Monthly Spending Trends

Get one month wrong and the trend lies in a specific way: a missing debit makes that month look cheaper, which reads as a category falling when nothing fell. The groundwork isn't the chart. It's proving each month reconciles before any go into the same table.

Why one month tells you almost nothing

A single statement is a snapshot of thirty-odd days, too short to separate signal from noise. Was March's 2,100 marketing spend high or low? On its own you can't say. Put it next to January's 900 and February's 1,300 and it's obviously climbing, fast.

Trend work asks a different question. Not "how much did I spend?" but "is this heading somewhere I should worry about?" Stacking months gives three reads:

• Direction. Rising, falling, or flat over the run? A subscriptions line that goes 180, 210, 260, 310 has a clear slope, though no single month looks alarming.
• Seasonality. Some costs repeat on a calendar, not a budget: stock builds before a busy quarter, energy climbs in winter, the accountancy fee lands once a year. Knowing the shape stops you panicking at a spike you should expect.
• Step changes. A figure that sat flat for months, then jumped and stayed there — usually a new contract, a price rise, or a subscription that slipped through. The easiest win, because it has a date attached.

These live in the gaps between months, not in any single one.

Every month has to reconcile before it joins the table

Here's the bit people skip, and it poisons the whole comparison. Before any month goes into the table, prove it's complete. Banks print a running balance after every transaction: start at the opening balance, apply each credit and debit in order, and you should land exactly on the closing balance. Land there and the month reconciles. Come up short, say by 480, and a transaction is missing or misread, so that month is now lighter than it really was.

In a single month that error is bad enough. In a trend it's worse, because it creates a fake movement. Say April lost a 480 supplier debit in extraction: April now reads cheaper, the chart shows costs "falling", and you relax about a category that never moved. The error didn't just dent one figure; it bent the line.

Extraction slips in predictable spots: a row lost in the seam between pages, a 1,290.00 debit read as 1,920.00 when OCR transposes digits, a payment in the wrong column. Across twelve months those slips compound.

Export Bank Statement runs this check before you ever see a spreadsheet. It walks the running balance from opening to closing and confirms each statement reconciles — or flags the month that doesn't, rather than handing you an export that's quietly short a payment. Most converters extract and leave the checking to you. For trend work that gap is fatal: you're no longer comparing months, you're comparing one good month against one that lost a transaction and calling the difference a trend.

Build the trend table: a column per month

Once each month reconciles, the analysis is mechanical. The shape you want is a pivot: categories down the side, months across the top, one number per cell. You can build it two ways.

By hand, drop every categorised transaction into a pivot table — Category as rows, Month as columns, sum of amounts as the value — and chart a row in two clicks. Fine for a handful of months; tedious past a year.

With the built-in analyser, which after conversion groups each month's transactions into categories, surfaces recurring payments and merchants, and totals each group — no pivot to build by hand. What it can't read is your intent: a payment tagged "supplier" that's really a one-off equipment buy will distort the trend until you re-tag it. On a trend that matters more, because one mislabelled big payment can invent a spike.

A few rules keep a trend table honest:

• Compare like with like. A 28-day February against a 31-day March looks lighter for nothing more than the calendar, so remember the shorter month before you read meaning into the dip.
• Watch for timing, not spending. Rent paid on the 31st of one month and the 1st of the next puts two payments in one month and none in the other — a calendar artefact, not a trend.
• Keep categories stable. If "software" means different things in different months, the row is noise. Decide categories once and apply them everywhere.

How to track monthly spending trends, step by step

1. Gather consecutive months. Download the PDFs from online banking; six to twelve months gives trends and seasonality room to show. A scan or phone photo works too, read by OCR.
2. Convert each month to Excel or CSV, so each comes out as clean, columned data instead of text you'd re-key.
3. Let each month's reconciliation clear. Any flagged month gets fixed first — a trend with one short month is worse than no trend.
4. Stack them into a pivot. Categories as rows, months as columns. Use the analyser's grouping or build it by hand, then re-tag miscategorised lines so each category means the same thing every month.
5. Read across the rows for direction, seasonal shape and step changes. Chart the two or three that move most — those are worth acting on.

One verified month usually takes well under a minute, which is the real reason trend work is worth doing at all. When each month is that quick, you check monthly and catch a rising line in month two — a quick cancellation, not an expensive story told at year-end. The expense analysis guide covers categorising a single month well — the building block every trend sits on.

A worked example

A small agency, eight months of statements. Stacking them, two rows stood out. Software climbed steadily — 210, 240, 280, 330 — a slow creep no single month would have flagged, driven by three trial subscriptions that had quietly rolled into paid. The fix was four cancellations.

The second row was the trap. Marketing appeared to halve in month five, which looked like a win until the reconciliation flag on that month explained it: a 1,900 ad invoice had dropped in the seam between pages during extraction. The "fall" was a missing transaction. Once fixed, marketing was flat, and the real story stayed with the software creep. Read the broken table and you'd have claimed a saving that never happened while the subscriptions kept climbing. The business cash flow analysis guide sets trends like these against the income coming in, and the recurring transactions guide goes deeper on subscription creep.