Bar stock has a combination of properties that makes it harder to control than almost any other retail inventory.
It is high value per unit. It is consumed in measures rather than units, so a bottle is rarely a whole number. It is poured by hand under time pressure. It is portable and desirable. And it is counted at the end of a shift by someone who has been working since evening.
Any one of those is manageable. Together they mean that a bar which does not measure variance deliberately will not notice it — the loss distributes itself across hundreds of pours and appears as slightly lower margin, which gets attributed to pricing.
What variance actually consists of
Before assuming theft, the categories worth separating:
Over-pouring. The largest category in most venues, and rarely deliberate. A free-poured measure varies by a meaningful percentage, and across a busy night that compounds into bottles.
Spillage and wastage. Spills, returns, a bottle broken, a keg line purged.
Unrecorded staff and management drinks. Legitimate and frequently uncounted.
Comps and remakes. A drink returned and remade is two pours against one sale.
Billing errors. Items served and not entered, particularly during a rush, which is a till problem rather than a stock one — what breaks at peak hour.
Theft. Real, and usually a smaller share than the first two.
The reason to separate them is that they have entirely different remedies, and a venue that treats all variance as theft addresses the smallest category while the largest continues.
Counting in a way that produces a usable number
A monthly count tells you that something happened over a month. It cannot tell you which shift, which bar, or which product — which makes it nearly useless for action.
What produces action:
Count by shift, not by month, at least for the high-value lines. A number attributable to a shift is a number someone can explain.
Count consistently. The same method, the same time, the same units. Weighing partial bottles is more accurate than estimating, and the accuracy matters because the whole exercise is a difference between two numbers.
Count a short list frequently rather than everything occasionally. The top twenty products by value usually account for most of the exposure. A daily count of twenty lines is achievable; a daily count of four hundred is not, and one that is attempted badly produces worse data than not counting.
Compare against what was sold, which requires the till and the stock record to be the same system. Where they are separate, the reconciliation is manual and it will lapse — the pattern described in why multi-outlet businesses lose money between systems.
The general discipline is the same as any inventory operation, and it is set out in what actually prevents stock variance.
What multiplies at the second venue
A single bar is manageable by presence. The owner is there, notices things, and knows the staff.
The second venue breaks that, and it breaks it in specific ways:
Comparison becomes impossible without common definitions. If two venues record measures differently, or use different product codes, their variance figures cannot be compared — and comparison across venues is the single most powerful diagnostic available, because it isolates what is venue-specific.
Stock moves between venues. A transfer that is recorded at one end and not the other creates a shortage in one place and a surplus in another, and both look like variance.
Pricing and promotions diverge. Manageable, provided it is deliberate rather than accidental.
Staff move between venues. Which is useful operationally and requires that the system works the same way in each, or every transfer costs retraining.
Nobody is present everywhere. The owner's attention, which was the control mechanism, no longer covers the operation. Something has to replace it, and that something is a reliable daily number rather than a monthly report.
Staff and shift, which is the other half
Labour is the second-largest controllable cost in a bar, and it is usually planned from memory.
The numbers that make rostering rational are ones most venues already generate and do not use: sales by hour and by day, over several weeks. That pattern is stable enough to roster against, and it typically reveals that staffing is uniform across hours that are not.
What is worth recording: hours actually worked rather than scheduled, sales per labour hour by shift, and which shifts run over. These are the inputs to a decision that is otherwise made by feel.
What to put in place, in order
One: one system for till and stock, so sold and consumed are comparable without manual reconciliation. Two: a short daily count of high-value lines, by venue and shift. Three: common product definitions and measures across venues. Four: transfers recorded at both ends. Five: one figure per venue per day, visible to whoever can act on it — the discipline in operational dashboards that people actually use.
Truffaire has built inventory, billing and multi-outlet operations systems across ten client deployments, including a multi-ERP system for a retail chain of over thirty outlets. The multi-venue problems above are the ones that recur; we have not published outcome figures for those engagements and will not estimate them here.
Frequently asked questions
Should we use measured pourers?
They reduce the largest variance category directly. The trade-off is speed and the feel of service, which matters in some venues more than others. Measuring first tells you whether the problem is large enough to justify it.
How often should we count?
High-value lines daily or per shift; the full range weekly or monthly. Frequency on a short list beats completeness on a long one.
What variance is normal?
It varies enough by venue type and service style that published benchmarks mislead. Your own trend is the meaningful comparison — establish a baseline, then watch movement.
Do we need one system across venues?
You need common definitions across venues. One system is the simplest way to guarantee that, and separate systems with a manual reconciliation step reliably drift.
How do we handle staff drinks and comps?
Record them as a category rather than absorbing them into variance. Unrecorded legitimate consumption is why so many venues conclude they have a theft problem.
Where to start
Pick your five highest-value products and count them at the end of every shift for two weeks, against what the till says was sold.
The pattern in those ten lines will tell you whether you have a pouring problem, a recording problem or a specific shift problem — and it costs ten minutes a night to find out.
The wider question of how much of this is sector-specific and how much is ordinary operations software is in choosing technology for your sector.
If you want multi-venue operations run on one set of numbers, get in touch.