The ecommerce dream versus reality.
online shops formed in 2023 is still on the Companies House register
20.1 per cent (16,883 of 83,992) on 1 October 2026, against 30 to 35 per cent of cleaning companies and 45.4 per cent of all other companies: ninth-lowest of 192 industry codes.
Selling online is often sold as the easy route to working for yourself. The Companies House register tells a harder story. Online shops are now nearly one in ten new companies, and they leave the register faster than almost any other kind of business. Most are not going bust. They are being struck off.
Key findings
- 20.1%
of online shops formed in 2023 still on the register on 1 October 2026
Cleaning companies 30 to 35 per cent, depending on the code; all other companies 45.4. Ninth-lowest of 192 industry codes.
- 5.7 times
as many online shops formed in 2023 as in 2017
Every company type: 84,184 against 14,881 (the 83,992 above is private limited only). From 2.4 to 9.7 per cent of new companies.
- 1 in 4
online shops active and outside default addresses in September 2025 had gone a year later
24.7 against 9.8 per cent of other companies. By main code, first of 92 industry codes; electricians are 73rd, at 7.2 per cent.
- 0 of 400
sampled dissolved online shops went through liquidation or administration
A dissolved sample cannot hold firms still in liquidation. Two in three were struck off by Companies House, one in three by owners.
- 32.8 years
median age of a new UK-resident online-shop founder
Against 39.5 years for other new companies. 17.9 per cent were under 25, against 8.2 per cent.
- 17.8%
of online shops aged two to three vanished in a year, with shell-like companies removed
Against 9.9 per cent of other companies the same age, the age at which every shell filter applies.
Which figure is which
Show the basesHide
The online-shop rates on this page sit on different populations. Each one is right on its own base; do not compare figures across bases.
Survival of the 2023 intake
- 20.1%
- Formed in 2023, still on the register on 1 October 2026. Private limited companies listing SIC 47910 in any of their four industry codes, Companies House search counts: 83,992 formed. The 84,184 in the key findings counts Companies Act companies of every kind (excluding LLPs and limited partnerships), from the Companies House search on 1 October 2026.
- 9th of 192rank
- The same measure across every industry. Every industry code with 1,000 or more private limited companies formed in 2023, ranked lowest first by the share still on the register on 1 October 2026 (Companies House public search, any of the four codes). Most of the eight lower codes are other ways of trading goods, and a company can list more than one code.
- 26.7%chart
- Formed in 2023, not dissolved by 31 December 2025. The same private limited population, measured at a fixed age so each year’s intake is compared like for like.
- 34.5%
- UK-resident founders, outside shared postcodes. 2023 online shops on the register on 1 September 2024 whose founders all live in the UK and whose postcode is shared by fewer than 100 companies, still there on 1 September 2026.
- 30.2%UK founders
- Formed in 2023, every founder UK-resident, still on the register on 1 September 2026. Private limited online shops formed in 2023 whose founding persons with significant control all declared UK residence, including those dissolved before September 2024, whose founders were read from the Companies House API. 82,476 of the 83,992 are covered.
Gone within a year, September 2025 to September 2026
- 24.7%
- Gone within a year, 1 September 2025 to 1 September 2026. Companies of every director-run type, active and outside Companies House default addresses on 1 September 2025. SIC 47910 in any code.
- 25.9%
- League table. The same population, by main (first-listed) industry code only, so each company counts in one industry.
- 21.1%shells
- Shell test. The 24.7 per cent population after removing dormant companies, those that never filed first accounts when due, and those at postcodes shared by 100 or more companies. Under 21 months old, no first accounts are due yet, so only the address and postcode filters apply; the two-to-three-year figure (17.8 per cent) is the one every filter reaches.
Founder age
- 17.9%
- Founders under 25. UK-resident founders of companies formed in the year to 1 September 2025. The founder chart counts every founder, wherever resident, so its under-25 share is higher.
How many are still standing
Of every year’s new online shops, how many are still on the register today? Take 2023, the peak year for new online shops. 20.1 per cent are still there. Every group we compared did better: other retail 29.9 per cent; cleaning companies 30 to 35 per cent, depending on the code (general cleaning of buildings 30.3, other cleaning services 34.7); painting and haulage about a third; and all other companies 45.4 per cent, where the typical industry code kept 42.8 per cent. Across the 192 industry codes with 1,000 or more new companies that year, online shops rank ninth-lowest; most of the eight below them are other ways of trading goods, such as market stalls, wholesale and sales agents. Among 2023 companies on the register in September 2024 whose founders all live in the UK, online shops were the least likely of the nine groups compared to be there two years later.
Formed in 2023: share still on the register on 1 October 2026
For companies formed in 2020, online shops (22.2 per cent) are level with cleaning, takeaway and haulage companies; hairdressing (29.7), IT consultancy (34.5) and all other companies (34.4) are clearly higher.
More on this chartLess
Measured at the same age (the fixed-age chart), online shops formed from 2017 to 2020 were within about five points of cleaning companies (and of painting, in the download) and behind all other companies in every intake. They were the lowest of the nine groups in 2019 and in every intake from 2021, and from 2022 the gap widened sharply. Cleaning here is other cleaning services (81299), the cleaning code with every year in the study; general cleaning of buildings (81210), the largest cleaning code, kept 30.3 per cent of its 2023 intake (public search, 1 October 2026).
Same age, different intakes
Share of each year’s new private limited companies not dissolved by 31 December of the second year after formation. For 2023, that is 31 December 2025.
Of the 2023 intake, 26.7 per cent were not dissolved by the end of 2025, against 42.6 per cent of cleaning companies; the 2024 intake, one year on, shows the same pattern (56.7 against 69.9 per cent).
More on this chartLess
Much of the widening comes with a change in who set the companies up. Among online shops still on the register in September 2024 with a founder on record, the share with a founder living outside the UK rose from 10.2 per cent of the 2020 intake to 46.8 per cent of the 2023 intake (both counted in September 2024, so the older intake had longer to lose its quickest closers). Where every founder lives in the UK and the postcode is not crowded, online shops trailed cleaning companies by about five points in each of the 2021, 2022 and 2023 intakes over the same two years. The 2023 intake, and much of the 2022 intake, reached their first accounts deadline after 4 March 2024, when Companies House gained new powers over registered office addresses, so stricter enforcement may also play a part. Earlier intakes looked different. On a like-for-like random sample of companies that carry an industry code, online shops formed from 2010 to 2016 were within about four points of other companies, above or below, in every year (130 to 452 online shops sampled a year); the gap widens from the 2017 intake. The chart starts in 2017 because companies formed earlier that closed early often carry no industry code, which flatters every industry group in the full counts. The cohort CSV carries both measures.
Does the gap survive the obvious objections?
Three tests on the 2023 intake. Each row sits on its own base, so compare across a row, not down a column.
Every company formed in 2023
Still on the register on 1 October 2026
- Online shops
- 20.1%
- Other retail
- 29.9%
- Cleaning 81299
- 34.7%
- All other
- 45.4%
Only those that had filed a real set of accounts
On the register on 1 September 2025 with accounts that were not dormant or missing; still there a year later
- Online shops
- 76.2%
- Other retail
- 85.3%
- Cleaning 81299
- 82.9%
- All other
- 88.0%
Only UK-resident founders, outside shared postcodes
On the register on 1 September 2024, every founder UK-resident, postcode shared by fewer than 100 companies; still there on 1 September 2026
- Online shops
- 34.5%
- Other retail
- 40.8%
- Cleaning 81299
- 39.1%
- All other
- 51.2%
Online shops are the lowest of the nine groups compared in every row.
More on this chartLess
Among companies that had filed real accounts, 23.8 per cent of online shops were gone a year later, against 17.1 per cent of cleaning companies and 12.0 per cent of all other companies; only about half of the 2023 online shops still on the register in September 2025 had filed such accounts (54.8 per cent). Nearly half of the 2023 online shops on the register in September 2024 with a founder on record (46.8 per cent) had a founder living outside the UK, and fewer than one in ten of those companies (9.7 per cent) remained two years later, close to the rate for other retailers founded from abroad (11.0 per cent). Measured from formation, as the headline is, 30.2 per cent of the 2023 online shops whose founders all live in the UK were still on the register on 1 September 2026, so about seven in ten had closed; for UK-founded cleaning companies (other cleaning services) it was 34.2 per cent. General cleaning of buildings, which kept 30.3 per cent counting every founder, was not measured on this basis. Residence is self-declared, and none of this says why a company closed.
Show the figures
| Group | Formed 2023 | Formed 2020 |
|---|---|---|
| Online shopsSIC 47910 | 20.1% | 22.2% |
| Other retail (not online) | 29.9% | 25.0% |
| Other cleaning servicesSIC 81299 | 34.7% | 23.1% |
| Road haulageSIC 49410 | 34.1% | 23.1% |
| PaintingSIC 43341 | 35.7% | 25.1% |
| Takeaways and food standsSIC 56103 | 38.2% | 22.4% |
| IT consultancySIC 62020 | 42.8% | 34.5% |
| All other companies | 45.4% | 34.4% |
| Hairdressing and beautySIC 96020 | 47.0% | 29.7% |
| Formed | Online shops | All other |
|---|---|---|
| 2017 | 25.8% | 31.8% |
| 2018 | 27.0% | 35.2% |
| 2019 | 29.4% | 40.3% |
| 2020 | 22.2% | 34.4% |
| 2021 | 20.7% | 37.4% |
| 2022 | 19.1% | 39.7% |
| 2023 | 20.1% | 45.4% |
| 2024Too young to judge | 32.2% | 57.0% |
| 2025Too young to judge | 77.6% | 86.5% |
| Formed | OnlineOnline shops | Other retailOther retail (not online) | CleaningOther cleaning services | All otherAll other companies |
|---|---|---|---|---|
| 2017 | 57.3% | 59.2% | 52.4% | 61.8% |
| 2018 | 58.3% | 59.3% | 56.4% | 65.3% |
| 2019 | 55.7% | 62.9% | 57.1% | 66.0% |
| 2020 | 49.8% | 51.0% | 46.6% | 59.5% |
| 2021 | 40.9% | 44.6% | 43.1% | 57.5% |
| 2022 | 32.2% | 38.0% | 40.7% | 54.3% |
| 2023 | 26.7% | 36.3% | 42.6% | 52.8% |
“On the register” means not dissolved and not in an insolvency process on 1 October 2026. It includes companies already under a proposal to strike off. The fixed-age table counts a company as surviving if it was not dissolved by 31 December of the second year after formation. 2010 to 2016 are in the cohort CSV, flagged: industry codes are incomplete for those years. On a phone the fixed-age table shows online shops, other retail, cleaning and all other companies; every group is on a wider screen and in the CSV.
The disappearance league table
Share of companies active on 1 September 2025 that had gone a year later, by main industry code, for the 92 industry codes with at least 10,000 companies. Online retail is first, at 25.9 per cent. Adjusted for company age it is 17.4 per cent, still first and level with other non-store retail. Much of the raw gap reflects how young online shops are.
Bars show the raw rate; the dark tick shows the rate adjusted for company age. Active companies outside Companies House default addresses, by main (first-listed) industry code. Two non-trading codes, dormant companies (99999) and non-trading companies (74990), are left out of the chart but counted in the ranks; both are in the CSV. Comparators are the largest one-person service codes. General cleaning of buildings (81210) is the largest cleaning code; other cleaning services (81299) has fewer than 10,000 companies under its own main code.
Show the figures
| Industry | Gone in a year | Age-adjusted |
|---|---|---|
| Online shopsSIC 47910, rank 1 | 25.9% | 17.4% |
| Other non-store retailSIC 47990, rank 2 | 21.5% | 17.4% |
| Sales agents, mixed goodsSIC 46190, rank 3 | 20.5% | 15.5% |
| Clothing and footwear wholesaleSIC 46420, rank 4 | 20.0% | 15.6% |
| Takeaways and food standsSIC 56103, rank 5 | 18.9% | 13.6% |
| Clothing shopsSIC 47710, rank 6 | 18.6% | 14.7% |
| General cleaning of buildingsSIC 81210, rank 7 | 18.1% | 13.0% |
| General wholesaleSIC 46900, rank 8 | 17.4% | 14.5% |
| Other software publishingSIC 58290, rank 9 | 17.0% | 13.2% |
| Other general storesSIC 47190, rank 10 | 16.7% | 13.3% |
| Hairdressing and beautySIC 96020, rank 29 | 11.8% | 9.8% |
| IT consultancySIC 62020, rank 31 | 11.8% | 12.0% |
| LandscapingSIC 81300, rank 58 | 9.4% | 9.3% |
| Plumbing and heatingSIC 43220, rank 72 | 7.6% | 8.1% |
| Electrical installationSIC 43210, rank 73 | 7.2% | 7.8% |
Struck off, not bust
Official insolvency statistics count only formal procedures such as liquidations, administrations and voluntary arrangements, so they show few online shops failing. The register shows what happens instead. We read the filing history of 400 randomly sampled online shops formed from 2019 to 2024 and since dissolved. None went through liquidation or administration. All 400 were struck off, either by Companies House or at their owners’ request.
How 400 dissolved online shops left the register
- Struck off by Companies House
- Closed by their owners
- Liquidation or administration: 0%
Other dissolved companies, sampled the same way: 5 of 400 (1.25 per cent) went through liquidation or administration. A sample of dissolved companies cannot include companies still in liquidation, which can take years to be dissolved. On the register itself, 1,288 of the 207,770 private limited online shops on the 1 September 2025 file were in liquidation, and 117 of the 44,651 online shops formed in 2020 (0.26 per cent) were in liquidation or administration on 1 October 2026.
- 67.0%
of the 400 sampled online shops had been dissolved within 21 months of being formed, before a first set of accounts is normally due (other dissolved companies: 53.5 per cent)
- 77.5%
of the 400 never filed a set of accounts. Most of them (65.3 per cent of the sample) closed before their first accounts were normally due (other companies: 66.0 per cent never filed)
- 331
formal insolvencies of online retailers in England, Wales and Scotland, September 2025 to August 2026 (Insolvency Service). A separate count, not a share of the companies that left the register
Is it a young person’s game?
Online-shop founders are younger. Among UK-resident founders of companies formed in the year to 1 September 2025, nearly one in five online-shop founders (17.9 per cent) was under 25, against 8.2 per cent for other companies, and the median age at incorporation was 32.8, against 39.5. The age chart below counts every founder, wherever resident; the disappearance chart counts UK-resident founders.
Founders by age at incorporation
Companies formed in the year to 1 September 2025. All founders, wherever resident. 77,957 online-shop founders, 853,745 others.
Gone within a year, by founder age
1 September 2025 to 1 September 2026. Companies formed since 6 April 2016 with UK-resident founders. Each chart has its own scale.
In both groups, companies with the youngest founders disappear fastest, and online shops disappear faster at every founder age. Among UK-resident founders, online shops with a founder under 25 disappeared at 28.2 per cent in a year, against 20.3 per cent for other companies; with a founder in their forties, 18.8 against 11.0 per cent, before allowing for company age.
Founder age accounts for little of the online-shop gap. For UK-resident founders, online shops disappeared at 21.9 per cent in the year, against 12.7 per cent for other companies (the all-ages bars in the chart). Given other companies’ founder-age mix, the online rate falls only to 20.7 per cent, and with company age also held constant it is still about one and a half times the rate. Counting every founder, wherever resident, the gap is wider and the pattern is the same.
“Founder” means a founding individual person with significant control, usually someone holding over 25 per cent. Ages are known for companies incorporated from 6 April 2016. The disappearance figures cover companies formed since then that were active and outside default addresses on 1 September 2025, with every founder declaring UK residence, which is why the all-ages rate in the chart differs from the 24.7 per cent headline population. The download also carries the same figures for every founder, wherever resident.
Show the figures
| Founder age | Gone, online (UK founders) | Gone, other (UK founders) |
|---|---|---|
| Under 25 | 28.2% | 20.3% |
| 25 to 29 | 24.3% | 15.3% |
| 30 to 34 | 20.7% | 12.7% |
| 35 to 39 | 19.1% | 11.6% |
| 40 to 49 | 18.8% | 11.0% |
| 50 to 59 | 18.9% | 11.3% |
| 60 and over | 23.1% | 12.7% |
| All ages | 21.9% | 12.7% |
Not just shell companies
Are these just shell companies that never traded? We split every active company into four groups, applied in order. Online shops are more likely to sit at a postcode shared by 100 or more companies (37.2 against 28.9 per cent, after dormant and never-filed companies are taken out), and those vanish fast. But the online shops left after removing all three shell-like groups still disappeared faster than other companies of the same age.
| Group | Online gone | Other gone |
|---|---|---|
| At a Companies House default address | 83.6% | 65.4% |
| Dormant, or never filed first accounts when due | 27.7% | 15.5% |
| At a postcode shared by 100 or more companies | 28.8% | 8.1% |
| Everything else | 21.1% | 9.4% |
Groups are applied in order, so each company sits in one group only. “Never filed first accounts when due” only applies to companies old enough to have a first deadline. The postcode group is what remains at crowded postcodes after dormant and never-filed companies are removed.
Higher at every company age, shell-like companies removed
Gone within a year, by company age, after removing default-address, dormant, never-filed and crowded-postcode companies (the 21.1 per cent population). Under 21 months, only the address and postcode filters can apply.
The cleanest comparison is companies aged two to three years, old enough for every filter to apply: 17.8 per cent of the online shops left disappeared in the year, against 9.9 per cent of other companies. At five to ten years it is 7.2 against 4.7 per cent. Across all ages it is 21.1 against 9.4 per cent, but more than half of those online shops are under 21 months old, too young to have missed a first accounts deadline. Adjusted for company age, the all-ages rates are 14.5 and 9.5 per cent.
A stricter test: keep only companies whose last accounts on the register were a real set (micro-entity, abridged, full and similar), not dormant and not missing. Adjusted for company age, 8.4 per cent of those online shops disappeared in the year, against 5.1 per cent of other companies, about one and a half times the rate (11.3 against 5.1 per cent before allowing for age).
Show the figures
| Company age | Online, shells removed | Other, shells removed |
|---|---|---|
| Under 12 months | 27.5% | 18.7% |
| 12 to 18 months | 38.2% | 24.9% |
| 18 to 21 months | 33.9% | 22.3% |
| 21 to 24 months | 18.4% | 10.5% |
| 2 to 3 years | 17.8% | 9.9% |
| 3 to 5 years | 13.3% | 7.8% |
| 5 to 10 years | 7.2% | 4.7% |
| 10 years and over | 4.0% | 3.2% |
| All ages | 21.1% | 9.4% |
“Setting up the company takes an afternoon. Keeping it takes a habit.”
“Picture the advert: a laptop on a beach and a shop that runs itself. The register tells a plainer story. A cleaning company will rarely make anyone rich overnight, but it is more likely than an online shop to still be there about three years on. Most online shops that close are not wound up by creditors. They are quietly struck off. The sellers I see who last treat it as a business from the first week: books kept, fees and returns counted, numbers checked every month.”
Method, sources and dates
Sources
Companies House Free Company Data Product (Basic Company Data) of 1 September 2024, 2025 and 2026; Companies House People with Significant Control snapshots of 1 September 2024, 2025 and 2026; Companies House public API advanced-search counts taken on 1 October 2026 (01:05 to 01:41 UTC), and the same counts re-taken the same morning on the public Companies House advanced search, which anyone can use without a key; the filing histories of 800 randomly sampled dissolved companies; the persons with significant control records, read from the Companies House API on 1 October 2026, of the 13,146 online shops and other cleaning services companies formed in 2023 and dissolved before 1 September 2024; Insolvency Service company insolvency record-level data, August 2026 release (England, Wales and Scotland). The 1 October 2026 bulk file was not published when the analysis ran.
Online shop
A company listing SIC code 47910, “Retail sale via mail order houses or via Internet”, in any of its four industry-code fields. The league table and the insolvency count use the first-listed code only. Industry codes are self-declared.
Cohort survival
Private limited companies incorporated in each calendar year from 2010 to 2025, counted by status on 1 October 2026 through the Companies House API. The comparators (other retail, meaning any division 47 code but not 47910; other cleaning services 81299; painting 43341; hairdressing and beauty 96020; takeaways 56103; road haulage 49410; and, as an extra, IT consultancy 62020) are counted the same way, by any of their industry codes. A company is counted in every group whose code it lists, so the trade groups can overlap with online shops; the overlap is small (of the 2023 intake on the register in September 2024, 0.4 per cent of cleaning companies and 3.3 per cent of IT consultancies also listed 47910, and 1.7 per cent of online shops listed any of the six trade codes). Other retail and all other companies exclude online shops by definition. The fixed-age chart uses dissolution dates: a company counts as surviving if it was not dissolved by 31 December of the second year after the year it was formed, so every window closes by 31 December 2025. Companies formed before 2017 that closed early often carry no industry code, which flatters every industry group, so the charts start at 2017. For 2010 to 2016 the cohort CSV adds a like-for-like comparison from a random sample of each year’s companies (130 to 452 online shops a year), set against companies that carry an industry code; on it, online shops were within about four points of other companies, above or below, in every one of those years. General cleaning of buildings (81210) appears for the 2023 intake only, from the public search.
One-year disappearance
Companies of all director-run types with status Active on 1 September 2025 and not at a Companies House default address, classified on that file, then looked up in the 1 September 2026 file. “Gone” means absent from the later file, which in practice means dissolved. “Other companies” excludes online shops. Age adjustment is direct standardisation to the age mix of all companies in the same population. Accounts category: the same population, limited to companies whose last accounts on the 1 September 2025 file were a real set (not dormant, not “no accounts filed”, not “accounts type not available” and not blank), age-adjusted the same way.
Tests of the 2023 intake
Filed accounts: private limited companies formed in 2023 that were active or under a proposal to strike off on the 1 September 2025 file, whose accounts category on that file was a real set of accounts (not dormant, not “no accounts filed”, not “accounts type not available” and not blank), still active or under a proposal on the 1 September 2026 file. UK-resident founders: the same intake on the 1 September 2024 file, with a founding individual person with significant control notified within 30 days of incorporation, every founder declaring UK residence, and a registered postcode shared by fewer than 100 companies on that file, still active or under a proposal on 1 September 2026. Earlier intakes: the same filters, over the same two years, applied to the 2020, 2021 and 2022 intakes (in the downloads). From formation: the same founder rules applied to every 2023 online shop and other cleaning services company, with the founders of the 12,276 online shops and 870 cleaning companies dissolved before 1 September 2024 read one by one from the Companies House API on 1 October 2026; still on the register means active or under a proposal on the 1 September 2026 file. Residence is self-declared and is published only as UK or outside the UK.
How they left
Filing histories of 400 dissolved online shops and 400 other dissolved companies, randomly sampled from those formed in 2019 to 2024. A private company’s first accounts are normally due 21 months after incorporation (Companies House guidance, “Life of a company: part 1 accounts”). A sample of dissolved companies cannot include companies still in liquidation, so the live register is quoted alongside it.
Founder age
Age at incorporation of founding individual persons with significant control, from month and year of birth. Founders of companies formed in the 12 months before each snapshot; the 2024 and 2026 cohorts show the same pattern.
What this does not show
The register cannot say why a company closed, or whether it ever traded. Every finding here is a correlation, not a cause. Strike-off enforcement has varied over time, which affects when companies are dissolved; since 4 March 2024 Companies House has had new powers over registered office addresses. Compare cohorts with care.
Reproducibility
Every figure was produced by a script from the files above. The 35 headline online-shop figures were independently recomputed with separate code: 33 confirmed and 2 corrected, with the corrected forms used here. The fixed-age figures and the 2020 and 2023 counts for online shops, the five trades, IT consultancy and all other companies were re-taken on the public Companies House advanced search, which needs no key. The fixed-age figures and formation counts matched exactly; the counts still on the register were within 47 companies of the earlier run, taken a few hours before, and every published rate matched to one decimal place. Other retail cannot be queried there, so it rests on the API run. The UK-founder, postcode and founder-residence tests of the 2020 to 2023 intakes were computed twice, by separately written scripts reading separately extracted copies of the files: 77 of 80 counts were identical and the other three differed by one company. The filed-accounts and accounts-category tests, and the from-formation UK-founder figures, were computed once.
Running an online shop, or about to start one? See how we work as accountants for ecommerce sellers, from Amazon and Shopify to eBay and Etsy.