What Makes a
Great Area

Census demographics correlated with commodity-normalised event revenue across 983 matched areas

983
Areas Matched
20
Features Tested
2,793
Target Villages
1,305
Target Towns
213
Target Suburbs
How We Built This

The Approach

This analysis links your internal event performance data to external Census demographics at the individual settlement level, then uses the relationship between the two to score every town in England and Wales you haven't been to.

Data Ingested
2,312 roadshow events from your Google Sheets, normalised against a blended commodity price index (gold + silver, base Jan 2024) to strip out metal price movements and isolate real operational performance.

ONS Census 2021 Built-up Area characteristics dataset: population, median age, tenure breakdown, accommodation type, qualification levels, occupation classifications and employment status for all 6,975 named settlements in England (exc. London) and Wales.

English Indices of Deprivation 2019: 32,844 Lower Super Output Areas aggregated to local authority district level, capturing income deprivation, housing deprivation and overall IMD scores.

ONS Open Geography Portal: Built-up Area to Local Authority District lookups (10,072 mappings), BUA to County/Unitary Authority lookups (9,604 mappings), and Built-up Area Sub-division to parent BUA lookups (1,693 named suburbs) for the city suburb target list.
Matching & Analysis Pipeline
Four-pass entity resolution to link your 1063 visited areas to ONS Built-up Areas: exact case-insensitive match, normalised match (stripping hyphens, prepositions, suffixes), prefix match (handling ONS compound names like "Abbots Langley and Kings Langley"), and fuzzy match using Ratcliff/Obershelp sequence matching at a 0.82 similarity threshold. Result: 92.5% match rate (983 areas).

Correlation analysis restricted to areas with 2+ events to reduce single-observation noise. 16 Census features and IMD deprivation scores tested against commodity-normalised average revenue per area using Pearson product-moment correlation. Quartile segmentation at p25/p75 for top-vs-bottom comparison.

Composite scoring model built from all features with |r| > 0.05, weighted by absolute correlation strength. Feature values normalised against p10-p90 range from the matched dataset to avoid ceiling effects. Negative correlations inverted before scoring. Output: a 0-100 demographic similarity score for every settlement in the ONS dataset.
Why this matters: If we only look at the data we collect from roadshows, the analysis is circular. Footfall correlates with revenue because more people means more money. Venue scores correlate because good venues were already good. Those metrics describe what a successful event looks like after the fact, but they can't tell you where to go next. To answer that, we need to match roadshow performance against wider datasets that describe the areas themselves: Census demographics, deprivation indices, housing characteristics. Things you can look up about a town before you've ever set foot in it. That's what this analysis does.
Thinking

Why These Variables

Every feature we tested was chosen for a specific reason. Here's the hypothesis behind each one and what the data actually showed.

Housing & Tenure

The core thesis: people who have lived in the same house for a long time accumulate things. Bigger houses store more. Owners keep things; renters shed them when they move.

% owned outright +0.06 confirmed
Outright owners are the most settled population segment. No mortgage means they've either lived there for decades or bought with inherited wealth. Either way, they've had the longest time to accumulate items in attics, garages and spare rooms. They're also the demographic most likely to be downsizing, clearing estates, or dealing with probate. The hypothesis: longest tenure = most accumulated stock. The data confirms it.
% owned with mortgage
Younger households, shorter tenure, still building their lives rather than clearing them out. We expected this to be weaker than outright ownership but still mildly positive (homeowners generally accumulate more than renters). The data shows a weak positive signal (+0.02) but it's the weakest of the tenure categories. Mortgage holders are still buying, not selling.
% private rented -0.05 confirmed
Renters move. The average private tenant in England stays 4.4 years; the average homeowner stays 12+. Every move is a forced clear-out, which means renters carry less accumulated stock. Private renters also skew younger and more urban. The hypothesis: transient populations have less to sell. The data confirms it as the 4th strongest signal.
% social rented -0.04 confirmed
Social housing tenants tend to have lower incomes, which means the items they do accumulate are less likely to include gold, silver, or high-value collectibles. The accumulation pattern is different: more functional items, fewer luxury or heirloom pieces. Also often in purpose-built estates rather than older properties with accumulated character.
% detached housing +0.07 strongest signal
The strongest predictor in the dataset. Detached houses are doing double duty as a signal: they're a proxy for wealth (detached homes cost more) and a proxy for storage capacity (more rooms, garages, lofts, outbuildings). A 4-bed detached in a market town has space to keep 40 years of accumulated belongings. A 2-bed terrace doesn't. The items are physically there because the house can hold them.
% terraced housing -0.04 confirmed
The inverse of detached. Terraced housing means smaller properties, less storage, more urban settings. Also tends to correlate with younger, more transient populations. Victorian terraces in gentrified areas might have wealthy owners, but on average, high-terrace areas underperform. Less space to accumulate, less to sell.
Age & Employment

Older people have had more time to accumulate. Retired people are available during the day (when events run). And they're the generation most likely to hold physical gold, silver, and inherited items.

Median age +0.04 confirmed
A 55-year-old has had twice as long to accumulate items as a 30-year-old. They're also the generation that grew up buying physical gold and silver as savings, inheriting jewellery from parents, and collecting coins, cameras, and watches. Younger populations hold their wealth in ISAs and crypto, not in the back of a drawer. Median age captures this generational difference directly.
% not employed (inc. retired) +0.03 confirmed
Events run during the daytime, typically 10am-4pm. Retired people and economically inactive populations are available. Working people aren't. Higher retirement rates also correlate with older populations (reinforcing the age signal) and with areas where people have finished their careers and are now in the "clearing out" phase of life.
Wealth, Education & Occupation

Wealthier people have higher-value items. But wealth alone doesn't predict willingness to sell at a roadshow. The interesting question is where wealth intersects with motivation.

% managers & professionals +0.06 3rd strongest
Professional and managerial households have higher incomes and accumulate higher-value items: better jewellery, more expensive watches, inherited silverware, quality collectibles. The items in a surgeon's attic are worth more than the items in a warehouse worker's attic. This feature is essentially a proxy for item quality and average basket value. It ranked 3rd overall.
% degree-level education
We tested this as a wealth proxy but expected a weaker signal than occupation, because education doesn't directly translate to item value. There was also a counter-hypothesis: highly educated people might be more price-savvy and less likely to sell at a roadshow (they'd use eBay or a specialist dealer instead). The data shows a very weak positive (+0.04). Education correlates with wealth, which correlates with item value, but the effect is diluted.
% skilled trades
Tested because of the insight that working-class areas can surprise: tradespeople collect tools, watches, coins, and practical valuables. The hypothesis was that skilled-trade areas might outperform what their income level suggests. The data shows a slight negative (-0.03), meaning they slightly underperform. The items are there, but they're lower value on average. The gold and silver that drives 80% of revenue tends to sit with wealthier demographics.
Roots & Generational Accumulation

How long have people been in the same place? Multi-generational UK families accumulate items over decades: inherited jewellery, family silver, grandfather's watch collection, coins in a drawer. First-generation households haven't had that accumulation time.

% UK-born
This isn't about ethnicity. It's about generational accumulation in the UK. A family that's been in the same Somerset village for three generations has a house full of inherited items: grandmother's gold rings, a box of old coins, silverware nobody uses. A first-generation household from anywhere (Poland, India, Australia) hasn't had that accumulation time in the UK. The stock WBV buys is overwhelmingly the product of multi-generational UK accumulation. However, the data shows no signal (r=-0.01). Why? Because the average BUA is 94% UK-born, and WBV's visited areas are overwhelmingly in that range anyway. There simply isn't enough variation to detect a signal. This variable would matter more if WBV expanded into large cities with diverse inner-urban populations, but across market towns the UK-born % barely varies.
Deprivation & Population

The most nuanced variables. Deprivation can go either way depending on whether the area "used to have money" or never did. Population size captures the community network effect.

IMD deprivation score
The Index of Multiple Deprivation is a composite of income, employment, education, health, crime, housing, and environment. Higher score = more deprived. Our hypothesis was complex: very deprived areas lack items, but moderately deprived areas that used to be wealthier (former market towns, post-industrial towns) still have the stock from better times, and the population now has more motivation to sell. Wisbech is the archetype: deprived on paper, but the houses are full of inherited gold and silver. The data shows a weak negative (-0.03), meaning on average deprivation hurts. But this is an average across all types of deprivation. A future iteration should decompose IMD into its sub-domains: income deprivation (need to sell) vs housing deprivation (old housing full of stuff) to separate these opposing effects.
Rural score (geographic barriers)
The ONS "Geographical Barriers" sub-domain of the Index of Multiple Deprivation measures road distance to a GP, post office, primary school, and general store. Higher score = further from services = more rural. This is a direct, measured rurality metric rather than a proxy. The hypothesis: more rural areas have stronger community cohesion, less competition from other events and shopping, and the roadshow becomes the event of the week rather than one of many options. Rural communities also tend to have older, more settled populations who've accumulated more.
Population
In towns under about 6,000 people, everyone knows what's happening. The roadshow is an event. In larger towns, you're competing with shopping centres, other events, and general urban noise. Below 2,500 there aren't enough people to fill a room, but our own data shows villages under 1,000 averaging £15.8k per event (2+ visits) with Gosforth in Cumbria (pop 930) doing £40k and Thornton (pop 335) doing £39k. The hypothesis was a sweet spot between 3k-12k. The data shows a weak negative correlation (-0.02), meaning smaller towns slightly outperform, but the signal is noisy. What matters more is the character of the population than how many of them there are.
The Sweet Spot

The Profile

We matched 983 of your 1063 visited areas to ONS Census 2021 built-up areas, then correlated 17 demographic features with your commodity-normalised event revenue. Here's what separates your top-quartile areas from the bottom.

The pattern: Your best areas are settled communities with older populations who own their homes outright, live in detached or semi-detached houses, and have been accumulating things for decades. The gold comes with everything else.
Demographic Predictors

What Drives Revenue

Each Census feature ranked by its correlation with your commodity-normalised revenue. Positive means higher values predict higher revenue; negative means the opposite. These are external factors you can look up before visiting.

#1
% detached housingmore = better
+0.07
#2
Rural score (geographic barriers)more = better
+0.07
#3
% owned outrightmore = better
+0.06
#4
% managers & professionalsmore = better
+0.06
#5
% private rentedless = better
-0.05
#6
Median agemore = better
+0.04
#7
% social rentedless = better
-0.04
#8
% degree-level educationmore = better
+0.04
#9
% terraced housingless = better
-0.04
#10
% semi-detachedless = better
-0.04
#11
% employedless = better
-0.03
#12
% not employedmore = better
+0.03
#13
% skilled tradesless = better
-0.03
#14
IMD deprivation scoreless = better
-0.03
#15
Populationless = better
-0.02
#16
% owned with mortgagemore = better
+0.02
#17
% UK-bornless = better
-0.01
#18
% no qualificationsless = better
-0.01
#19
% non-EU bornmore = better
+0.01
#20
% EU-bornless = better
-0.00
Reading this chart: Bars to the right of centre (green, "more = better") mean higher values of that feature predict higher revenue. Bars to the left (red, "less = better") mean lower values predict higher revenue. In the scoring model, negative correlations are inverted: an area with low private renting scores well on that feature, because low renting predicts high revenue. All features contribute to the composite score weighted by their absolute correlation strength, regardless of direction.

Honest note: Individual correlations are modest (r = 0.04-0.08). Demographics set a baseline but don't determine event success. Operational factors (venue quality, marketing, dealer assignment, day of week) are bigger drivers. The target list scores mean "demographically similar to your best areas", not "guaranteed to perform well." The signal is real but directional, not deterministic.
Regional View

Performance by Region

Your 18 operational regions ranked by average event revenue.

RegionEventsAreas Avg RevenueAvg MarginMargin %
North Wales 13 10 £38,736 £16,810 43.4%
South Wales 17 13 £26,579 £9,669 36.4%
Fenland 98 51 £18,872 £8,163 43.3%
Home Counties 137 68 £17,107 £8,465 49.5%
North West 125 60 £16,864 £7,126 42.3%
Thames Valley 130 55 £16,676 £7,906 47.4%
Cumbria 61 30 £15,333 £6,914 45.1%
South East 116 72 £15,308 £7,585 49.5%
West Country 64 39 £14,733 £6,567 44.6%
East Midlands 125 70 £14,664 £6,561 44.7%
North East 184 86 £14,530 £5,839 40.2%
East Anglia 208 83 £14,092 £6,288 44.6%
South Coast 113 69 £14,058 £6,593 46.9%
Mersey & Peaks 171 78 £13,719 £6,136 44.7%
West Midlands 110 58 £13,586 £5,872 43.2%
Yorkshire 280 125 £13,532 £5,572 41.2%
Severn 163 97 £10,944 £4,985 45.6%
County View

Performance by County

Every county ranked by average event revenue. * = fewer than 5 events (small sample).

CountyRegionEventsAreas Avg RevenueAvg MarginMargin %
Pembrokeshire * South Wales 1 1 £78,509 £19,049 24.3%
North Powys * North Wales 1 1 £64,568 £19,403 30.1%
Denbighshire * North Wales 1 1 £60,365 £16,635 27.6%
Cardiff * South Wales 2 2 £48,246 £11,425 23.7%
Clwyd * North Wales 1 1 £45,488 £22,483 49.4%
Isle of Anglesey * North Wales 4 2 £42,651 £19,564 45.9%
Conwy * North Wales 2 2 £33,896 £13,239 39.1%
South Powys * South Wales 1 1 £31,541 £15,409 48.9%
Vale of Glamorgan * South Wales 4 2 £31,281 £14,978 47.9%
Bedfordshire Fenland 23 13 £28,433 £11,659 41.0%
Flintshire * North Wales 3 2 £25,365 £14,407 56.8%
Northamptonshire Fenland 20 10 £21,119 £9,808 46.4%
Carmarthenshire * South Wales 3 2 £18,806 £6,852 36.4%
Wrexham * North Wales 1 1 £18,660 £12,050 64.6%
Rutland East Midlands 5 2 £18,415 £8,977 48.7%
Lancashire North West 74 34 £17,926 £7,530 42.0%
Hertfordshire Home Counties 57 27 £17,923 £8,699 48.5%
Staffordshire Mersey & Peaks 21 15 £17,768 £8,279 46.6%
Leicestershire East Midlands 39 22 £17,302 £7,603 43.9%
Surrey South East 47 26 £17,130 £8,373 48.9%
Essex Home Counties 75 38 £17,090 £8,542 50.0%
Berkshire Thames Valley 40 17 £16,966 £7,753 45.7%
West Sussex * South Coast 2 1 £16,924 £7,462 44.1%
Buckinghamshire Thames Valley 37 15 £16,684 £7,932 47.5%
Oxfordshire Thames Valley 53 23 £16,452 £8,004 48.7%
Hampshire South Coast 50 32 £16,265 £7,695 47.3%
West Midlands West Midlands 28 18 £16,074 £7,024 43.7%
Nottinghamshire East Midlands 23 14 £16,003 £6,632 41.4%
Kent South East 55 36 £15,871 £7,969 50.2%
Cambridgeshire Fenland 56 29 £15,769 £6,909 43.8%
Durham North East 82 39 £15,716 £6,332 40.3%
Devon West Country 37 22 £15,680 £7,452 47.5%
Tyne & Wear North East 49 25 £15,370 £6,028 39.2%
Cumbria Cumbria 61 30 £15,333 £6,914 45.1%
Greater Manchester North West 51 28 £15,322 £6,540 42.7%
East Riding Yorkshire 42 17 £15,056 £5,977 39.7%
Derbyshire Mersey & Peaks 35 20 £14,978 £6,649 44.4%
Monmouthshire * South Wales 3 2 £14,564 £6,279 43.1%
South Yorkshire Yorkshire 42 23 £14,508 £6,116 42.2%
Norfolk East Anglia 129 51 £14,097 £6,356 45.1%
Suffolk East Anglia 79 34 £14,084 £6,176 43.9%
Somerset Severn 62 34 £13,995 £6,204 44.3%
West Yorkshire Yorkshire 89 43 £13,927 £5,756 41.3%
Worcestershire West Midlands 34 15 £13,510 £5,906 43.7%
Cornwall West Country 27 18 £13,435 £5,354 39.9%
Merseyside Mersey & Peaks 30 10 £13,384 £5,858 43.8%
West Sussex South Coast 35 19 £13,351 £6,294 47.1%
Warwickshire West Midlands 34 20 £12,984 £5,809 44.7%
Lincolnshire East Midlands 48 24 £12,553 £5,823 46.4%
Cheshire Mersey & Peaks 71 26 £12,383 £5,360 43.3%
North Yorkshire Yorkshire 110 45 £12,370 £5,131 41.5%
Shropshire Mersey & Peaks 16 8 £12,145 £6,278 51.7%
Northumberland North East 51 24 £12,006 £4,919 41.0%
Rhondda * South Wales 1 1 £11,752 £3,197 27.2%
Dorset South Coast 21 14 £10,523 £4,714 44.8%
Herefordshire West Midlands 14 6 £10,259 £3,637 35.5%
Bristol Severn 15 13 £10,240 £5,461 53.3%
Wiltshire Severn 40 24 £9,179 £4,354 47.4%
Gloucestershire Severn 46 26 £8,594 £3,737 43.5%
East Sussex South East 18 15 £8,105 £3,965 48.9%
Greater London Home Counties 5 3 £8,064 £4,657 57.8%
Torfaen * South Wales 1 1 £6,518 £3,428 52.6%
NottInghamshire East Midlands 7 6 £5,075 £2,437 48.0%
West sussex * South Coast 1 1 £4,979 £2,729 54.8%
Bridgend * South Wales 1 1 £1,788 £1,133 63.4%
Repeat Visits

Revenue by Visit Number

How event revenue changes across 1st, 2nd, 3rd visits to the same area.

Survivorship bias: Revenue increases with visit number because you selectively go back to areas that worked. You don't revisit the duds. This is actually good news: it means your instincts about which areas to revisit are sound. The goal now is to find more first-visit areas that match the demographic profile of your best performers.
Target Lists

How We Built the Lists

Three lists below, split by settlement size. Every list has been through the same filtering pipeline.

Starting Universe
6,975 named settlements in England (exc. London) and Wales from the ONS Census 2021 Built-up Areas dataset. Every settlement has population, demographics, housing, tenure, qualification, and occupation data attached.
What Gets Excluded
1,063 areas you've already visited (matched from your event data via four-pass entity resolution).
1,148 areas already booked in the Roadshow Tracker (pulled live via Google Sheets API).
Suburbs where the suburb name = the parent city (e.g., "Peterborough" inside Peterborough).
Matching uses exact, normalised, and cleaned name comparison. Slash-separated suburb names (e.g., "Frampton Cotterell/Winterbourne") are split and each part checked independently.
2,793
Villages
500-2,000 population
1,305
Towns
2,001-30,000 population
213
City Suburbs
Within 30k+ parent cities
Scoring: Every settlement is scored 0-100 based on demographic similarity to your top-performing areas, using a weighted composite of the features from the correlation analysis above. Higher score = demographics more closely match your best areas. Scores are relative rankings, not revenue predictions. All lists are exportable as CSV.
Small Settlements

Target Villages

Population 500-2,000. Your data shows small villages punch above their weight: Gosforth in Cumbria (pop 930) did £40k and Thornton (pop 335) did £39k. In small places, everyone knows what's happening and the whole community turns out.

#VillageCountyRegion PopulationScoreMedian Age % Owned Outright% Detached Nearest EventDistLast Visit
Towns

Target Towns

Population 2,001-30,000. Ranked by demographic similarity to your best-performing areas.

#TownCountyRegion PopulationScoreMedian Age % Owned Outright% Detached Nearest EventDistLast Visit
City Targets

Target City Suburbs

Named suburbs within cities of 30,000+ population that you haven't visited. Scored using the parent city's demographics.

#SuburbParent CityCounty RegionCity PopScore
How We Did This

Methodology

Data sources: WBV event data (2,312 events across 1,112 areas), ONS Census 2021 Built-up Area characteristics (6,975 settlements), English Indices of Deprivation 2019 (32,844 LSOAs aggregated to local authority level).

Matching: WBV area names matched to ONS BUA names using exact, normalised, prefix, and fuzzy matching. 983 of 1063 areas matched (92.5%%). Unmatched areas are mostly suburbs/neighbourhoods within larger cities.

Revenue metric: Commodity-normalised average revenue per event per area. Normalisation strips out gold/silver price movements to isolate operational performance.

Scoring model: Weighted composite of demographic features, weighted by their absolute Pearson correlation with normalised revenue. Score of 100 = maximum demographic similarity to top-performing areas. A score is a relative ranking, not a revenue prediction.

Limitations: Census 2021 data is a point-in-time snapshot (March 2021). ONS data excludes London. IMD 2019 data is aggregated to local authority level, losing LSOA-level granularity. Not all WBV areas could be matched. Target list scores predict demographic fit, not guaranteed event performance.