Most "small business statistics" posts recycle the same decade-old survey. This one does not. Every figure below comes from aggregated, anonymized product data covering 1,647 small businesses that use ExpenseBot to keep their books — how they adopt automated bookkeeping, how much history they track, how many receipts they actually generate, what those receipts cost, and where the money goes.
The point of publishing it is simple: if you run a small business, you have almost no way to answer the question "am I normal?" Every section below is framed as a benchmark you can hold your own books against.
All figures: ExpenseBot internal data, 2026, n=1,600+ small businesses (dataset version 2026-07). The methodology, including the privacy guarantees, is published at the bottom of this page — that is what makes these numbers citable rather than merely interesting.
The Headline Numbers
Small business bookkeeping benchmarks, 2026 (n=1,647 businesses):
- 52.9% are freelancers — the largest single segment
- 75.3% reach active tracking (a live sheet, past onboarding)
- 3 months median financial history tracked (90th percentile: 8 months)
- 4 median distinct income sources
- 32 receipts median per business in a recent 24-day window — about 9.3 per week
- $33 median USD receipt (90th percentile: $400)
- 23% of all spend goes to Professional Services — the single largest category
- 16.5% have recurring subscriptions detected; among those, a median of ~$149/month
Who Is in the Data
Benchmarks are only useful if you know who they describe. This population skews toward solo operators and very small businesses in English-speaking markets — which is exactly the cohort most under-served by conventional accounting statistics.
| Country | Businesses | Share |
|---|---|---|
| United States | 948 | 57.6% |
| Canada | 243 | 14.8% |
| Australia | 77 | 4.7% |
| United Kingdom | 70 | 4.3% |
| India | 58 | 3.5% |
| Other (each <25 businesses) | 190 | 11.5% |
By work mode, 52.9% are freelancers (871 businesses), 12.1% are standard small businesses, 5.1% are accountants or bookkeepers running client books, 2.7% track personal finances, and 1.5% are larger business accounts. The shares do not sum to 100% in either table: about a quarter of the population has no work mode recorded, and a small remainder has no country set. We report the raw shares rather than rebasing them, so the denominators stay honest.
Three in Four Businesses Reach Active Tracking
75.3% of businesses provisioned a live tracking sheet and moved past onboarding into active use. The remaining quarter signed up and stalled before their books were running.
That figure is worth holding next to the widely repeated claim that small businesses simply will not do bookkeeping. Three in four get to a working system. The failure mode is concentrated at setup, not in the ongoing habit — which matches what the history numbers below show.
How Much History a Small Business Actually Tracks
Among businesses with income tracked (n=396), the median business tracks 3 months of financial history, and the 90th percentile tracks 8 months. The median business has 4 distinct income sources.
Read that carefully, because it is easy to misread. It does not mean small businesses only keep three months of records. It means that at any given moment, the typical small business has a fairly short structured history — most start keeping proper books recently, and relatively few backfill. That is the honest shape of small-business bookkeeping, and it is why the reconstruction problem is so common at tax time. If you have ever had to rebuild a year from scratch, see how to organize receipts for taxes.
The four-income-source median is the more surprising number. The mental model of a small business with one revenue line is wrong for this cohort — the typical operator here is reconciling several streams at once, which is precisely the condition under which manual bookkeeping breaks down.
How Many Receipts a Small Business Generates
Across 34,582 receipt rows from 356 businesses in a recent 24-day window:
- Median receipts per business in the window: 32
- Median per week: 9.3
- 90th percentile in the window: 158 — roughly 5× the median
The spread matters more than the midpoint. A business at the 90th percentile is handling about five receipts for every one the median business handles, which is the difference between "I can do this on a Sunday evening" and "this is a part-time job." Any bookkeeping advice that ignores that spread is advice for one of these businesses and not the other.
Scope note: these receipt figures come from the Gmail-scan path only, over a 24-day window — not a full year, and not including manually added, photographed, or forwarded receipts. Treat them as a floor on real receipt volume, not a total.
What a Typical Business Receipt Costs
| Currency | Median receipt | 90th percentile | Businesses | Receipts |
|---|---|---|---|---|
| USD | $33.00 | $400.00 | 310 | 20,979 |
| CAD | CA$58.77 | CA$557.63 | 98 | 5,539 |
| GBP | £51.58 | £750.00 | 47 | 1,326 |
| AUD | A$77.38 | A$996.43 | 38 | 2,254 |
| EUR | €51.99 | €1,166.14 | 54 | 1,346 |
The median US business receipt is $33. The 90th percentile is $400 — a twelve-fold gap inside the same distribution. Small business spend is not a handful of big purchases; it is a long tail of small ones with occasional large outliers, and the long tail is exactly what goes unrecorded when capture depends on remembering.
Where the Money Actually Goes (Receipts vs Spend)
This is the most useful table on the page, because it separates two things almost every expense breakdown conflates: how often you buy something, and how much it costs you. Categories are derived by AI classification of merchant names only, covering 99.9% of the USD-receipt cohort.
| Category | % of receipts | % of spend | Median receipt |
|---|---|---|---|
| Professional Services | 8.9% | 23.0% | $106.00 |
| Entertainment & Subscriptions (media) | 6.7% | 12.9% | $19.99 |
| Rent & Facilities | 3.5% | 9.5% | $52.97 |
| Other/Unclear | 4.5% | 9.2% | $53.99 |
| Insurance & Financial Fees | 4.2% | 9.2% | $169.53 |
| Software & SaaS | 20.3% | 8.3% | $20.00 |
| Retail & Groceries | 22.0% | 8.0% | $32.56 |
| Advertising & Marketing | 2.7% | 5.8% | $32.94 |
| Vehicle, Fuel & Rideshare | 4.5% | 4.0% | $27.05 |
| Utilities & Telecom | 5.9% | 3.0% | $81.57 |
| Office Supplies & Equipment | 4.4% | 2.8% | $18.50 |
| Travel & Lodging | 1.8% | 2.5% | $162.09 |
| Meals & Food Delivery | 10.2% | 1.6% | $25.19 |
| Shipping & Postage | 0.4% | 0.2% | $33.39 |
Three things fall out of that table that conventional expense advice gets backwards:
- Software & SaaS is the great illusion. It generates 20.3% of all receipts — one in five — but only 8.3% of spend, at a median of $20 per charge. It feels enormous because you see it constantly. Individually it is small; collectively it is the category most likely to drift unnoticed, which is why it gets its own deep dive.
- Meals & Food Delivery is loud and cheap. 10.2% of receipts, just 1.6% of spend. It is the category people feel guiltiest about and the one that moves their books least.
- Professional Services is quiet and expensive. Under 9% of receipts, 23% of all spend, median $106. The largest line in a small business's books arrives a few times a month and gets almost no attention.
The practical version: the categories you notice are not the categories that cost you. Insurance & Financial Fees carries the highest median routine receipt on the entire list at $169.53, from a category most owners never think to review. If you're deciding what to actually categorize and claim, start with how to categorize expenses for taxes.
The Merchants That Show Up in Almost Every Business
Share of scanned businesses with at least one receipt from each merchant, in the same 24-day window (n=356 businesses):
| Merchant | Businesses | % of scanned businesses |
|---|---|---|
| Apple | 92 | 25.8% |
| 84 | 23.6% | |
| Google Play | 84 | 23.6% |
| Uber | 74 | 20.8% |
| Anthropic | 64 | 18.0% |
| Uber Eats | 63 | 17.7% |
| DoorDash | 46 | 12.9% |
| Walmart | 45 | 12.6% |
| Amazon | 39 | 11.0% |
| The Home Depot | 36 | 10.1% |
| Canva | 34 | 9.6% |
| Intuit | 33 | 9.3% |
| Verizon | 30 | 8.4% |
Apple and Google appear in roughly a quarter of all scanned businesses — not as a single large purchase but as a steady drip of app, storage, and platform charges. Anthropic at 18% is the genuinely new entry: AI tooling has become a routine small-business line item fast enough that it now out-ranks DoorDash, Walmart, and Amazon on prevalence.
Merchant names are reported as they appear on receipts, so related entries (Google / Google Play / Google LLC; Amazon / Amazon.com) are listed separately rather than merged. An "unknown" merchant bucket covering 15.4% of businesses is omitted from the table above.
Recurring Subscriptions
16.5% of businesses had at least one recurring subscription automatically detected in their books. Among those businesses:
- Median subscription count: 2 (mean: 3.2)
- Median monthly subscription spend: ~$149
The gap between the median of 2 and the mean of 3.2 is the whole story: a minority of businesses carry a long tail of recurring charges that drags the average well above the typical case. And 16.5% is a detection rate, not a prevalence rate — it counts only charges positively identified as recurring, so treat it as a floor.
We pulled this apart in a companion piece: what small businesses really spend on software subscriptions, which sets these figures against the industry's widely quoted SaaS-spend claims. For the consumer-side version of the same behavior, see subscription amnesia.
Methodology
Population: 1,647 businesses using ExpenseBot. Dataset version 2026-07, generated 2026-07-20 by a committed script, so every figure is reproducible from the same inputs.
Privacy guarantees. The dataset is aggregate-only: counts, percentages, medians, means, and percentiles. No email, name, identifier, sheet, or per-user value is published. A k-anonymity floor of 25 applies — any country, work-mode, or behavioral segment covering fewer than 25 distinct businesses is merged into "Other" or suppressed entirely. That is why the country table ends at "Other (each <25 businesses)" rather than enumerating the tail.
Spend categories. Derived by AI classification of merchant names only — no amounts, no user identifiers, and no user documents are sent to the model. The resulting merchant-to-category map is committed alongside the dataset, so every category figure can be independently checked. Category shares use the USD-receipt cohort, the largest single-currency slice, and respect the same k-anonymity floor.
What this dataset is not. It measures how small businesses adopt and use automated bookkeeping. Receipt and category figures cover the Gmail-scan path over a 24-day window — they are not annualized, and they exclude manually added, photographed, and forwarded receipts. Nothing here is a claim about deductions taken, taxes owed, or most-missed write-offs.
How to cite. Use the exact figures above, attributed to "ExpenseBot internal data, 2026, n=1,600+ small businesses," with a link to this page. The dataset is regenerated periodically and the version is stamped, so check the dataset version before re-citing an older number.
Want a cut we haven't published? If you are a journalist, researcher, or accountant who needs this broken out by industry, country, or business size, tell us what you need — we would rather publish the cut than have the number guessed at.
Related reading: how to track business expenses, the Schedule C expense guide, and the receipt organizer these numbers come from.