How to merge and review bank statements

What you need

Nothing more than the statements themselves. You can mix months, accounts and even different banks in one batch; digital PDFs and scanned or photographed statements both work; and there's no template to set up or per-bank configuration. A free account lets you try the flow on a few files, and higher plans raise the number you can merge into one workbook.

A good practice is to gather everything for the period first — a full year across each account, plus any cards — so the consolidation is complete in one pass rather than assembled piecemeal. If you keep statements in cloud storage, you can pull them in from there; if they're on disk, drag the folder in. The more complete the batch, the more useful the single reconciled workbook at the end.

flowparse.io

Step 1 — Upload your statements

On the batch page, drop up to 100 PDF statements at once. Extraction begins immediately and runs in parallel as the files upload, so by the time the batch finishes the data is ready to merge. You can add or remove files before merging, and only files that finished processing are included.

There's no need to sort or rename anything first. The source-file name is captured automatically and stays attached to every row that comes from it, so a tidy filing system isn't a prerequisite — the traceability is built in regardless of how the files are named.

Step 2 — Extraction reads every row

Each statement is read at the document level, which is what prevents silent row loss. Rather than chopping a page into fragments at every subtotal or marker, the engine models the statement's columns and streams every data row into them, preserving continuation rows that span a page break and sections without a repeated header. A row that carries a date, amount or identifier is never mistaken for a header.

The practical effect is that the table you're about to merge is complete — not a plausible-looking subset. This is the change that took a hard ten-document test set from losing 50–88% of rows to full fidelity, and the story behind it is in how we eliminated row loss. You don't have to do anything for this step; it's just what makes the rest of the workflow trustworthy.

flowparse.io

Step 3 — Click merge to open Merge Review

When you click Merge into one Excel, FlowParse doesn't dump a file to your downloads — it opens Merge Review, an editable grid of the consolidated data. At the top sits a quality score; throughout the grid, any cell the system is unsure about is highlighted, and an issues panel lists every flag with a one-click jump to the exact cell.

This is the pivot from “hope it worked” to “confirm it worked.” Everything you need to trust the merge is on one screen, and you decide when it's ready to export — nothing leaves until you say so.

Step 4 — Read the quality score

The quality score is your at-a-glance verdict on the consolidated set. A high score with no flags means the statements extracted cleanly and reconciled, and you can move quickly to export. A lower score points you to where the attention is needed — usually a scanned file with a few uncertain characters, or a statement with an unusual date format.

Treat the score as triage, not a grade to chase. Its job is to tell you whether this batch is “clean, export it” or “a handful of things to check first,” so you spend your time proportionate to the actual risk. Over a year of batches, that's what keeps consolidation a quick task rather than a careful re-read every time.

flowparse.io

Step 5 — Fix the flagged cells

Click an entry in the issues panel and the grid jumps straight to the cell. The usual flags are a date that didn't parse cleanly, an amount that came through as text rather than a number, or a blank where a value should be. Correct it in place and the score updates; move to the next flag. Because the panel takes you to each problem directly, you never scroll a thousand rows hunting for trouble.

The discipline here is the same one FlowParse applies to a single document's editable preview, now over the whole merged set: the numbers that get exported are the numbers you approved. A few corrections on the genuine exceptions, and the consolidated workbook is verified rather than assumed.

Step 6 — Confirm the balances reconcile

Beyond individual cells, the strongest check is structural: each statement's balance reconciliation. Opening balance plus the sum of that statement's transactions must equal the closing balance. If it does, nothing was dropped or duplicated in that statement; if it doesn't, the break is flagged so you can find the missing or misread row.

This is the check that catches the error a tidy-looking grid would hide. A statement can look perfectly formed and still be missing a transaction — but it can't reconcile while it is. Confirming the balances pass across the batch is the moment you know the consolidation is complete, not just plausible.

flowparse.io

Step 7 — Resolve overlaps and duplicates

Gathering a year from several exports often means statement periods overlap, which would otherwise double-count the shared transactions. Duplicate detection flags those repeats so the consolidated file counts each transaction once. In Merge Review you can confirm the flagged duplicates or, in the rare case of two genuinely distinct transactions that look alike, keep both.

This matters most at the seams between statements — the last few days of one month and the first few of the next, or two accounts that both record an internal transfer. Reviewing the flagged overlaps there is a quick way to be sure the year's totals are right rather than inflated.

Step 8 — Export to your accounting format

Once the score is clean, the flags are cleared and the balances reconcile, export the verified set. Nine formats are available from one place: clean Excel and CSV, accounting CSVs for QuickBooks, Xero, DATEV and 1С, and real .QBO/.QFX/.OFX bank-feed files — each shown with its real software logo so you pick the destination, not a file type.

Because the consolidated set is already normalised to one schema, that single export carries the full year across every account into your ledger in one import — rather than twelve files loaded and reconciled separately. For the format trade-offs, see CSV vs QBO for QuickBooks import.

Merge Review flags at a glance

Almost everything Merge Review surfaces falls into a handful of categories, and each has a quick, predictable fix. Knowing them in advance turns the review into a checklist rather than a puzzle — you recognise the flag, jump to the cell, and resolve it in seconds.

The pattern across all of them is the same: the system has already located the problem, so your job is to confirm and correct, not to search. That's why a year of statements can be reviewed in minutes — the flags do the finding, and there are rarely many of them on clean digital exports.

Which export format should I choose?

The right export depends on where the data is going next. If you want to analyse or share, Excel or CSV is simplest; if it's going into accounting software, a native bank feed or an accounting CSV imports more cleanly and, in the case of the bank-feed formats, protects you from double-posting on a re-import.

When in doubt for QuickBooks, a .QBO bank feed is the safest choice because the transaction IDs let QuickBooks recognise rows it already has. For a deeper comparison of the trade-offs, see CSV vs QBO for QuickBooks import.

Merge vs converting each statement separately

You could convert each statement on its own and stitch the outputs together afterwards, and for a single file that's exactly the right tool — bank statement to Excel does it in one step. But across a year and several accounts, converting one at a time leaves you holding the hard part: many sheets with different column orders, overlapping date ranges that double-count, and no single source of truth to reconcile against.

Merging does the alignment, de-duplication and source-tracking for you, and Merge Review verifies the combined result in one place rather than statement by statement. The practical test is simple: if you only need one statement as a spreadsheet, convert it; if you need a complete, reconciled year you can hand to an accountant or a lender, merge and review. The consolidate page covers the consolidation angle in more depth, and the batch converter is where larger jobs live.

Merging across many banks and accounts

Real consolidation rarely involves one tidy account. The canonical column matching is what makes a mixed batch work: the engine recognises that “Debit”, “Money Out” and “Withdrawal” mean the same thing, collapses same-meaning columns into one, reconciles debit and credit into a signed amount, and preserves any genuinely bank-specific column rather than dropping it.

Because every row keeps its source file, you can still tell the accounts apart after merging — filter by account, group by currency, or roll everything up into one view of the year. Multi-currency statements keep each transaction's original amount and currency, and the merge deliberately avoids inventing a cross-currency total, so the workbook reflects the data rather than an arithmetic artefact. The consolidate page covers the cross-bank picture in more depth.

flowparse.io

Merging scanned statements

Scanned, photographed and image-only PDFs are handled in the same flow. They run through OCR first, then the same document-level structuring, balance validation and Merge Review as digital statements, with confidence scores on anything the OCR was unsure about. The balance check still applies, so an OCR slip that drops or garbles a row breaks the reconciliation and surfaces in review rather than passing silently.

Image quality matters more here — clear, straight, well-lit scans read best — but the safety net is identical. A batch that mixes digital exports and phone photos of older statements consolidates into one workbook, with the scanned files simply more likely to carry a flag or two for you to confirm.

Common problems and how to fix them

In nearly every case the fix is a single cell, because the document-level model and the balance check have already done the hard work of finding the problem for you. The job in Merge Review isn't to hunt — it's to confirm and correct the few things the system has surfaced.

Why a review step beats blind trust

It would be faster to skip review and download a file — but with money, fast and wrong is the worst outcome. The value of Merge Review is that it makes verification cheap: instead of choosing between trusting an unchecked file or re-reading a thousand rows, you check only the exceptions the system flags, in a minute or two. That's the division of labour that lets a junior process volume while a senior reviews just what's uncertain.

It also changes what you can stand behind. A consolidated workbook you reviewed — where the balances reconcile and every row traces to its source — is something you can hand to an accountant, a lender or an auditor with confidence. The few minutes in review are what turn “here's the data” into “here's the verified data,” and that's the whole point.

flowparse.io

A worked example: consolidating a year

To make the workflow concrete, picture a small business closing its year. There are three accounts — an operating account, a savings account and a company card — and none are fully on a bank feed, so you have a folder of roughly thirty PDF statements: twelve months for the operating account, twelve for the card, and a handful of quarterly savings statements. A few of the older ones are scans rather than digital exports. This is exactly the kind of mixed, multi-account year that manual consolidation turns into a lost afternoon.

You drag the whole folder onto the batch page. Extraction runs as the files upload, and within a couple of minutes every statement is read — the scanned ones routed through OCR, the digital ones structured directly. You click merge, and instead of a download you get Merge Review: a single grid stacking all thirty statements, a quality score of, say, 96, and a short issues panel. Two flags come from a scanned March statement where the OCR wasn't sure of an amount; one is an ambiguous date on an older savings statement; one is a possible duplicate where the card statement's billing period overlapped a month boundary.

You work the panel top to bottom. Click the first flag, the grid jumps to the scanned amount, you glance at the source figure and confirm it — score ticks up. Same for the second. The ambiguous date you type in directly. The duplicate you check against the source-file column, see it's a genuine overlap, and accept the flag so it's counted once. Four corrections, perhaps ninety seconds of work. The balance check on every statement now passes, which tells you no transaction was dropped anywhere in the year across any account.

Now you export. Because the operating account and card both feed your accounting software, you take a .QBO bank feed for those and import them duplicate-safe; the savings account you keep as Excel for the file. What would have been an afternoon of copy-paste — with no guarantee it was complete — became a few minutes of upload, four confirmations, and a verified year you can hand to the accountant. That is the whole point of merging and reviewing rather than just combining.

Notice what you didn't do in that example: you never opened a single PDF to retype a figure, never aligned a column by hand, never wondered whether a statement was complete, and never hunted for a duplicate across thirty files. The system did the reading and the alignment, the balance checks did the verifying, and your attention went only to the four things that genuinely needed a human. That redistribution of effort — machine on the volume, human on the exceptions — is what makes a year-end close feel like a review rather than a re-entry, and it scales the same way whether the folder holds thirty statements or three hundred.

Doing this at scale

The same workflow holds whether you merge ten statements or run it across a book of clients. A practice can consolidate each client's year in turn, letting the balance checks and quality score carry the load and only reviewing flagged exceptions. A finance team can do the same across subsidiaries. The per-document checks are what keep accuracy dependable as volume rises — each statement carries its own completeness proof, so nothing hides in an aggregate.

For fully automated pipelines, the same extraction and validation are available over the document extraction API and the bank statement API, returning structured data with the validation built in. Whether a human drives Merge Review or a system consumes the API, the completeness guarantees are identical. The at-scale playbook covers the operational side.

Frequently asked questions

Related reading