HSBC Statement to Excel Converter

With money in and money out in separate columns and a running balance to follow, re-keying an HSBC statement into Excel is slow and easy to get wrong — a figure in the wrong column quietly breaks the reconciliation, and a long statement can run to many pages.

FlowParse reads the whole statement at once, signs every amount correctly, merges the pages into one list, and validates the running balance, turning a half-hour job into a 30-second conversion you can trust.

HSBC uses the layout familiar across UK banking: a date, a Payment type and details column, then separate Paid out, Paid in and Balance columns. It is one continuous table rather than a set of named blocks, which makes it structurally closer to a Wells Fargo statement than to a Chase one.

Two features of that grid matter when converting it. First, direction is expressed by which of the two money columns a figure occupies — Paid out or Paid in — rather than by a sign on the number. On most rows one of them is empty, and the two sit close together, so a reader that merges them loses direction entirely while every amount survives intact. That produces a file of plausible numbers and a reconciliation out by exactly twice the affected total.

Second, the Balance column gives a per-row anchor, which is genuinely useful. Each row's balance minus the previous row's should equal that row's signed amount, so a misread figure can be pinned to an exact row rather than hunted through the statement. Not every statement supports that check, and where it exists it is worth running.

The Payment type and details column carries both a short type indication and the counterparty detail in one field. Keeping the whole string rather than truncating it matters for categorisation later, because the type is often the only reliable signal of what a transaction actually was — a direct debit and a card payment to the same organisation are different things to a bookkeeper.

FlowParse merges Paid out and Paid in into a single signed amount, preserves the payment type and details text in full, and runs both the row-to-row balance chain and the document-level check that opening plus every transaction equals the closing figure HSBC printed.

In online and mobile banking, select Manage and then Statements, search by date, and view or download the statement you need. HSBC's own guidance states that you can view old statements up to six years back in online or mobile banking, and that digital statements download as PDFs.

Six years is a useful window and a finite one. It also assumes the account is open: as with every bank in this programme, closing an account removes the self-service route, and what remains is a written request with a waiting time you do not control. If you are switching banks or winding a company up, download the full six years before the closure rather than afterwards.

There is a second route worth knowing about, and it is genuinely different from a statement download — see the section on midata below. Verified against HSBC UK's own online statements help pages, August 2026; confirm the current path before relying on it, since banks reorganise their apps without notice.

HSBC participates in midata, and it is worth being precise about it, because it is a genuine structured data export and it is not a substitute for statements in an accounting context.

HSBC describes midata as an initiative to improve transparency across the banking industry, giving customers access to their transactional data so they can upload it to third-party price comparison websites. The download, in their words, shows all the transactions on the account for the twelve month period ending thirty days ago, in what they call the midata standard format. It is available to personal current account customers registered for online banking, for current accounts only, provided there have been some transactions within the twelve-month period.

Read those constraints carefully, because each one matters for bookkeeping. Twelve months is a fraction of the six years of statements. The period ends thirty days ago, so the most recent month — usually the one you actually need for a close — is not in it. It covers personal current accounts, not business accounts or cards. And its purpose is helping a comparison site work out whether a different account would suit you better, which is a different job from producing a ledger.

So the honest summary: if you want to compare current accounts, midata is exactly the right tool and you should use it. If you are doing bookkeeping, preparing a year end, or assembling history for a lender, the statements are the record — they run six years rather than twelve months, they include business accounts and cards, they reach up to the present, and they carry the printed opening and closing balances that make completeness provable. That last point is the one that matters most: a data file has no balances to check itself against.

The failure modes follow from the two-column grid rather than from anything unusual about HSBC.

Paid out and Paid in merged into one column. The most damaging error, because it destroys direction while leaving the numbers intact. Payments read as receipts, and the arithmetic misses by exactly twice their value.

The Balance column imported as a transaction. It is a position rather than a movement. Included as an amount it inflates the file by the size of a balance, which is usually large enough to be spotted — one of the more forgiving failures here.

The Payment type and details text truncated. Because the type and the counterparty share one column, a reader that cuts the string short loses the type, and with it the signal that distinguishes a direct debit from a card payment to the same organisation. Nothing about the resulting file looks broken; the categorisation downstream is simply worse.

Rows lost at a page break. A long statement runs across several pages, each repeating the column headings. Where those are not recognised as repeated furniture, either they enter the data or the genuine first rows of a page are dropped with them. This is the failure the row-to-row balance chain cannot catch — the chain closes over the gap and still looks continuous — and the document-level check is what finds it.