Statement AI is live  ·  Reconciling Canadian PDFs to the cent

Guide

The three-sheet workbook every accountant secretly wants

Summary, transactions, vendors, categories. What goes where, why the shape never changes, and how to plug Balancely output into QuickBooks or Excel pivot tables.

The Balancely Team5 min read

We export every job as the same three-sheet workbook. Summary, All Transactions, Vendor Summary. In that order, with that capitalization, with the same column headers every time.

The shape doesn't change because the people who consume the workbook don't want it to. Bookkeepers build pivot tables on top of column B. Accountants point Power Query at column C. VLOOKUPs reference column D. The minute we shuffle a column, someone's monthly process breaks.

Here's what each sheet does, why we put it there, and what we deliberately left out.

Sheet 1: Summary

One row per statement. Twelve statements in your upload, twelve rows here. The columns are:

ColumnExample
Statement Date2026-03-15
IssuerRBC
Card (last 4)4511
Opening Balance$1,204.55
Total Purchases$2,810.32
Total Payments($1,204.55)
Closing Balance$2,810.32
ReconciledYes (0.00)

The Reconciled column is the one your accountant looks at first. It says “Yes” with a delta in cents if the statement balanced within tolerance, or “Check” with a signed delta if not. Conditional formatting paints the row amber when it's “Check.”

A grand-total row at the bottom sums the period: total purchases, total payments, net change. We don't hide it under a collapsed group, and we don't put it on a separate sheet. If you uploaded a year of statements, you want the year-total visible on the same screen.

Sheet 2: All Transactions

One row per transaction across every statement in the upload. This is the sheet most people actually live in. The columns:

  • Date — ISO format yyyy-mm-dd, always.
  • Statement — the source statement's date, so you can filter to one cycle.
  • Description — the line as it appeared on the statement, cleaned up (extra whitespace collapsed, location codes preserved).
  • Vendor — our normalized vendor name. “TIM HORTONS #4521 OTTAWA ON” becomes Tim Hortons.
  • Category — payment, purchase, fee, interest, return.
  • Amount — signed. Purchases positive, payments negative.

Auto-filter is enabled on the header row. Conditional formatting paints negative amounts green (those are payments — money flowing in your favor) and red rows for any transaction whose source statement failed reconciliation. Frozen first row, frozen first column.

Auto-filter on, freeze panes on, ISO dates. Three things we get asked to add to every accounting tool that ships without them.

Sheet 3: Vendor Summary

One row per normalized vendor, aggregated across every transaction in the upload. The columns are Vendor, Transaction Count, Total Spend, and % of Total.

The normalization is where most of the work happens. A real-world Tim Hortons charge looks like one of:

TIM HORTONS #4521 OTTAWA ON
Tim Hortons 4521 OTTAWA
TIMHORTONS-4521 ON CA
TIM HORTON 4521

Our regex pipeline normalizes all four to Tim Hortons. Same for Uber, Petro-Canada, Loblaws, every Amazon variant, and the long tail of franchise-numbered retailers. The rules are conservative: when in doubt, keep them separate. Pro and Business tiers can override the rules with custom mappings if their tax categorization needs differ.

The % of Total column is a formula, not a value. We write =C2/SUM(C:C) rather than computing the percent and writing it out, because the moment you delete a row to filter out a vendor, you want the percentages to update. Hard-coded percentages are a footgun.

What we deliberately left out

  • Charts. Pretty, useless. People who want a chart will pivot the data themselves; people who don't want one don't want one cluttering the workbook.
  • Inferred categories. “This Tim Hortons charge is meals & entertainment” is a tax decision your accountant makes, not a label we should pre-fill.
  • A fourth sheet. We've had requests for a budgets sheet, a forecast sheet, a flagged-transactions sheet. Each of them belongs in a different tool. The workbook is for the ground truth; the analysis happens elsewhere.

The point isn't that three sheets are enough for everyone. It's that the same three sheets, in the same shape, every time, are enough for almost everyone — and the consistency is more valuable than any single feature we could add.

Try Balancely on your next statement.

Two free conversions to start — no credit card required.