How to Digitise Client Invoices and Receipts for MTD: A Step-by-Step Guide

To digitise invoices and receipts for Making Tax Digital (MTD), you need to extract specific transaction data from each document and store it in a digital format that meets HMRC's record-keeping requirements. Scanning to JPEG is not enough. MTD requires structured, searchable data — not image files.

This guide walks through the exact workflow for receipts, invoices, and bank statements. It covers what data to extract, how to handle each document type, and how to run this process across a client portfolio without building a data entry team.

What 'Digitising' Actually Means for MTD (Not Just Scanning)

Digitising for MTD means capturing structured data from a document — not creating a digital image of it.

HMRC's Making Tax Digital for Income Tax (MTD for IT) mandates that businesses maintain digital accounting records containing the date, amount, and category of every business transaction. A scanned image stored in a folder does not meet this requirement. The data inside the document must be accessible, structured, and held within software that connects to HMRC's systems.

Under MTD for IT — which became mandatory in April 2026 for sole traders and landlords with gross income over £50,000 — maintaining paper records no longer satisfies the legislative requirement. Firms that sign up clients must ensure those clients are feeding structured digital data into MTD-compatible software such as Xero, QuickBooks, or Sage, not simply emailing photos of receipts.

The distinction matters practically. A JPEG of an invoice sitting in a Google Drive folder cannot be queried, categorised, or submitted. A structured record with supplier name, invoice number, date, net amount, VAT, and expense category can be pushed directly into a bookkeeping system.

The Problem With Basic Scanning — Why a JPEG Isn't Enough

Scanning a document gives you an image. MTD requires data.

Most small business clients already scan or photograph documents — but the output is typically a flat image file (JPEG, PNG) or an unsearchable PDF. These files:

• Cannot be parsed by accounting software
• Cannot be linked to digital records as required by HMRC's digital link rules
• Cannot be queried for VAT, date ranges, or supplier names
• Do not constitute "digital records" under MTD legislation

A second problem is accuracy. Manual data entry error rates sit between 1% and 4% across experienced staff under normal conditions — higher under time pressure. For an accounting firm processing hundreds of documents per client per year, that error rate compounds quickly across a portfolio of 50 or 100 clients.

The requirement is not to store a copy of the document. It is to extract the underlying transactional data and hold it digitally in a compliant format. Receipt digitisation for MTD accounting means getting from a physical or photographed document to a structured data record — automatically, at volume, and with a verifiable audit trail.

What Data Needs to Be Extracted From Each Document Type

Before building any workflow, you need to know exactly which fields MTD requires — and which additional fields are commercially useful.

Receipts

Invoices (Received — Accounts Payable)

Bank Statements

These fields form the foundation of every MTD-compliant digital record. Any document digitisation workflow should capture all MTD-required fields at minimum. The best-practice fields are what make those records useful inside a bookkeeping system.

Step-by-Step: Receipt Digitisation Workflow

Here is the complete process for digitising a client's receipts for MTD compliance.

Step 1 — Collect the source documents. Ask clients to photograph receipts immediately after purchase using a mobile app, or submit physical receipts in a weekly batch. Establish a consistent submission channel: email, shared folder, or a dedicated document portal.

Step 2 — Convert images to a processable format. If documents arrive as photos (JPEG/HEIC), convert to PDF or pass them directly to an extraction tool. Ensure images are clear, well-lit, and include the full document. Poor image quality is the single biggest cause of extraction errors.

Step 3 — Run automated data extraction. Pass each document through a document intelligence tool that identifies and extracts structured fields: date, supplier, total, VAT, category. Tools like Taxiom extract these fields automatically from receipts in any format — printed, handwritten, or photographed.

Step 4 — Review extracted data. Human review remains important. Check for missing fields, unrecognised suppliers, and ambiguous category assignments. Flag documents that need client clarification.

Step 5 — Categorise each transaction. Map the expense to the correct MTD-compliant category. Use a standardised category list that aligns with your bookkeeping software's chart of accounts.

Step 6 — Push data to your bookkeeping system. Export structured data into Xero, QuickBooks, Sage, or your MTD-compatible platform of choice. Most tools support CSV export or direct API integration.

Step 7 — Archive the source document. Retain the original image or PDF against the digital record. HMRC recommends retaining records for at least five years after the MTD submission deadline for each tax year.

Step-by-Step: Invoice Digitisation Workflow

Invoices contain more structured data than receipts and are typically more consistent in format — which makes them more amenable to high-accuracy automated extraction.

Step 1 — Establish an intake channel. Ask clients to forward supplier invoices to a designated email address or upload to a shared folder. Set a weekly submission cadence. Inconsistent intake is where most firms lose time.

Step 2 — Classify the document type. Determine whether the invoice is a purchase invoice (accounts payable), a sales invoice (accounts receivable), or a credit note. Each has a different extraction template and bookkeeping treatment.

Step 3 — Extract structured fields. Run invoice scanning through a document intelligence platform. For invoice scanning and MTD compliance, the key fields are: invoice number, supplier name, invoice date, net amount, VAT rate, VAT amount, gross total, and payment due date. Taxiom extracts all of these fields in a single pass, regardless of invoice layout or format.

Step 4 — Match to supplier records. Cross-reference the extracted supplier name against your client's supplier master list. Flag new suppliers for onboarding. This step prevents duplicate supplier records in the bookkeeping system.

Step 5 — Validate totals. Confirm that net + VAT = gross. This is a simple arithmetic check but catches a significant proportion of extraction errors, particularly on handwritten or poorly formatted invoices.

Step 6 — Apply expense category. Assign the correct nominal code or expense category based on the supplier and invoice type. For recurring suppliers, this can be automated using a rules engine.

Step 7 — Post to accounting software. Export to your bookkeeping platform. Where direct integration exists, this step is instantaneous. Where it doesn't, a structured CSV handles the transfer cleanly.

Step 8 — File and link the original. Attach the source PDF to the transaction record inside your accounting software. This completes the digital link between the physical document and the MTD submission.

Step-by-Step: Bank Statement Digitisation Workflow

Bank statements are often the most time-consuming documents to process manually. They contain high transaction volumes, inconsistent descriptions, and variable formats across banks. A structured workflow cuts processing time substantially.

Step 1 — Obtain the statement. Clients should provide bank statements as PDF downloads from online banking, not photographs. PDF statements have consistent structure and extract far more accurately than photographed pages.

Step 2 — Extract all transactions. Run the PDF through a document intelligence tool. For bank statements, the required output is a transaction-level dataset: date, description, debit, credit, balance. Taxiom processes multi-page bank statements and returns a clean, structured dataset ready for import.

Step 3 — Identify business vs. personal transactions. For sole traders with mixed-use accounts, flag transactions that require client clarification before posting. Build a reference list of known personal transactions to automate this over time.

Step 4 — Auto-categorise where possible. Use a rules engine or bookkeeping software's bank feed matching to assign categories. Transactions from known suppliers can be categorised automatically. Novel transactions require manual review.

Step 5 — Reconcile against invoices and receipts. Match bank transactions to processed invoices and receipts. Unmatched transactions need supporting documents before they can be posted as compliant MTD records.

Step 6 — Import to bookkeeping software. Import the structured transaction file into Xero, QuickBooks, or your MTD platform. Most platforms accept CSV uploads with date, description, debit/credit columns.

Step 7 — Flag and resolve exceptions. Unmatched, uncategorised, or flagged transactions should be resolved within the same processing cycle — not left to accumulate before the quarterly submission deadline.

How to Do This for 50 Clients Without a Data Entry Team

Scaling receipt digitisation for accounting firms across a large client base requires a different approach to per-client manual processing.

The economics do not work at volume. A firm processing documents for 50 clients manually — at even 30 minutes per client per week — is spending 1,300 hours per year on document intake alone. That is the equivalent of more than half a full-time employee, doing work that generates no billable output.

The answer is a standardised, automated intake-to-export pipeline:

• One intake channel per client. Each client submits documents to a dedicated folder or forwarding address. This eliminates collection chasing and document mixing.
• Automated extraction. Every document entering the pipeline runs through extraction automatically. No manual triggering. No queue management.
• Exception-only human review. Staff review flagged documents, not every document. The workflow surfaces problems — it doesn't require a person to verify clean data.
• Batch export. Structured data exports to each client's bookkeeping system on a fixed schedule — daily, weekly, or on demand.

Taxiom is built specifically for this model. It processes documents at any volume, extracts structured data across receipt, invoice, and bank statement types, and integrates with the bookkeeping platforms UK accounting firms already use. The accounting-specific workflow handles the full pipeline from document intake to structured export — without requiring a data entry team.

Try Taxiom free — automate steps 4–7 completely. Start at taxiom.co

Accuracy and Error-Checking Best Practices

Even automated extraction requires quality controls. Here is what a robust checking layer looks like.

Set field-level confidence thresholds. Most extraction tools return a confidence score per field. Flag any field below your threshold (typically 90–95%) for human review. Do not post low-confidence fields without verification.

Run arithmetic validation rules. For invoices: net + VAT must equal gross. For bank statements: running balance must reconcile with the prior period closing balance. These checks catch extraction errors before they reach your bookkeeping system.

Build a supplier normalisation list. Supplier names on receipts and invoices are often inconsistent: "AMZN MKTP UK", "Amazon.co.uk", and "Amazon UK Ltd" are the same supplier. A normalisation table maps variants to a canonical name, which keeps your client's bookkeeping clean.

Use a pre-posting review queue. Before any extracted data enters the bookkeeping system, route it through a short review step. A trained bookkeeper can review 50 extracted records in under ten minutes. This is the quality gate between extraction and posting.

Retain the source document against every record. The source image or PDF should be linked to the digital transaction record. This provides the audit trail HMRC may request in a compliance check, and gives clients a single point of reference if a transaction is ever queried.

Conduct monthly exception analysis. Track which document types, clients, or suppliers consistently produce extraction errors or flagged fields. Use this data to improve intake quality — better image guidance for clients, better templates for recurring formats, better rules for categorisation.

Conclusion

Digitising invoices, receipts, and bank statements for MTD is not a scanning exercise — it is a structured data workflow. The requirement is extracted, categorised, and digitally linked transaction data, delivered into MTD-compatible software on time, every quarter.

For accounting firms, the challenge is doing this at scale without building a dedicated data entry function. The answer is an automated intake-to-export pipeline with exception-based human review.

The steps above give you the framework. The practical question is which tool handles the extraction layer. Taxiom was built to answer that question for accounting firms: any document, any format, structured output, direct integration.

FAQ

What does HMRC actually require when it says 'digital records' for MTD?

HMRC requires that the date, amount, and category of each business transaction are recorded and held digitally in software compatible with MTD. A scanned image is not sufficient on its own. The transaction data must be in a structured, queryable digital format that can be submitted — or used to generate a submission — through HMRC's systems. ICAEW's MTD guidance confirms that maintaining paper records no longer satisfies the legislative requirement for mandated taxpayers.

Is a photographed receipt compliant with MTD?

Not by itself. A photograph of a receipt is not a digital record under MTD rules. The data within the receipt — supplier name, date, amount, VAT — must be extracted and held in your bookkeeping software. You may retain the image as supporting documentation, but the image alone does not constitute a compliant digital record.

How long does it take to digitise invoices and receipts per client per month?

This depends heavily on transaction volume and the tools used. Manual processing of 100 documents typically takes two to four hours per client. With an automated extraction and review pipeline, the same volume can be processed in under 30 minutes — mostly comprising exception review. The time saving compounds significantly across a multi-client portfolio.

Which accounting software platforms accept structured data from document digitisation tools?

Xero, QuickBooks Online, Sage Accounting, and FreeAgent are the most commonly used MTD-compatible platforms in the UK. Most document intelligence tools, including Taxiom, support CSV export compatible with all four. Some platforms also offer direct API integration for push-based data transfer.

Do I need to digitise historical documents before MTD compliance begins?

No. MTD requires digital record-keeping from the point of mandation onwards. You are not required to retrospectively digitise historical records. However, many firms use the MTD transition as an opportunity to digitise prior-period records for operational efficiency — particularly for clients who are likely to face HMRC compliance checks.

What happens if a client submits a poor-quality photo of a receipt?

Poor image quality is the leading cause of extraction errors in any document digitisation workflow. Blurred images, low light, partially cropped documents, and folded receipts all reduce extraction accuracy. The best mitigation is client guidance: provide a one-page briefing on how to photograph receipts correctly, and build a rejection workflow for images that fall below your quality threshold.