Your POS is a data goldmine: what a bar owner should pull every week (and ignore)
By Jack Hawkins · Updated
A modern till records every sale with a timestamp, a product, a person and a discount code attached, and then presents it as a summary that may not answer the next operating question. The fix isn't more reports; it's six specific pulls, done the same evening each week, that answer questions you can act on before the trade changes without you. Here's the working set we run venues through, and the two reports everyone checks that deserve to be ignored.
Before the list, the unglamorous prerequisite: consistency. Your menu items need stable IDs (keep a product-ID mapping when names or recipe versions change), discounts need reason codes, and staff need to actually ring comps. Without those three, every report below is a rumour with decimals.
Pull one: contribution league table, not a bestseller list
Sort products by (net price − ingredient cost) × units sold, over the last 28 days. Contribution pounds, not volume, not percentage.
This one table replaces the three arguments most weekly meetings waste time on. The bestseller might be your third-worst earner per serve; the "slow" craft pour might be top-five contribution and just needs to be visible on the menu. If your POS can't export item-level sales to a sheet you can cost against, this pull alone justifies finding a tool that can. Our menu engineering matrix does the join for you.
Pull two: the daypart curve against your rota
Hourly net sales, four weeks deep, same weekday columns side by side. You're not looking for the peak (you live there). You're looking for the shoulders: the two hours before peak where you're fully staffed and empty, and the two hours after last orders' worth of demand where you're closing lights while someone finishes a table.
Put the rota next to it. If labour coverage is flat while the curve isn't, you're paying for hours that produce nothing and thinning the two hours that produce most. Calculate paid-hour alternatives and check service outcomes before assuming the comparison will save a whole shift.
Pull three: discount, void and comp value with names attached
Total value, by reason code, by staff member, by hour. This is not a surveillance exercise; it's arithmetic with a paper trail. A venue running 4% discounts has a different problem than one running 4% with £2,900 of it in unexplained "manager" voids between 23:00 and 02:00. The first is a policy. The second is the variance conversation arriving early, which is the good version of the conversation.
Pull four: mix shift, this week versus the same week last quarter
Top twenty products by units, ranked, both weeks. Then just look at what moved more than five rank places. Menu drift is invisible week-to-week and obvious quarter-to-quarter: the tequila programme you launched is either genuinely eating gin's lunch or it was a fortnight of weather. Decisions about the next menu should be made on the shift, not on how much the brand ambassador talked.
Pull five: the dead-list audit
Every item with fewer than one sale per day for fourteen days. Each one costs a slot on your menu (attention is a finite resource per table), and behind the bar it costs stock depth, glass space, or a keg fading in the back. Three columns decide each one: contribution per serve, whether it anchors a category you claim to have, and when it last made anyone happy. A focused review may identify items to rebuild, retain for a deliberate role or remove, and the remaining list gets better purely by getting shorter.
Pull six: the Friday-to-Sunday share
Wet sales, weekend versus weekday, as a rolling eight-week percentage. This number is quietly the boss of the others. A weekend share above 60% means your cashflow, staffing and buying all hang on three nights, and a disrupted busy weekend can materially affect the period, depending on contribution and fixed costs. Venues that know their share make better decisions about weekday offers (or deliberately refuse them) than venues that don't.
What to ignore
The all-time charts. Long historical series can show seasonality and structural changes, but a next-service decision also needs a recent matched comparison. Everything that matters is last-28-days versus prior-28-days or same-week-last-quarter. If a report can't be phrased as a comparison, it's decoration.
Revenue-only dashboards. Revenue up 6% with contribution down 4% is a bad week that feels like a good one, usually because a low-margin category carried it. Every report worth opening has cost or contribution attached, or it lies politely.
The honest limit of what your till knows
The POS knows what was rung, not what was poured, not what was given, and not what was drunk by people who never got rung in. It knows a Margarita sold at 23:41; it can record a double, modifier or comp when correctly configured and rung; it cannot prove the actual physical pour. Which is why every pull above is a starting question, and why the stock cycle remains the lie detector. POS for behaviour, stock for physics; venues that only run one of the two are always slightly confused.
For a practical weekly cadence, use Better Bartender's weekly review workspace to bring a sales summary, recipe costs and a stock reconciliation together. Our bar sales analytics page explains the invitation-only pilot: permitted POS access is tested per venue and provider edition, rather than assumed to work universally.
Match the six pulls to actual inputs
Sales exports provide revenue, units and timestamps where the provider exposes those fields. Recipe costs supply contribution; paid hours and loaded wage costs supply the labour overlay; physical counts, receipts and waste supply variance. Join them with stable product and venue identifiers and a documented business-day cutoff. Midnight is not always the end of a bar service, so compare trading nights consistently. Reconcile refunds, modifiers and authorised comps before ranking staff or products. Label a missing input as unavailable rather than filling it with zero: a zero cost and an unknown cost produce very different decisions.
FAQ
What reports should a bar owner check weekly? Six pulls: contribution league table, daypart curve versus rota, discount/void/comp value by reason and staff, matched-period mix shift, dead-list audit, weekend share. Ten minutes each, same evening weekly.
How do I see which cocktails make the most money? Rank by (net price − ingredient cost) × units sold, not by units sold. A bestseller list ranks popularity; a contribution table ranks money, and the two lists disagree in almost every venue.
Can I export POS data to a spreadsheet? Most modern UK systems (Square for Restaurants, Epos Now, Lightspeed, Tabology and similar) export item-level sales as CSV from the back office; legacy systems often email scheduled reports instead. If yours can't export line-level detail, that's a reason to raise it with the vendor, not to give up.
How often should I change my drinks menu based on data? Item decisions on 28-day windows; structural menu changes quarterly, on the mix-shift pull. A one-week spike may reflect events, weather or availability; four-week patterns still need the same contextual checks before being treated as a structural change.
Method and assumptions
All thresholds (rank moves, dead-list windows, weekend-share levels) are working defaults we suggest venues test against their own history, not industry standards; nothing here is a claim about typical venue performance. Comparisons to stock-cycle reporting follow our methodology.
Last reviewed 19 October 2026.