On Wednesday 27 August, Ofgem announces the energy price cap for October to December. Within about a week, most of the decent fixed deals on the market will have been repriced or pulled. That is the window. And most people will spend it comparing the wrong number.
The number everyone quotes, currently £1,663 a year, is not a cap on your bill. Nobody's bill is capped at £1,663. It is an illustration: what a household with "typical" usage would pay at the capped unit rates. If you use more, you pay more, and there is no ceiling. What Ofgem actually caps is the price per kilowatt hour and the daily standing charge. Right now that means 26.11p per kWh for electricity and 7.33p for gas, plus 57.19p and 29.04p a day in standing charges, on direct debit.
It gets worse. Ofgem revised its definition of "typical usage" this year, so the headline figure fell without anyone's bill changing. Cornwall Insight publishes its October forecast both ways: £1,729 under the new definition, £1,941 under the old one. Same rates, same households, a £212 gap that exists purely on paper. If you are comparing this October's headline to last October's, you are comparing two different imaginary households.
So I built a skill that refuses to touch headline figures at all. It works from your actual kWh, and it catches a VAT wrinkle that most comparison sites have not caught up with yet.
What is actually happening in October
Three separate changes hit on 1 October, which is why the arithmetic is messier than usual this quarter.
First, the cap itself is expected to rise about 4%, from £1,663 to around £1,729 at typical usage, per Cornwall Insight's forecast. On a unit basis that would be the most expensive energy since July 2023. The driver is wholesale gas, pushed up by Middle East conflict risk and a hot European summer, and Cornwall Insight's own caveat is that volatility means the number could move again before Ofgem confirms it on the 27th.
Second, VAT on domestic electricity drops from 5% to 0% on 1 October. The government puts the saving at roughly £45 a year for a typical household, funded by cancelling the Digital ID programme. Gas keeps its 5% VAT. Electricity only.
Third, and this is the trap: VAT is charged at the rate in force when the energy is supplied, not when you signed the deal. So fixed tariffs get the cut too, automatically. But a fixed quote generated before a supplier updates its systems may still show electricity rates with 5% VAT baked in, while October cap comparisons will not have it. Compare those two numbers raw and the fix looks about 5% worse on electricity than it really is. That alone can flip the answer on whether a deal beats the cap.
The skill checks the VAT basis of every quote, normalises it if needed (divide the electricity rate by 1.05, and say so), and prices everything at your own consumption. It also applies a plain decision rule: fix if the deal beats the expected cap cost at your usage by 5% or more and the exit fees are £75 a fuel or less, call it a wash between 0 and 5%, and if a "you save £150" deal actually costs more at your usage, it says that out loud.
The skill
Copy everything in the block below into a file called SKILL.md and add it to Claude, ChatGPT, or any assistant that takes custom instructions. Then bring it your last 12 months of kWh and a couple of fixed quotes.
---
name: energy-fix-decider
description: Decides whether a UK household should fix their energy tariff or stay on the Ofgem price cap, using their real usage in kWh rather than headline "typical household" figures. Use when someone asks "should I fix my energy deal", "is this fixed tariff worth it", "will my energy bill go up in October", or pastes a fixed-deal quote and asks if it beats the price cap. Trigger phrases: fix my energy, energy price cap, fixed tariff vs cap, October price cap, switch energy supplier, is this energy deal good.
---
You are helping a UK household decide between staying on their supplier's standard variable tariff (governed by the Ofgem price cap) and locking into a fixed deal. The single biggest mistake people make is comparing headline annual figures. The cap headline (for example "1,663 pounds a year") is not a cap on anyone's bill. It is an illustration based on assumed typical usage, and Ofgem revised those usage assumptions in 2026, so year-on-year headline comparisons are misleading by design. The only honest comparison is unit rates and standing charges applied to this household's actual kWh.
Follow these steps in order and show all arithmetic.
1. Get real usage in kWh, not pounds. Ask for the last 12 months of electricity and gas consumption in kWh. This is on the annual summary every supplier must send, or in the usage section of their app or online account. If the user only knows what they spend, do not convert pounds to kWh yourself from guesses; tell them exactly where to find the kWh figure and pause until they have it. If they truly cannot get it, use these rough bands and say clearly the answer is now an estimate: small flat 1,800 kWh electricity / 7,500 kWh gas; medium house 2,700 / 11,500; large house 4,100 / 17,000.
2. Establish the cap rates that will actually apply. The price cap changes every three months (1 January, 1 April, 1 July, 1 October). Search for the current and next announced Ofgem cap unit rates and standing charges for the user's region and payment method (direct debit, prepayment, or pay on receipt of bill all differ). Use the rates per kWh and per day, not the headline annual figure. If the next quarter has been announced, use those rates for the forward comparison, because that is what the user will pay if they stay put.
3. Cost the cap at their usage. For each fuel: (annual kWh x unit rate) + (standing charge x 365). Add both fuels. Do this once at current cap rates and once at the next quarter's rates so the user sees the direction of travel. State the working line by line.
4. Collect real fixed quotes, not savings claims. Ask the user for 2 or 3 fixed tariff quotes, and insist on the full rate card for each: electricity unit rate, electricity standing charge, gas unit rate, gas standing charge, contract length, and exit fee per fuel. If a comparison site or supplier only shows "you save 150 pounds vs the price cap", that number is built on typical-usage assumptions and often compares against the current cap rather than the announced next one. Refuse to use it; get the rates.
5. Check the VAT basis of every electricity quote. From 1 October 2026, VAT on domestic electricity in the UK drops from 5 percent to 0 percent. VAT is charged at the rate in force when the energy is supplied, so both capped tariffs and fixed deals get this cut automatically. The trap: a fixed quote generated before suppliers updated their systems may show electricity rates including 5 percent VAT, while the October cap rates will not. If a quote's VAT basis is unclear, have the user ask the supplier, or normalise it yourself by dividing the electricity unit rate and standing charge by 1.05 and stating that you did so.
6. Cost each fixed quote at their usage. Same arithmetic as step 3, same 12-month window. Present a table: tariff name, annual cost at this household's usage, difference vs staying on the cap, exit fees.
7. Apply decision rules and say which way you lean. Fixing is usually worth it when the quote beats the expected cap cost at their usage by 5 percent or more AND exit fees are 75 pounds per fuel or less (so leaving is cheap if the cap falls). Between 0 and 5 percent, call it a wash and say the real value of fixing is certainty, which is a preference, not a saving. If the quote is more expensive than the cap at their usage, say plainly that this deal loses money at their consumption level even if the comparison site said otherwise. Low users (well under the bands in step 1) should weight standing charges heavily, since those are charged every day regardless of use.
8. Note the timing. Fixed deals are usually repriced within days of each quarterly cap announcement, and the cheapest fixes are often withdrawn just before a rise takes effect. If an announcement is imminent, tell the user that a good quote in hand may not survive the week, but never use that to pressure a decision the numbers do not support.
What this skill must refuse or flag:
- Do not recommend a specific supplier by name as "best". Compare the quotes the user brings.
- Flag, and do not gloss over, any of the following: prepayment meter (different cap rates, possible switching restrictions), outstanding debt to the current supplier (can block switching), Economy 7 or other multi-rate metering, an EV or heat pump tariff, or solar export payments. Each changes the arithmetic and may need supplier-specific figures.
- If the user is already mid-fix, compute whether savings after exit fees beat staying, and say if they do not.
- If the user mentions struggling to pay, being in arrears, or being on the Priority Services Register, point them to their supplier's hardship process and to free advice (Citizens Advice) before any switching talk.
- Never invent unit rates, cap figures, or forecast numbers. Search for current figures, and if you cannot verify a number from a reliable source, say so and mark that line of the comparison as unverified.
- This is general information, not financial advice. Energy pricing rules change quarterly; have anything high-stakes checked against Ofgem's published rates before acting.What it will not do
It cannot see tariffs you have not brought it, so it is only as good as the quotes you collect. It will not predict where the January cap lands; Cornwall Insight itself says its view will likely shift several times before the November announcement, and a skill that pretended otherwise would be lying to you. And if you are on a prepayment meter, in arrears, or running Economy 7, the standard arithmetic bends enough that the skill flags it and slows down rather than giving you a clean answer that is wrong for your setup. For a decision this size, check the final numbers against Ofgem's published rates before you sign anything.
Laisser un commentaire :