Median or average house prices: which one is the market?

“The average house price” is two different numbers wearing one name, and the gap between them is not pedantry. It decides what you believe a market costs. When a headline says average, it usually means the mean: add every price, divide by the count. When this site says median, it means the middle sale: half of sales came in above it, half below. In a big, calm market the two sit close together. In a postcode district, they routinely do not.

Why does the mean drift upwards?

The mean uses every pound of every sale, which means one exceptional transaction moves it. A district that trades forty homes a year around £250,000, plus one converted rectory at £2 million, gets a mean about £43,000 above what the typical buyer paid. Nobody bought at the mean. The median barely notices the rectory: the middle sale is still the middle sale. That resistance to extremes is the whole argument, and it matters most exactly where samples are small, which is what a postcode district is. How small is measured rather than guessed: every median here carries an interval that says so.

What do the two look like in one district?

The gap is visible in the live data. In SW11, Battersea, the 12 months to August 2026 show a median sold price of £720,000, while the mean across the same sales runs at £853,002, dragged up by a tail of multi-million pound houses between the commons. Neither number is false. The median says what the middle buyer paid; the mean says what the market turned over per sale. If you are deciding whether an asking price is reasonable, you want the first. If you are a treasurer summing stamp duty receipts, you want the second.

How far apart do the two get in practice?

Far enough to change a decision, and by an amount that varies district by district. Five districts, both measures, in cash and per square metre, every cell live from the current dataset:

The ordering is not an accident. Districts with grand houses scattered through ordinary stock, Kensington being the extreme case, hold the widest cash gaps, because a handful of very large sales pulls the mean and leaves the median where the market actually trades. Districts whose homes resemble each other, new-build heavy Canning Town, terrace-heavy Middlesbrough, hold the two measures close together. Notice that the per-square-metre columns sit closer than the cash columns everywhere: dividing by floor area removes the part of the tail that is simply bigger homes, which is much of it, and what remains is the genuinely dearer-per-metre stock.

Why does the choice matter twice for a rate?

Price per square metre is a ratio, so extremes get in through either the price or the floor area, a mismeasured EPC can make one sale look absurdly cheap per metre. RealScout therefore publishes the median price per square metre as the headline everywhere, shows the mean beside it where the sample is robust, and fences out physically implausible ratios before either is computed (the sales stay in the counts and the lists; only the broken ratio is excluded). The methodology page states every rule.

What does it mean when the two disagree?

A mean sitting well above the median tells you the market has an expensive tail: a few big houses in a district of flats, as in most of inner London. A mean below the median is rarer and usually means a tail of distressed or unusual cheap sales. Either way, the gap is a description of the market’s shape, which is why a district page also shows the middle 60% band (p20–p80) wherever the sample supports one, rather than pretending a single number can carry a whole distribution. In L1, central Liverpool, that band ran £1,980 to £3,620/m² over the 12 months to August 2026, nearly a factor of two inside one district, and no single “average” would have told you that.

Every term above is defined in the glossary, and every figure comes from the August 2026 dataset of HM Land Registry sold prices matched to EPC floor areas, rebuilt monthly.