Energy is the last large line most venues never divide by covers. European half-year price series, published by consumption band and excluding taxes, give a restaurant a defensible outside benchmark; the venue's own kilowatt-hour price comes from its own invoice. Dividing the period's energy spend by covers served turns both into a figure a manager can act on.
The energy bill is the one large line a venue never divides by covers
Food gets divided. Labour gets divided. A restaurant owner can usually say what a portion of the flagship dish costs in ingredients, and what an hour on the pass costs in wages, because both numbers are argued about every week. Energy is not argued about. It arrives as one figure at the end of the month, it is paid, and it is filed.
That is odd, because energy behaves like neither of the two lines it sits between. It is not proportional to sales the way food is: a cold room runs at the same temperature on a dead Tuesday and on a full Saturday. It is not proportional to headcount the way labour is: the extra dishwasher does not double the extraction. It is a line driven mostly by how many hours your machines are switched on, and only secondarily by what you pay per kilowatt-hour.
So the useful questions are two, and they are separate. The first is external: is what I pay per unit roughly what a buyer my size pays, or am I clearly off? The second is internal: what does energy add to a single guest sitting down at my table, and which lever moves that? This page answers both, and it keeps them apart on purpose, because the first is answered with published statistics and the second is answered only with your own invoice and your own cover count.
One warning before the numbers. A tariff rate printed in an article rots. It is right on the day it is typed and wrong a quarter later, and nothing on the page tells the reader which of the two states it is in. That is why this page prints no retail tariff for Poland in any currency. It prints published half-year series that carry their own period label, and it teaches you to read your own rate off your own bill, which is the only rate that was ever true for you anyway.
Where comparable prices come from: half-year series for industrial consumers
Eurostat publishes energy prices for non-household consumers twice a year, by country, by consumption band, and separately for prices with and without taxes. Two datasets matter here: nrg_pc_205 for electricity and nrg_pc_203 for natural gas. Both were re-read for this page on 28 August 2026, both answered with HTTP 200, and both carry the same series update stamp of 11 August 2026.
Half-year price series — a published price for a country, a consumption band and a tax cut, covering six months of contracts rather than a moment. Two observations exist for every year, labelled S1 and S2, and a new observation replaces neither of the old ones: the series grows, it does not get corrected in place.
This is the property that makes the series usable in a document that will be read for years. Every figure below carries the half-year it belongs to. When the next observation lands, the figures here do not become false — they become older, which is a different thing, and the reader can see it without being told.
What the series is not is a price for restaurants. There is no activity dimension in either dataset: the cut is by how much you buy, not by what you do with it. A sweep of the full Eurostat catalogue, read as plain text on 28 August 2026, returns no dataset that prices energy by NACE activity for food service (Eurostat, catalogue of datasets, plain text, opened 28 August 2026). There is therefore no official figure of the form "the average restaurant pays X", and this page does not invent one.
Consumption bands: why a venue and a factory pay different prices for the same kilowatt-hour
Prices are published per band because buyers of different size genuinely pay different prices, and averaging across them produces a number that describes nobody. A band is defined by annual volume, and it is stated in the dataset itself rather than interpreted by us.
Consumption band — a class of buyer defined by annual volume. Electricity band IB covers consumption from 20 MWh to 499 MWh a year; gas band I1 covers less than 1 000 GJ a year and band I2 covers 1 000 GJ to 9 999 GJ. Comparing a venue with a national average that ignores the band is comparing it with a mix of buyers it does not belong to.
Which band is yours is arithmetic on your own invoices, not a judgement call. Add twelve months of metered kilowatt-hours from your electricity bills; if the total lands between 20 000 and 499 000 kWh, the electricity figures below are the ones written for a buyer your size. For gas the unit differs and one conversion is needed: one kilowatt-hour is 3.6 MJ by definition, so one gigajoule is 1 000 divided by 3.6, that is 277.8 kWh. Band I1 therefore ends at roughly 278 MWh of gas a year.
Working out your own band from twelve invoices
Take the example venue used throughout this page — invented numbers, chosen only to make the arithmetic visible. It meters 5 600 kWh of electricity in the month examined, which is 67 200 kWh over twelve comparable months, that is 67.2 MWh. Band IB. It meters 6 500 kWh of gas, which is 78 000 kWh over the year, and 78 000 multiplied by 3.6 MJ is 280 800 MJ, that is 280.8 GJ. Band I1.
Two remarks that cost people money. First, a venue near a band boundary should check the boundary before quoting any comparison, because the neighbouring band is a different price. Second, twelve months is the right window even if your contract is shorter: the band is annual, and a summer quarter annualised will put a restaurant with a garden in the wrong class entirely.
Electricity in Poland and the EU: four half-years in a row
All four figures below are prices excluding taxes and levies, in euro per kilowatt-hour, for electricity band IB — consumption from 20 MWh to 499 MWh a year (Eurostat, nrg_pc_205, band IB, excluding taxes and levies, EUR per kWh, series updated 11 August 2026, opened 28 August 2026).
| Half-year | Poland | EU27 | Difference |
|---|---|---|---|
| 2024-S1 | 0.1760 | 0.1933 | −0.0173 |
| 2024-S2 | 0.1822 | 0.1921 | −0.0099 |
| 2025-S1 | 0.1936 | 0.1900 | +0.0036 |
| 2025-S2 | 0.1906 | 0.1818 | +0.0088 |
Read the Polish column on its own first. From the first half of 2024 to the first half of 2025 the Polish price for this band rose by 0.0176 EUR per kilowatt-hour, which is 10.00 per cent, and then eased by 0.0030 in the second half of 2025, which is 1.55 per cent down. Over the same two years the EU27 figure fell from 0.1933 to 0.1818, that is 5.95 per cent down.
Two movements in opposite directions produce the whole story of this page, and neither of them is visible if you look only at your own invoice.
The 2025 crossover: Poland passed the EU from below, and what it does to a budget
In the difference column above the sign changes exactly once, between 2024-S2 and 2025-S1. In the first half of 2024 this band cost 8.95 per cent less in Poland than the EU27 average; in the second half of 2025 it cost 4.84 per cent more. That is a description of a published series, not a forecast, and nothing here says the sign will stay where it is.
What it does to a budget is easiest to see as money. On the example venue's 67 200 kWh a year, the 2024-S1 gap of 0.0173 EUR per kilowatt-hour was worth 1 162.56 EUR of advantage a year; the 2025-S2 gap of 0.0088 is worth 591.36 EUR of disadvantage. The swing between the two half-years is 0.0261 EUR per kilowatt-hour, which on the same volume is 1 753.92 EUR a year — a sum that appears in no line of the profit and loss account under the word "energy prices", because it hid inside a bill that also changed volume.
That is the practical use of an outside series. It separates the part of your bill that moved because the market moved from the part that moved because you did something, and no internal report can do that separation for you.
A budget note that follows from the shape of the data rather than from opinion: because the series is half-yearly, a plan built on a single observation carries six months of unmeasured drift by construction. Use the direction across four observations, not the level of the last one.
Gas across two consumption bands: the smaller buyer usually pays more, and once did not
The gas dataset is read for two bands at once, which makes the size effect visible in the same table. All figures are excluding taxes and levies, in euro per kilowatt-hour (Eurostat, nrg_pc_203, bands I1 and I2, excluding taxes and levies, EUR per kWh, series updated 11 August 2026, opened 28 August 2026).
| Band | Half-year | Poland | EU27 |
|---|---|---|---|
| I1, under 1 000 GJ | 2025-S1 | 0.0840 | 0.0753 |
| I1, under 1 000 GJ | 2025-S2 | 0.0739 | 0.0735 |
| I2, 1 000 to 9 999 GJ | 2025-S1 | 0.0800 | 0.0674 |
| I2, 1 000 to 9 999 GJ | 2025-S2 | 0.0765 | 0.0638 |
In the EU27 the expected pattern holds in both half-years: the smaller band costs 11.72 per cent more than the larger one in 2025-S1 and 15.20 per cent more in 2025-S2. In Poland it holds in the first half of 2025 by 5.00 per cent — and then reverses. In the second half of 2025 the Polish small band at 0.0739 is 3.40 per cent cheaper than the Polish large band at 0.0765.
That reversal is stated here because the data says it, not because it is convenient. The Polish small band fell 12.02 per cent between the two half-years while the large band fell 4.38 per cent, and the order of the two flipped as a result. Anyone quoting "small buyers always pay more" from a single half-year would have been repeating a rule the next observation broke.
The Poland-to-EU27 gaps in gas run the other way from electricity: Poland is above the EU27 average in all four cells, by 0.0087 and 0.0004 in band I1 and by 0.0126 and 0.0127 in band I2. Note how small the 2025-S2 gap in band I1 is. A gap of four ten-thousandths of a euro per kilowatt-hour is not a fact a restaurant should build a decision on; it is a reminder that the two series nearly met.
Prices excluding taxes and levies — why that is the cut to compare
Both datasets are read here in the cut labelled by Eurostat itself as "excluding taxes and levies". That choice is not a technicality, and it changes what a comparison means.
Price excluding taxes and levies — the price of the commodity and its delivery before any tax, excise or levy the state adds. It is the cut that compares countries, because the tax component is a decision of each government and a difference in tax is not a difference in the price of energy.
For your own venue the opposite is true. You pay the gross figure at the bottom of the invoice, and it is the gross figure that leaves the bank account, so your cash plan is built on it. Keep both, and never mix them in one sentence: the net price is what you compare with the series, the gross amount is what you divide by covers.
The single most common error in energy comparisons is putting one venue's gross rate next to a published net rate and announcing a gap. The gap in that comparison is the tax, and it says nothing about whether the venue buys well. If your finance layer holds one of these numbers, it should hold both with the cut labelled — that is the sort of thing a finance contour exists to keep honest.
From the bill to kilowatt-hours: taking your own invoice apart
An energy invoice is not one price. It is a bundle of a metered volume, a commodity charge, distribution charges that are partly fixed, and taxes. The volume is the only part that responds to what happens in the kitchen, so the first job is separating it from everything else.
The three parts to separate, and why the fixed part stays in
Split the invoice into: metered kilowatt-hours; every charge that scales with those kilowatt-hours; every charge that does not, including the standing and capacity components. Then add the second and third groups together, net of tax, and divide by the kilowatt-hours.
Your own kilowatt-hour price — the net amount of the invoice divided by the metered kilowatt-hours of the same invoice, with the fixed components included. Leaving the fixed part out produces a lower figure with no cause behind it, and that figure will not compare with anything.
Your own kilowatt-hour price = Net invoice amount ÷ Metered kilowatt-hours
Net invoice amount— everything the supplier charges for that period before tax, both variable and fixed components, PLN;Metered kilowatt-hours— the volume read from the meter for the same period, kWh.
Worked on the example venue, with invented figures: 4 480 PLN net over 5 600 kWh gives 0.80 PLN per kilowatt-hour. The gas invoice for the same month, 1 950 PLN net over 6 500 kWh, gives 0.30 PLN per kilowatt-hour. Both are that venue's prices for that month, and neither is a market rate for anybody else.
Comparing 0.80 PLN per kilowatt-hour with 0.1906 EUR per kilowatt-hour requires converting one into the other's currency at a rate for the month of the invoice. This page prints no exchange rate, for the same reason it prints no tariff: a rate typed into an article is wrong the following week. Take the rate from the central bank for your invoice month, convert your own figure, and compare inside one band and one tax cut.
Energy per cover: the formula and the denominator that decides it
The outside series answers "am I buying badly". This formula answers "what is energy doing to my plate", and it is the number a manager can actually move.
Energy per cover = Energy spend for the period ÷ Covers served in the same period
Energy spend for the period— electricity plus gas plus any other metered energy for the period, gross if you are planning cash, net if you are comparing with the series, PLN;Covers served— guests seated in the same period, counted from the reservation and floor records, guests.
Energy per cover — energy spend for a period divided by the number of guests served in that period. It is a management figure and not an official statistic: nobody publishes it, and its only meaning is against your own previous months.
Covers, not checks — the denominator changes the answer by the size of a party
The example venue spends 6 430 PLN on energy in the month, serves 3 200 guests and issues 1 280 checks. Per cover that is 2.01 PLN. Per check it is 5.02 PLN. Both are arithmetically correct and they differ by exactly 2.5, which is the average party size, because that is the whole content of the difference.
Substituting checks for covers therefore overstates the figure by the average party size, and the error is invisible: both versions look like reasonable numbers. Say which denominator you used in the same sentence as the result, every time. The same discipline applies to every per-guest figure in the section — the boundary between an average check and an average guest has a page of its own in average check versus average guest.
Two more denominators are wrong in ways worth naming. Covers booked rather than served counts no-shows as guests. Covers for the calendar month against an invoice for a billing period that is not the calendar month mismatches numerator and denominator by however many days the periods differ; align the periods first, even if it means using the previous closed month.
What moves this number more than price does: equipment running hours
Here is the part that surprises owners who have spent a month negotiating with suppliers. The consumption side of the bill is usually the larger lever, and it is entirely inside the building.
Equipment contribution = Rated power × Running hours × Your own kilowatt-hour price
Rated power— the nameplate power of the appliance, kW;Running hours— hours the appliance is actually drawing power, measured rather than taken from the roster, h;Your own kilowatt-hour price— the figure from the previous section, PLN/kWh.
On the example venue: an appliance of 5 kW running 12 hours a day at 0.80 PLN per kilowatt-hour contributes 5 × 12 × 0.80 = 48 PLN a day, which is 1 440 PLN a month. Switching that one appliance on three hours later brings it to 5 × 9 × 0.80 = 36 PLN a day, saving 360 PLN a month — 8.0 per cent of that venue's electricity bill, from one appliance and one habit.
Compare that with the price lever. A 5 per cent improvement on the same venue's whole electricity bill is 224 PLN a month. Three hours a day off one machine beat a five per cent price negotiation on everything, and the 360 PLN also lands as 0.11 PLN off the energy per cover, taking it from 2.01 to 1.90.
This is why the running-hours question comes before the supplier question. It also explains the gap between two venues with the same menu, the same size and the same supplier: they are not paying different prices, they are running different hours. Measuring those hours is the sort of thing an analytics layer and a dashboard are for, because nobody writes down when the extraction was switched on.
Energy also sits in the labour question from the other side: the hours a machine runs and the hours a person is paid for are set by the same shift plan, and moving one moves the other — see how many people a shift needs for the staffing half of the same decision.
Energy seasonality and the cash plan: the payment does not arrive when you expect it
The last section is about timing rather than level, because a correct annual figure can still break a week.
Energy consumption in a restaurant is seasonal in two directions that do not cancel. Heating and hot water rise in winter; cooling, ice and ventilation rise in summer. A venue with a garden adds a third season with its own load. Billing adds a fourth pattern on top, because settlement periods, estimated readings and true-ups mean the month you consume and the month you pay are frequently not the same month.
The consequence is ordinary and it still catches people: the largest energy payment of the year often lands in a week that also carries a supplier settlement. That is a cash question rather than a cost question, and it is planned the same way supplier terms are planned — see supplier payment terms and the cash plan for the weekly grid this payment goes into.
Build the series before you plan against it. Twelve closed months of energy spend, covers served and energy per cover, on one sheet, is enough to see the shape; anything shorter shows you a season and calls it a trend. A calculator on your own numbers is a reasonable place to keep it, and external factors such as temperature belong next to it rather than inside it — that is what external signals around a venue are for.
One thing this page deliberately does not do: it does not tell you whether to change supplier, and it does not compute the payback of new equipment. Both are real questions with real arithmetic, and both need inputs — offer terms, appliance specification, financing — that no published series contains.
Frequently asked questions
What did electricity cost for a mid-sized venue in Poland in 2025?
For the band that fits a mid-sized venue — band IB, 20 MWh to 499 MWh a year — electricity excluding taxes and levies cost 0.1936 EUR per kilowatt-hour in the first half of 2025 and 0.1906 EUR in the second half, from the Eurostat series nrg_pc_205 updated on 11 August 2026. Those are band prices, not your price: yours comes from dividing your own net invoice by your own metered kilowatt-hours.
How does that compare with the EU average?
In the same band and the same tax cut, the EU27 figure was 0.1900 EUR per kilowatt-hour in the first half of 2025 and 0.1818 in the second, so Poland stood 1.89 per cent and then 4.84 per cent above the EU27 average. Two half-years earlier the relation was the other way round: in the first half of 2024 Poland was 8.95 per cent below.
Why compare prices excluding taxes?
Because the tax component is set by each government separately, and a difference in tax is not a difference in the price of energy. The cut labelled "excluding taxes and levies" strips it out, which is what makes two countries comparable. Your own cash planning still runs on the gross amount you actually pay, and the two figures must never be mixed in one comparison.
What is a consumption band and why does it matter?
It is a class of buyer defined by annual volume — electricity band IB runs from 20 MWh to 499 MWh a year, gas band I1 sits under 1 000 GJ and band I2 from 1 000 GJ to 9 999 GJ. It matters because prices genuinely differ between bands: in the EU27 in the second half of 2025 the smaller gas band cost 15.20 per cent more than the larger one. A comparison that ignores the band compares you with buyers you are not.
How do I turn my energy bill into a cost per cover?
Two divisions. First, net invoice amount divided by metered kilowatt-hours gives your own price per kilowatt-hour, with the fixed charges left in. Second, total energy spend for the period divided by guests served in the same period gives energy per cover. Align the periods before you divide: an invoice for a billing period is not automatically an invoice for your calendar month.
Should the denominator be checks or covers?
Covers. A party of four produces one check and four guests, so using checks overstates energy per guest by exactly the average party size — 2.5 in the worked example on this page, which turns 2.01 PLN into 5.02 PLN on identical money. Both figures are arithmetically correct and neither is self-describing, so name the denominator in the same sentence as the number.
What moves energy cost more, the price or the running hours?
Usually the running hours. On the worked example, taking three hours a day off a single 5 kW appliance saves 360 PLN a month, while a 5 per cent improvement on the entire electricity price saves 224 PLN. The price is negotiated once a year with a counterparty; the hours are decided every day by your own staff, and that is where the larger and faster lever sits.
Take your last closed month and do the two divisions: your own price per kilowatt-hour from the invoice, and energy per cover from the same period's guest count. From then on compare only those two figures — against the published band series, and against your own previous months. The rest of the section is built on numbers you own: restaurants.
Related reading in the section and on the blog: restaurant prime cost for where energy sits and why it is not inside prime cost, alcohol licence fees in Poland for another payment that arrives in steps, Polish restaurant economics from official data for what else is published about the sector, reporting automation and the numbers an owner actually looks at, what process automation costs, restaurant automation across reservations, suppliers and reviews, CRM or ERP: which layer to add next and the four thresholds small companies start from.