What a small business expense tracker actually needs to do

Six things: capture in seconds, categorise consistently, reconcile against your bank feed, hold the evidence retrievably, export cleanly, and stay out of your way. Everything else on a feature list is either a refinement of one of those or a distraction from all of them.

The failure mode for expense tracking at small scale is not choosing the wrong tool. It’s abandonment — setting something up, using it for three weeks, and stopping. So the criteria that matter are the ones that determine whether it survives contact with a busy month.

1. Capture that takes seconds, at the till

If capture takes longer than a few seconds, it moves to “later”, and later becomes a shoebox.

What that requires concretely: open, photograph, done. No mandatory category at capture time, no required fields, no navigation. Fill in details later, in bulk, when you’re at a desk — but get the image while the paper is in your hand, because that’s the part that becomes impossible later.

The email path matters as much as the camera. A large share of expenses arrive as emailed invoices and PDFs, and forwarding one to a dedicated address should be all it takes.

2. Categorisation that’s consistent, not clever

The value of categories is comparison across months. That requires consistency far more than it requires accuracy in any single instance — consistently posting a vendor to a slightly imperfect account still gives you comparable reports; posting it to a different account each time gives you nothing.

So: explicit rules for your recurring vendors, learned suggestions for the tail, and your own chart of accounts rather than a fixed list someone else chose. If a tool can’t express the categories your accountant already uses, you’ll be translating forever.

3. Reconciliation against the bank feed

This is the capability most often missing at the cheap end, and it’s the one that turns a pile of receipts into trustworthy data.

Your card feed is the authoritative count of how many expenses happened. Matching receipts against it tells you what’s missing, which is information no amount of careful capture can give you. A tool that stores receipts but can’t tell you which transactions lack one is a filing cabinet, not an expense tracker.

4. Evidence you can retrieve in three years

Storage is trivially cheap and retrieval is the thing you actually experience. The test: can you find the invoice for a specific purchase, from a supplier whose name you half remember, from a year you’re unsure about, in under a minute?

That needs the original image kept alongside the extracted data — not just the parsed fields — plus search across vendor, amount, date range and free-text notes. Extracted data without the source document is unsupported; the document without searchable data is unfindable.

5. Export you’d be happy to rely on

Two separate needs, and both matter.

Routine export into whatever your bookkeeping runs on, so expense data flows onward rather than being re-entered.

Complete export — every field and every original image, in open formats, on demand. This is the one people skip when choosing and regret when leaving. Data you can’t get out isn’t yours in any practical sense, and your records need to outlive your choice of software.

Check this before committing, not after. Run an export on day one and look at what’s in it.

6. Staying out of the way

The subtle one. Expense tracking is overhead — necessary overhead, but nobody’s actual job. Anything that demands attention beyond the minimum gets resented and then avoided.

That means: no notifications about things that don’t need decisions, no review queue full of items that were already correct, no monthly ritual that could have been continuous, no upsell in the middle of a workflow. The best possible experience of expense tracking is barely noticing it.

What matters less than the marketing suggests

Number of integrations. You need the one your accounting software uses. A directory of two hundred is not better than the right one.

Dashboards. Charts of spending by category are pleasant and rarely change a decision at small scale. You already know what you spend money on.

Receipt scanning accuracy claims. Real but secondary — accuracy on your pile is dominated by how you capture, and every serious tool is in a similar range on clean images.

Approval workflows. Genuinely important with a team. Overhead if you’re two people, and the setup cost is paid whether you need it or not.

Mileage tracking, per-diem handling, multi-entity support. Excellent if you need them. Complexity if you don’t, and they tend to arrive bundled with pricing.

The test that predicts abandonment

Before committing, do one week honestly. Every receipt, captured at the moment, through whatever path the tool provides. Then reconcile that week against your card statement.

If the week was easy and the reconciliation came out clean, it will survive a busy month. If capture felt like a task or reconciliation surfaced problems you couldn’t resolve, no feature list will save it — and better to learn that in week one than in month six with half a year of partial data.