Importing Q1 Expenses: CSV Format and AI Categorisation

With 58 days until the Q1 deadline on 7 August, most sole traders and landlords face the same problem: months of expense receipts stuffed in shoeboxes, scattered across spreadsheets, or buried in bank statements. The good news? Modern MTD software can import your CSV files and use AI to categorise expenses automatically. The bad news? Most people get the formatting wrong and create more work for themselves.

This guide walks you through importing expenses via CSV, how AI categorisation maps your transactions to HMRC fields, and the formatting mistakes that turn a 10-minute job into a 2-hour headache.

Why CSV Import Matters for Q1

If you've been tracking expenses in Excel, Google Sheets, or downloading bank statements, you're probably looking at hundreds of transactions that need categorising. Manual data entry would take days. CSV import takes minutes - if you get it right.

Most MTD software accepts CSV files because it's the standard format for financial data. Your bank exports CSV files. Excel saves as CSV. Even ancient accounting software exports CSV. It's the common language between different systems.

But here's what most people don't realise: the AI categorisation only works well if your CSV is properly formatted. Get the column headers wrong or mix up date formats, and you'll spend hours fixing mistakes that could have been avoided.

Essential CSV Format Requirements

Before you export anything, your CSV needs these five columns as a minimum:

• Date: When the expense occurred
• Description: What you bought or paid for
• Amount: How much you spent (numbers only)
• Category: Optional but helpful (office supplies, travel, etc.)
• Reference: Invoice number, receipt reference, or bank transaction ID

Your column headers must be clear and consistent. Use "Date" not "Transaction Date" or "Expense Date". Use "Amount" not "Cost" or "Value". The simpler the headers, the better the AI categorisation works.

Date Format Rules

This trips up most people. UK MTD software expects dates in DD/MM/YYYY format, but many systems export in MM/DD/YYYY or YYYY-MM-DD. Check your dates before importing.

Safe date formats for UK MTD software:

• 01/06/2026 (1 June 2026)
• 1/6/2026 (also 1 June 2026)
• 2026-06-01 (ISO format - usually fine)

Dangerous formats that cause problems:

• 06/01/2026 (could be 6 January or 1 June)
• Jan 1, 2026 (text dates often fail)
• 1st June 2026 (ordinal numbers confuse systems)

Amount Formatting

Keep amounts simple. Numbers only, with a decimal point for pence. No pound signs, no commas, no spaces.

Good: 125.50

Bad: £125.50 or 1,250.50 or £1,250.50

If your spreadsheet has formatted currency, copy the amounts to a new column and use "Paste Special > Values" to strip the formatting.

How AI Categorisation Works

When you import a CSV file, the AI categorisation engine analyses three things:

1. Description text: Keywords like "fuel", "office", "insurance"
2. Amount patterns: £50 monthly might be a subscription
3. Payee information: Company names it recognises

The AI then maps these to HMRC expense categories automatically. For sole traders, this includes office costs, travel expenses, marketing, professional fees, and equipment costs. For landlords, it covers repairs, insurance, letting agent fees, and property management costs.

Common AI Categorisation Results

Here's what the AI typically gets right first time:

• Fuel expenses: Shell, BP, Tesco petrol → Travel costs
• Office supplies: Staples, WHSmith, Amazon office items → Office costs
• Professional fees: Accountant payments, legal fees → Professional services
• Insurance: Direct Line, Aviva business policies → Insurance
• Phone bills: BT, EE, Vodafone → Telephone and internet

What it struggles with:

• Generic descriptions like "Payment" or "Direct Debit"
• Cash payments with no payee information
• Mixed purchases (office supplies and personal items together)
• Unusual or specialist suppliers

Step-by-Step CSV Import Process

Most MTD software follows a similar import process. Here's the typical workflow:

Step 1: Prepare Your CSV File

Clean your data first. Remove any personal expenses, fix date formats, and ensure amount columns contain numbers only. Save the file with a clear name like "Q1-expenses-2026.csv".

Step 2: Start the Import

Look for "Import" or "Upload" options in your MTD software. This is usually in the expenses section or main menu. Select your CSV file and upload it.

Step 3: Map Your Columns

The software will show your CSV columns and ask you to map them to its fields. Match "Date" to "Expense Date", "Amount" to "Cost", etc. This step is crucial - get it wrong and your import will fail.

Step 4: Preview and Check

Most software shows a preview of how your data will import. Check the first few rows carefully. Are dates correct? Do amounts look right? If something's wrong, cancel and fix your CSV file.

Step 5: Run AI Categorisation

After import, the AI categorisation runs automatically. This can take a few minutes for large files. You'll see expenses grouped by category with confidence scores.

Step 6: Review and Adjust

Go through each category and check the AI's work. Pay special attention to:

• High-value items that might need splitting
• Items marked "Uncertain" or with low confidence scores
• Personal expenses that shouldn't be claimed
• Categories that seem wrong for your business

Common Formatting Mistakes That Break Imports

After helping thousands of users import expenses, these are the mistakes we see repeatedly:

Mixed Date Formats in One File

Your CSV has 01/06/2026 in row 10 and 2026-06-15 in row 20. The import fails because the software can't work out which format you're using. Stick to one format throughout.

Merged Cells or Complex Headers

Excel users often export sheets with merged cells, multiple header rows, or formatted titles. CSV import needs simple, flat data. One header row, one value per cell.

Currency Symbols and Commas

£1,250.50 looks nice in Excel but breaks most import systems. Strip all formatting and use plain numbers: 1250.50.

Empty Rows and Columns

Blank rows between data or empty columns at the end cause import errors. Clean these up before saving your CSV.

Special Characters in Descriptions

Pound signs, quotes, and commas in description fields can break CSV parsing. If a description contains commas, the whole field needs quote marks: "Office supplies, paper, pens".

Maximising AI Categorisation Accuracy

To get the best results from AI categorisation, help the system by improving your expense descriptions:

Be Specific in Descriptions

Instead of: "Amazon purchase"

Write: "Amazon office supplies - printer paper and pens"

Instead of: "Fuel"

Write: "Shell petrol station - client visit fuel"

Include Business Purpose

The AI learns from context. "Hotel London client meeting" is better than just "Hotel London". It helps categorisation and provides better records for HMRC.

Use Consistent Naming

If you buy from the same suppliers regularly, use consistent names. "BT Business" not "BT" sometimes and "British Telecom" other times.

Add Category Hints

If your CSV has a category column, use it. Even rough categories like "Travel", "Office", "Marketing" help the AI make better decisions.

What Happens After Import

Once your expenses are imported and categorised, they become part of your MTD records. The software maps each category to the correct HMRC field on your quarterly update form.

For sole traders, categories map to boxes like:

• Office costs → Administrative expenses
• Travel → Car and travel expenses
• Marketing → Advertising and promotion
• Equipment → Capital allowances or repairs

For landlords, the mapping covers:

• Repairs → Property repairs and maintenance
• Insurance → Property insurance
• Letting fees → Letting agent fees
• Mortgages → Mortgage interest (if applicable)

The beauty of good MTD software is that you don't need to know these mappings. You just need to categorise expenses correctly and the software handles the HMRC form completion.

Troubleshooting Import Problems

When CSV imports fail, it's usually one of these issues:

File Too Large

Most MTD software has file size limits. If your CSV is over 10MB, split it into smaller files or remove unnecessary columns.

Encoding Problems

If your CSV contains special characters or was created on a Mac, save it as "CSV (UTF-8)" in Excel or Google Sheets.

Date Recognition Errors

The software can't parse your dates. Check the format and ensure consistency throughout the file.

Duplicate Detection

Some software flags potential duplicates during import. This is usually helpful, but check carefully - similar amounts on the same day might be legitimate separate expenses.

Planning for Q2 and Beyond

Once you've mastered CSV import for Q1, you can use the same process for future quarters. Many users set up monthly CSV exports from their bank or accounting system, making quarterly updates much easier.

Consider setting up regular data exports so you're not scrambling before each deadline. As we covered in our Q2 planning guide, getting ahead of the admin makes everything smoother.

CSV import with AI categorisation transforms the tedious job of expense entry into a quick, mostly automated process. Get the formatting right, help the AI with good descriptions, and review the results carefully. Your Q1 filing becomes much more manageable, and you'll have a system that works for every quarter ahead.