~/sajjad $ cat projects/draught-yield
← cd ../projects
Data & reporting
Draught yield check
Live · runs weekly7 draught lines
The beer-line meters measured every pint poured, but their link to the till had never worked, so the supplier’s own loss figure was meaningless. I built the match myself: every till key mapped to its meter line, so each week shows how many pints were poured but never rung through, and what they cost.
The problem
The venue already paid for flow meters on every draught line. Their dashboard reported “drinks tilled: 0” on every line, because the till had never been connected, and so its yield and loss figures were wrong. Nobody could say how much beer was going out unpaid.
How the match works
- Meter side: pints served per product, net of line cleaning, from the weekly CSV exports. The meter’s own £ and yield figures are thrown away.
- Till side: a parser maps every till key to its meter line. Halves count as half a pint, refunds come off, discounted “local” keys fold in, and repeated parent rows are never counted twice.
- Three cask brands share one metered handpull, so they are combined into one line.
- Till yield = till pints ÷ meter pints. The gap is costed at that week’s real average price per pint, net of VAT.
- Traffic lights: 97%+ green, 95–97% amber, under 95% red. Line-cleaning waste above 15 pints a week and pour rates under 98% are flagged too.
What it found
- Baseline week: 581 pints poured, 513 rung through, an 88% till yield. Every line lost pints.
- One line-clean wasted 89.5 pints in a single week, against a normal 6–13.
- The till had no wastage or staff-drink keys, so honest losses couldn’t be separated from unrung pints. Another product had two prices on two keys.
Result
A standing section of the weekly report with a named owner and targets: 95% within four weeks, 97% after that. It shows each product against last week and the daily gap, so busy-night losses are visible, and it is cross-checked against a monthly keg count. By week three the weekly loss was down 23%.