Most drinks retailers I have spoken to can tell you last week’s turnover to the pound. Far fewer can tell you which two metres of shelving earned it, or what it cost them to hold the stock that sat next to it doing nothing.
Duty went up again on 1 February 2026, with all alcohol duty rates uprated in line with RPI at 3.66%, and the increase lands on the retailer before it lands on the shopper. Wages, card fees and insurance have all moved in the same direction. Shelf prices have not moved as fast, because customers in this category are unusually good at noticing when they do.
So the margin has to come from somewhere else. In practice, it comes from ordering better and losing less, and both of those are measurement problems before they are anything else.
The argument here is simple. A till is a data collection device that happens to take payments. If you are only pulling sales totals out of it, you are using a fraction of what it already knows. Below are five numbers worth watching weekly, why each one moves profit, and what a sensible trigger threshold looks like.
1. Gross Margin Return on Inventory Investment, Measured by Bay
Gross margin percentage is a comfortable metric. It is also close to useless on its own in drinks retail, because it treats a bottle that sells twice a week the same as one that sells twice a year.
GMROII fixes that. Divide annual gross margin by average inventory at cost. A result of 3.0 means every pound tied up in stock returned three pounds of margin over the year.
Run it at bay level rather than store level. That is the whole trick. A store-wide figure of 2.8 can hide a premium spirits bay running at 1.1 while the ready-to-drink fixture next to it runs at 6.4.
Once you see that, the conversation changes from “should we cut prices” to “should that bay be four facings smaller”. Most independents find at least one bay where the answer is yes.
Threshold worth setting: any bay under 2.0 gets reviewed at the next range change.
2. Days of Supply, Split by Category Velocity
Days of supply is current stock on hand divided by average daily units sold. Straightforward enough. The mistake is applying one target across the whole store.
Fast lines can run lean because you can reorder them weekly and the supplier will always have them. Slow premium lines cannot, because a stockout of a £75 single malt on a Friday in December costs you the sale and probably the customer’s next three visits.
I would suggest three bands. Core fast movers at 14 to 21 days. Mid-velocity at 30 to 45 days. Long-tail premium at 60 to 90, with a hard cap on how much capital sits in that band overall.
A liquor store POS system that tracks unit velocity by SKU and flags days-of-supply breaches automatically will surface a slow-moving £4,000 tranche of stock in seconds – platforms built for this niche, SantePOS among them, tend to include case-break handling and age-verification prompts that generic retail tills leave to workarounds.
The forecasting question is not unique to drinks, either. Larger operators have been pushing the same logic much further, as coverage of AI-driven supply chain forecasting in major retailers has shown, though the underlying discipline is the same at any size.
Category demand shifts add urgency. Spirits volumes have been soft across several developed markets, a pressure visible in Diageo’s recent share price recovery and the strategy behind it, and retailers carrying deep premium ranges have felt it in their holding costs.
3. Shrink Variance from Cycle Counts, Not Annual Stocktakes
An annual stocktake tells you that you lost £11,000. It does not tell you where, when, or how, which makes it an accounting exercise rather than a management one.
Cycle counting does. Count one category per week, compare the physical count against the system’s expected on-hand, and record the variance as a percentage of that category’s sales.
Alcohol is a specific target rather than a general one. The British Retail Consortium’s 2026 Crime Report recorded roughly 5.5 million detected shop theft incidents across the year, with organised groups concentrating on high-value, easily resold goods. Spirits sit near the top of that list almost everywhere.
Breakage, mis-scanned multipacks, damaged returns to supplier that were never credited, and case-break errors all show up in the same variance line. Splitting the causes is what makes the number actionable, and automated tracking makes the split possible at all – a point made well in Coruzant’s piece on how automated inventory systems reduce shrinkage and holding costs.
4. Void, No-Sale and Manual Discount Rates by Operator
Every till logs voids, no-sale drawer opens, price overrides and manual discounts, tagged to an operator. Aggregated across a month, these produce a behavioural fingerprint that is very hard to fake.
You are not looking for a smoking gun. You are looking for outliers. If four staff run a 0.8% void rate and one runs 4.1%, that is either a training gap or something else, and both are worth a conversation.
Set permission tiers so overrides above a threshold need a manager code. Then review the exception report fortnightly. Most systems built for licensed retail, including SantePOS and rivals such as KORONA POS and Lightspeed, will generate this without custom work.
5. The Price Realisation Gap
Your planned margin is what the buying sheet says. Your realised margin is what the till actually collected after discounts, staff purchases, multibuy mechanics, and the odd sympathetic rounding-down at the counter.
The gap between them is where a surprising amount of profit disappears.
Calculate it monthly per category: planned gross margin percentage minus achieved gross margin percentage. A gap of 0.5 points is normal noise. Three points is a promotion that is being applied more broadly than intended, or a multibuy that is cannibalising full-price single sales.
Volume growth without margin discipline is a trap the majors know well – even a strong quarter like BJ’s Wholesale’s 15.9% sales rise is scrutinised on margin as much as on top line. Small retailers should apply the same test.
Making the Numbers Land
Five metrics are manageable. Twenty is not, which is why most dashboard projects die.
A workable cadence: voids and discounts reviewed fortnightly, shrink variance weekly by rotating category, days of supply weekly, price realisation monthly, GMROII quarterly at range review.
Two practical warnings. First, none of this works if the product file is dirty. Duplicate SKUs, missing cost prices, and mis-set case quantities will corrupt every figure above, and cleaning the file is a genuinely tedious week of work that nobody wants to do. Do it anyway.
Second, do not buy software before you know which of these five you actually intend to act on. Plenty of operators pay for reporting they never open.
Conclusion
Protecting margin in drinks retail is rarely about a single dramatic decision. It is about noticing a bay returning 1.1 instead of 3.0, a category leaking 2% to shrink, or a promotion quietly running 3 points wide – and correcting each one before it compounds across a full trading year.
The data to spot all three is already sitting in the till. The only real question is whether anyone is reading it.
