Measurement
Why a London attribution benchmark does not transfer to England's regions
Explains why a London attribution benchmark does not transfer to England's regions, covering population, publisher mix, time of day and a modelled rebuild.
What to take away
- No, a London attribution benchmark does not transfer to England's regions. Auction density, publisher mix and audience composition all change.
- Greater London held just under 8.9 million residents at Census 2021 and the highest output per head of any UK region, ahead of the larger South East.
- No UK body publishes regional cost per thousand or cost per acquisition figures, so any regional cost figure is your own measurement or a labelled assumption.
- Never move a benchmark without moving the population definition with it.
Why a London benchmark does not transfer to the regions
No. A benchmark is a measurement taken under stated conditions, and regional conditions differ.
Change the conditions and the number stops describing your plan. A dashboard shows a variance, someone writes an explanation, and the explanation hardens into folklore.
Auction density is higher where more bidders chase less inventory. Publisher mix decides what a plan can buy, because London plans buy different titles, and audience composition moves when impressions and clicks land.
The deeper problem is what a benchmark claims about a population, which most teams skip before copying a London figure into a regional report.
Regional population and output per head compared
Population does not settle the benchmark. Greater London generates roughly a fifth of UK output from about 13 per cent of the UK population.
The table sets population against output per head and ranks media cost pressure.
| Region | Population, 21 March 2021 | Output per head | Media cost pressure |
|---|---|---|---|
| Greater London | Just under 8.9 million | Highest in the UK | Highest |
| South East England | Just under 9.3 million | Above the UK average | High |
| East of England | About 6.3 million | Close to the UK average | Moderate to low |
| North West England | About 7.4 million | Below the UK average | Moderate |
| West Midlands | About 6.0 million | Below the UK average | Moderate |
| Yorkshire and the Humber | About 5.5 million | Below the UK average | Moderate to low |
Population source: ONS Census 2021 first results, reference date 21 March 2021, released 28 June 2022.
The cost column ranks pressure only. No UK body publishes regional CPM figures at this granularity, so it is not a price list. Add your own contracted rates, dated when you agreed them.
Use ONS output statistics as the denominator and quote the release date beside each figure. Releases get revised, and the Code of Practice for Statistics sets what producers must do.
Why regional media costs stay unmeasured
Advertising Association and WARC put UK ad spend above £35 billion in 2023, published April 2024. That sizes a national market, not a placement in Bristol or Sunderland.
The example below is therefore structural. It shows how to rebuild a forecast when only the London rate is measured. The regional rate is not a published figure, and your report should say so.
For supply and demand across this market, see England's marketing attribution market.
Example: a London cost per acquisition rebuilt for the North West
A London plan reports a cost per acquisition of £42 on display and paid social. An analyst applies it to a £200,000 North West budget. The £42 is measured. The regional side is modelled.
- Fix the population. The London plan targeted commuters aged 25 to 44 inside the M25. A North West plan covers a wider age band across Greater Manchester and Merseyside.
- Compare publisher mix and placement volumes. London bought national news and lifestyle stock. The North West leans on regional news brands and local radio digital inventory.
- Compare the time-of-day profile. London commuting creates sharp morning and evening peaks. Dispersed North West audiences spread activity across the day.
- Set a tolerance and decision threshold. Reuse £42 only for the share of spend matching on all three fields within your stated tolerance, for example 10 per cent.
- Rebuild the forecast as a modelled range, naming the matched share and your measured regional rate.
Publisher mix, modelled conversions and audience timing
Publisher concentration decides how much of a regional report is modelled rather than observed. Fewer groups means less inventory diversity, so a London plan may find no equivalent regional substitute at the same volume.
| Region | Publisher groups | Representative brands | Why the modelled share rises |
|---|---|---|---|
| Greater London | Reach plc, Newsquest | MyLondon and lifestyle titles | National stock holds enough observed conversions |
| South East England | Newsquest, Reach plc, National World, Iliffe Media | Oxford Mail, The Argus, The News | Fragmented titles split events into small cells |
| North West England | Reach plc, Newsquest, National World | Manchester Evening News, Liverpool Echo | Large titles deliver volume, but per-title cells stay thin |
| West Midlands | Reach plc, Midland News Association | BirminghamLive, Express & Star | Two groups carry most inventory, so one cell dominates |
| Yorkshire and the Humber | National World, Reach plc, Newsquest | Yorkshire Post, The Star, Yorkshire Live | Three groups split the observed events |
If a publisher-level cell holds fewer than a few dozen conversions, merge it and say why. This is why attribution audience research that keeps populations apart matters more than any single benchmark figure.
For business to business plans, regional employment composition decides who the audience can reach, and the ONS employment and employee types collection is the starting point.
Common questions
Do London attribution benchmarks transfer to other UK regions?
Only when population definition, publisher type and time-of-day profile all match. Official data covers population and output per head, not media prices.
Which regions differ most from London on media cost pressure?
Greater London ranks highest, with the South East next. Output per head sits below the UK average in the North West, the West Midlands, Yorkshire and the Humber and Wales.
How do I test whether a London benchmark can be reused regionally?
Run the five steps above. Reuse the London figure only for the share of spend matching on population, publisher type and time of day within your stated tolerance. Forecast the remainder from your own last 90 days and label it modelled.
Where does the regional evidence run out?
At price. Population, output and employment data carry a reference date, but no regional CPM or CPA series is published.
