The setup

Tom runs the books for eight small businesses. A pub, two consultancies, a small construction firm, a private GP practice, two creative agencies, and a farm shop. Different banks across the board — Barclays, HSBC, Starling, Monzo Business, a credit union, and a US-based dollar account for one client who invoices internationally. All eight use QuickBooks Online.

Every month, Tom downloads each client's bank statement as a CSV and imports it into QBO. The job description is "import the statement." The actual job is "reshape eight different CSVs into the format QBO is willing to accept."

The 3-column rule

QBO's Web Connect import requires a CSV in this shape: 

Date, Description, Amount
2026-05-02, Direct Debit BT Group, -54.99
2026-05-03, Customer payment 10421, 1640.00
2026-05-03, Card payment Pret, -8.40

Or four columns — Date, Description, Credit, Debit. That's the entire allowed format. Anything else and the import fails silently, partially, or with one of QBO's famously unhelpful error messages.

Here's what a Starling Business CSV actually looks like: 

Date,Counter Party,Reference,Type,Amount (GBP),Balance (GBP),Spending Category,Notes
02/05/2026,BT Group,DDR-BT-22418,Direct Debit,-54.99,3214.18,Bills,
03/05/2026,Acme Ltd,FPI-10421,Inbound,1640.00,4854.18,Income,Customer payment

Eight columns. Date in DD/MM/YYYY (QBO might or might not accept it depending on locale). The "Counter Party" column is closer to QBO's "Description" than the "Reference" column. The "Spending Category" wants to be in the description too. The other four columns are noise QBO will choke on if you leave them in.

Every bank ships a slightly different shape. 3,000 to 7,000 people a month search "import bank statement csv quickbooks." The QB Community is full of variations on the same question — a typical one is captured well in Conectier's guide, which has to explain the 3-vs-4-column rule before anything else.

What Tom used to do

For each of his eight clients, every month:

  • Open the bank CSV in Excel.
  • Delete the noise columns. Remember which column to keep for each bank.
  • Combine the right columns into "Description" (different combinations per bank).
  • Reformat dates because half his banks export DD/MM and his US dollar account exports MM/DD.
  • Save as CSV. Lose all formatting Excel quietly applied.
  • Import to QBO. If the date column is wrong, the entire batch rejects. Find the bad row. Fix. Re-import.

Best case: 20 minutes per client. Worst case (when Excel silently rewrites a leading-zero account reference, or strips a thousands separator): an hour. Average across eight clients: about four hours a month on column gymnastics.

What the market sells

There's a whole category of tools that exist because of this one problem. DocuClipper starts at $14/mo and handles the conversion in their cloud. MoneyThumb sells CSV2QBO as a one-time license. Bankstatementconverter.com is browser-based and free for occasional use. The native QBO importer is famously fiddly.

Each of these handles one part of the problem. None of them solves the "every bank exports a different shape, and they change quarterly" problem in a way that's free, fast, and lives on your laptop.

The Mappd recipe

Tom configured one Mappd recipe per bank. Eight banks, eight recipes — about three minutes each the first time. Each recipe encodes: which source column is the date (and what format it's in), which combination of columns becomes the Description, which column is the Amount, and what to do with the rest (drop them).

Now his monthly bank-statement routine looks like this:

  • Download each client's CSV.
  • Drag onto the matching recipe in Mappd.
  • Eyeball the diff (Mappd shows row counts in/out, date-format normalisation, dropped columns).
  • Export. Import to QBO.

Per client: about 3 seconds of Mappd-time, plus 1–2 minutes for the QBO import itself. Across eight clients: 24 seconds of Mappd, ~12 minutes of importing.

Before
~4 hrs / month
8 clients × ~30 min of Excel gymnastics. One batch-reject per quarter, on average.
After
~12 min / month
8 recipes, configured once. Diff catches batch-rejecting rows before export.

The math, over a year

Time back / year
~46 hrs
Roughly one extra client of capacity
Tool cost avoided
$168/yr
DocuClipper alternative subscription
Mappd cost
£144/yr
Pro plan (covers 8+ recipes)

The honest bit

What Mappd doesn't do: connect to your bank. You still have to download the CSV yourself — Mappd doesn't replace Plaid or your bank's API. The "always-broken" part of this workflow isn't the connection; it's the shape. Banks change column orders, add new fields, rename "Reference" to "Memo" once a year. Mappd's recipe is one place to absorb those changes — update the recipe once, every future month works again.

If your monthly close includes more than a couple of bank CSVs

The Bank CSV → QuickBooks preset is one of Mappd's six launch recipes. The first run per bank takes about three minutes; every subsequent run takes three seconds. Files under 5 MB are free — which is most bank statements unless you're processing six months at a time. Pro (£12/mo for one bookkeeper, ten recipes) covers a typical small practice.

Names changed for anonymity. Quotes, search-volume figures, and workflow described are real and sourced from forum evidence.