The Colour of Enterprise
Mapping business activity in England & Wales
As I write this we're all in lockdown.
Swathes of the workforce are now sat at home in Covid-19 imposed isolation and previously bustling public spaces are all but empty. While absorbing the quiet, it made me wonder what is it we normally do all day?
There's probably a thousand ways to answer that question but a dataset I know well is Companies House registrations. It tells us where and what our businesses do.
Read on for the method, but this is the visualisation I ended up with from the data:
This point map plots 2.1m registered office addresses across England & Wales. Bubble size shows the total registered businesses in an area. Colour denotes the most popular SIC (activity) code.
It's worth pointing out what this does and does not represent.
The map shows the most popular activity in any given location. It doesn't mean that's the only activity in a location so we shouldn't infer too much about the local economy as a whole.
In short, it's a "flavour" of what businesses do.
If you're on a desktop, it's worth visiting the interactive version. This throws up some remarkable insight by allowing us to see local activity by category.
Knowledge based businesses and those involved in transportation proliferate through the spine of the country and construction firms seem to cluster on the edge of cities. The City of London is clearly visible as a financial hub - but this isn't the only cluster of money movers. Meanwhile builders from Essex seem to have been very busy building London.
Hospitality, real estate, manufacturing and agriculture have a much smaller footprint.
Areas that rely on hospitality dot the coast and agriculture only appears in low density zones to the north. Estate agents are nearly invisible but for two enormous clusters in North London. If there's a housing downturn, expect the pain to be disproportionate here.
Be careful how you read these clusters though: They don't show there are low numbers of these types of firms across the country. It means there a very few places where these industries form the top activity.
The problem that consumed most time was choosing how to group business activities. Companies are categorised by SIC code (standard industrial classification) and whilst multitudinous, these codes are grouped into 12 broad categories.
I wanted this map to be super easy to read so further grouped categories where it made sense. For example Code G (Motor trade) and Code H (Transportation & Storage) both cover people who move things. They are grouped simply under "Transport" on my final visualisation.
Code J (information & communication) and Code S (Services) took a painful cut. These are enormously broad categories and would've resulted in a map that simply showed "Services" as our prime business activity.
This would've been a fair picture: We live in a service-based economy. But since it offered no insight and no "flavour of business", I made the decision to remove these categories in the interests of trying to tease out a picture of local specialities.
I think the insight in the final map confirms it was the right decision.
Making the visualisation
There are two noteworthy techniques in this visualisation.
The first is geocoding company addresses at scale. For this I joined postcodes in the Companies House data with National Statistics Postcode Lookup, which turns postcodes into a latitude and longitude.
They're both large datasets so I used Snowflake to do the heavy lifting.
The second is the dotted grid effect on the map. This was much simpler: I rounded the latitude/longitude of each business to one decimal place and plotted a standard point map. Rounding pushes the plot into a matrix effect. Nice.