What new British companies are betting on
Every company in Britain has to register a name and say what it does, and all of it is public. So I went through 8.4 million companies to see what people are starting businesses in right now.
Fewer people are starting companies to make things, sell clothes, hire staff or run care homes. More are starting them to write software, own property, give advice, or do something you have to turn up for, like a padel court, a café or a garage.
Here's each trade's share of new companies, in the last 12 months against 2021 to 2023.
- Software
2.0% to 5.4% - Holding companies
1.8% to 3.2% - Sports and gyms
0.4% to 0.7% - Taxis
0.5% to 0.7% - Consultancy
3.6% to 4.5% - Car repair
1.1% to 1.3% - Cafés and takeaways
2.9% to 3.6% - Property
8.0% to 9.6% - Online shops
8.3% to 7.8% - Making things
3.8% to 3.3% - Hiring
1.2% to 0.9% - Clothes
2.1% to 1.5% - Care homes
0.9% to 0.6%
Software is the biggest change by far, and the word AI is in 5 times as many names. That's a story of its own. The one I didn't expect is holding companies, up from 1.8% of new companies to 3.2%. Add property, and about one new company in eight exists to own something rather than sell anything.
Names can mislead, though. The word Online is down 45%, but online shops are still one new company in 13, so I'd put that down to fashion.
Search every company name in Britain
Names give you far more detail than trades, so I built a search. Each line counts a word per 10,000 new companies, and the map shows where those companies are.
Where Padel companies are. Darker means more common than you’d expect.
Most common, for its size
- Sutton4.2×
- Bromley3.3×
- Warrington3.0×
- Stockport2.9×
What they say they do
- 49%
- 10%
- 8%
Still alive
60% of those formed in 2023, against 44% of all companies.
A lot of the booms are places you go to
Padel is up about 10 times, reformer pilates 11 and saunas 8. It isn't just a fashion in names, because new companies calling themselves sports venues are up 2.8 times.
The rest is a stranger mix. Each chart shows a word since 2019, on its own scale, and the number is the last 12 months against 2021 to 2023.
Not all of the padel companies have got far: the word has been on 1,640 live companies, in a country with 682 padel venues.
TCG means trading card games, and it's still climbing. Spud is jacket potatoes, I'd guess, since 104 of the 121 new ones sell food.
Health is split: ADHD is up 2.7 times, while Healthcare is down 44%. Peptides are up the most, from almost nothing, and I don't know what most of those companies are selling.
Advisory is up 4.5 times, but the consultancy trade is only up 27%, so I'd guess a lot of that is consultants picking a grander word.
The newest is air conditioning. Words like aircon and cooling went up 2.9 times last quarter, during what looks like the hottest summer on record. It's only one quarter, but it shows how fast people move.
A typical fad lasts about two years
Most of these won't last. Here are the clearest fads I could find since 2010, in the order they peaked, each on its own scale. I picked them by hand, so treat the two years as a rough guide.
Padel is at the bottom because it may not have peaked. Pilates is already down 23% from last year's high, and the lockdown side hustles have mostly gone: Candles is down 59% and Handmade 66%.
Padel is in the suburbs and saunas are by the sea
Names tend to mean what they say. Surf is 44 times as common in Truro as across the UK, so a map of a word is a rough map of the trade. Here are three recent booms, darker where the word is more common than you'd expect.
Padel is strongest on the edges of big cities, in places like Sutton, Warrington and Stockport. Saunas lean towards the coast, and darts is 15 times as common in Dudley as across the UK.
Contains OS data © Crown copyright and database right 2020 and 2023. Contains Royal Mail data © Royal Mail copyright and database right 2020. Source: Office for National Statistics licensed under the Open Government Licence v.3.0. Postcode area boundaries by Mark Longair.
Names ending in -ora have jumped on three continents
Here's an odd one. From 2015 to 2023, about 5 in every 10,000 new British companies had a made-up name ending in -ora. Last quarter it was 46, not far off the 51 with AI in the name.
This chart counts those names, along with the same thing in Australia and Colorado, and other made-up endings like -ana and -ica in grey.
Names ending in -ora, per 10,000 new companies
They land on the same words too. Nexora barely existed anywhere before 2024, and it's now been used by 193 companies in Britain and 57 in Australia.
It's also what you get if you ask a chatbot for a unique, brandable, made-up name. gpt-oss, a model OpenAI gives away, ended 66 of its 1,688 ideas in -ora, including Nexora. So I'd guess that's part of why, though I can't prove it, and the line sat still for 18 months after ChatGPT came out.
More than half of 2023's companies have already gone
It costs £100 to start a company, so most of these are small bets, and most don't last. Here's how many were formed each quarter, split into the ones still on the register and the ones that have gone.
The recent bars only look healthy because those companies haven't had time to close. Companies House's own count was 868,000 in 2023 and 822,000 in 2025, so new companies aren't really surging. That's why I've used old copies of the register and shares rather than raw counts.
The word in the name is a decent guide to who lasts. Here's the share of 2023's companies still here, by word. The faint line is the average, 44%.
Mostly gone
- Jobs
9% - Ecom
11% - Crypto
13% - Mart
14% - Shopping
16% - Cleaning
30%
Mostly still here
- Architects
89% - Freehold
89% - Dental
88% - Psychology
87% - Solicitors
77% - Holdings
73%
So a list of names won't tell you what's going to work. It's more like a poll, where a vote costs £100. Right now the votes are going to software, to property and holding companies, and to things you have to turn up for. They're drying up for factories, clothes and hiring.
Apart from software, what most of the risers have in common is that you can't get them from abroad. Nobody can import a padel court, an MOT or a flat white. It isn't a clean split, because care homes can't be imported either and they're shrinking. But I'd wager people have noticed which businesses are hard to undercut, whether by a factory overseas or by a chatbot.
The survival numbers point the same way. The names that last took something to get started, like a dental degree, and the ones that vanish are the ones anyone can set up in an afternoon. So I wouldn't read too much into this year's names. The better question is which of them are still around in three years.