A cursory recheck of the count is obviously still worth doing, just to confirm, but a variance that size is never really going to be count issue. At that point you're dealing with a significant operational issue: unrecorded consumption, a calibration problem, a dispense fault or deliberate product loss. A variance of that size won't be explained by counting error alone.
The Thread Running Through Both Categories
Spirits and post-mix look like completely different categories, but they usually fail for the same reasons. In both cases, the stock system is only reporting a symptom. The real job is finding the cause.
The most valuable information in any variance report isn't the size of a single variance. It's repetition. One unusual result is worth noting. The same result, appearing month after month, is where the investigation starts.
I'm David Holden. I've spent 30 years working across hospitality audit and operations, helping hotels, pubs, restaurants and bars identify where profit is being lost and put practical controls in place to stop it.
The common thread through every category we've covered in this series is simple: stock doesn't go missing by magic. Every variance has a cause, and once you understand the maths, finding that cause becomes much easier.
If you're looking at recurring deficiencies and don't know where to start, I can help turn unexplained losses into actionable answers.
Comments