200 Japan Qualified Invoices, One
Spreadsheet: Batch Without T-Number Typing

New to the Qualified Invoice System? Start with our step-by-step guide to extracting Japan's qualified invoice data into Excel, covering the 6 mandatory fields, registration number format, dual-rate tax structure, and the October 2026 transition deadline.

Why Batch Processing Under the Invoice System Isn't About Speed — It's About Consumption Tax Filing Compliance

Under the post-2023 consumption tax filing rules, businesses have two calculation methods for their consumption tax return (消費税申告): accumulation calculation (積上げ計算) and deduction calculation (割戻し計算). The deduction method is simpler — you take total tax-included purchase amounts by rate and calculate tax from the aggregate. The accumulation method is more precise and often more favorable for businesses with many small-value transactions, because per-invoice rounding consistently favors the taxpayer.

Here's the critical constraint most finance teams discover only when they prepare their first post-2023 tax return: if you choose accumulation calculation for your sales tax, you must use it for your purchase tax as well. You cannot mix methods — accumulation for sales and deduction for purchases is not permitted under the NTA's Qualified Invoice System rules.

What this means in practice: for every qualified invoice you receive from a supplier, you need the exact consumption tax amount as printed on that document — not calculated from the total, but extracted as-is from the invoice itself. Multiply this by 200 invoices a month. The batch processing problem is not "I have 200 invoices to type." It's "I need 200 individual consumption tax amounts, per rate, before my tax return software can even begin calculating what I owe."

The math makes this concrete. Accumulation calculation works by taking the sum of all consumption tax amounts listed on your qualified invoices and multiplying by 78/100. If you use the deduction method, you take total tax-included purchases per rate and multiply by 7.8/110 (standard) or 6.24/108 (reduced). Both paths end at the same tax return — but the accumulation path requires you to first extract the per-invoice tax data from 200 documents before you can sum it. That's the batch extraction gap.

100–300 Invoices per Month Breaks Single-Document Workflows

A Japanese mid-size company with 50–100 active supplier relationships typically receives between 100 and 300 invoices per month. At the low end — 100 invoices, each requiring six fields under the Qualified Invoice System — you are extracting 600 data points. At 3 minutes per invoice for manual entry (locating the registration number on each supplier's unique layout, splitting line items by tax rate, typing consumption tax amounts), that's 5 hours of data entry. At 300 invoices: 15 hours — nearly two full working days per month spent on nothing but invoice data entry.

The format diversity makes it worse. A single month's supplier invoices can include:

• Clean PDFs from registered corporate suppliers with clearly labeled registration numbers in standard formats
• Scanned paper invoices from regional wholesalers using pre-printed forms with the T-number squeezed into a corner
• Mobile-phone photos of handwritten delivery slips from small sole proprietors that technically qualify as simplified qualified invoices (適格簡易請求書)
• Foreign-currency invoices from overseas suppliers who registered for QII status — where consumption tax must still be shown in Japanese yen

A template-based extraction approach — training a tool to recognize fields on Supplier A's invoice, then Supplier B's, then Supplier C's — collapses under this diversity. By the time you've configured the template for one supplier's December format, they've changed their invoice layout in January. The tool that works for one invoice becomes the bottleneck when there are 199 more behind it, each formatted differently.

Batch processing — uploading multiple files simultaneously and receiving a single merged spreadsheet where each row corresponds to one document — changes the workflow from "one invoice, one extraction session, one result" to "one upload, one column definition, 200 results in one file." The column names you define — Registration Number, Invoice Date, Total (10%), Total (8%), Consumption Tax (10%), Consumption Tax (8%), Supplier Name — are applied across the entire batch, regardless of how each supplier formats their invoice.

Why Japan's Accounting Platforms Don't Solve Incoming Invoice Extraction

Japan's three dominant cloud accounting platforms — freee, MoneyForward Cloud, and Yayoi (弥生) — all delivered full Qualified Invoice System compliance before the October 2023 deadline. Each supports automatic T-number insertion on outgoing invoices, dual-rate consumption tax display, consumption tax return preparation, and 7-year electronic storage compliant with the Electronic Books Preservation Act (電子帳簿保存法).

But there is a fundamental asymmetry in what these platforms do. They are built to issue compliant qualified invoices and file the resulting consumption tax returns. They are not built to receive and extract data from incoming supplier invoices that arrive as PDFs, paper scans, or mobile photos.

Here's what happens in the actual workflow: a supplier emails a PDF invoice. The finance clerk opens it. freee/MoneyForward/Yayoi cannot read that PDF — they have no document extraction capability for incoming files. The clerk must manually type the registration number, dual-rate totals, and consumption tax amounts from the PDF into the accounting software's vendor payment entry screen. The software then handles the rest: journal entries, tax calculation, filing. But the extraction step — the bridge between "receiving an invoice" and "the data is in the system" — remains entirely manual.

MoneyForward Cloud does offer CSV import for batch journal entries (仕訳インポート), and freee supports Excel-based data import. Both are useful — but they require the data to already be in structured CSV or Excel format. The hard part — getting 200 invoices' worth of data into that structured format — is the step neither platform addresses.

For international companies using SAP Japan, Oracle Japan, or Microsoft Dynamics 365 with Japan localization, the gap is wider. These ERPs require manual configuration just to display qualified invoice fields in their vendor master records — let alone extract them from incoming documents. Dynamics 365's Japan localization, for example, requires setting up separate sales tax groups for qualified vs. non-qualified vendors, defining item-level tax groups for standard and reduced rates, and configuring invoice layouts — all of which presumes the data is already available in structured form.

How to Extract 200 Qualified Invoices Into a Single Consumption-Tax-Ready Spreadsheet

ImageToTable.ai uses column-name extraction: instead of drawing bounding boxes around fields like template-based OCR tools, you type the field names you want — "Registration Number," "Total (10%)," "Consumption Tax (8%)" — and the AI locates each value anywhere on the page by understanding what it means. This is the mechanism that makes batch processing across diverse supplier formats possible: the extraction instruction is semantic ("find the T+13-digit registration number on this page") rather than positional ("look at coordinates X=150, Y=230").

The batch workflow for monthly qualified invoice processing:

The key advantage for consumption tax filing: the consumption tax columns in your export — the amounts exactly as printed on each qualified invoice — are the raw inputs for accumulation calculation (積上げ計算). You sum the "Consumption Tax (10%)" column, sum the "Consumption Tax (8%)" column, multiply each by 78/100, and you have your purchase tax credit for the filing period. No per-invoice calculator work, no rounding errors from manual entry.

Automatic 8% / 10% Tax Rate Classification Across 200 Invoices

The single most time-consuming manual task created by the Qualified Invoice System is the dual-rate split. Before October 2023, a finance clerk entered one total per invoice. Under the new system, line items must be sorted into two rate buckets — 8% reduced rate (food and beverages not consumed on-premises, newspapers) and 10% standard rate — with separate totals and consumption tax amounts recorded for each.

For a restaurant supply invoice mixing kitchen equipment (10%) and food ingredients (8%) across 15 line items, that sorting alone adds 2–3 minutes to the data entry for that one invoice. For 200 invoices with mixed-rate line items, the sorting overhead alone can consume 3–4 hours per month.

The tool handles this through inferred columns — a mode where AI reads document content and determines which category each item belongs to, even when the category isn't explicitly labeled. You define a column like Tax Rate (options: 8% Reduced / 10% Standard), and the AI reads each line item, identifies whether it's subject to the reduced rate based on Japan's consumption tax classification rules (軽減税率対象品目), and fills the result. Japan's reduced-rate classification — food and beverages not consumed on-premises and not alcoholic, plus newspaper subscriptions — is specific enough that a large language model can apply it from item descriptions.

The result at batch scale: instead of a finance clerk manually sorting through line items on 200 invoices, the spreadsheet output already has separate columns for 8% totals, 10% totals, and their respective consumption tax amounts — correctly classified, per invoice, per rate. The accumulation calculation inputs are ready to sum.

Registration Number Extraction Meets Electronic Books Preservation Act Searchability in One Pass

The Electronic Books Preservation Act (電子帳簿保存法) requires that all electronically stored transaction documents be searchable by three criteria: transaction date (取引年月日), transaction amount (取引金額), and counterparty (取引先). Furthermore, these three criteria must support range specification (e.g., "all invoices between ¥10,000 and ¥50,000") and combination search (e.g., "all invoices from Supplier X in March 2026").

For businesses with annual revenue exceeding ¥50 million, these searchability requirements are mandatory. For those under ¥50 million, the requirements are relaxed if the business can respond to a tax official's download request — a concession that doesn't eliminate the need for searchable records, only simplifies the technical implementation.

A folder of 200 supplier PDFs named however the supplier titled them — "invoice_march.pdf," "請求書20260315.pdf," "Scan001.pdf" — is not compliant. A spreadsheet where each row contains the structured data extracted from those PDFs — Registration Number, Supplier Name, Invoice Date, Total (10%), Total (8%), Consumption Tax (10%), Consumption Tax (8%) — is. Every criterion the law requires you to search by is a column in your spreadsheet. Date range filtering, amount filtering, counterparty search — all become native spreadsheet operations.

This is a secondary benefit of batch extraction that few discussions of the Qualified Invoice System acknowledge: the act of extracting structured data from invoices simultaneously satisfies the electronic storage searchability requirements. You still need to retain the original PDFs for the 7-year period mandated by the Consumption Tax Act — but for rapid retrieval during a tax audit, the extracted spreadsheet is the tool you actually use.

The extracted T-numbers (T+13 digits) also create a verification-ready dataset. The NTA's Qualified Invoice Issuer Publication Site allows lookup by registration number. With 200 T-numbers in a spreadsheet column, you can spot-check a sample against the registry — or, for high-risk suppliers, verify the entire column — without manually typing each 14-character code into a browser lookup tool.

October 2026: When the Batch Problem Gets 2.5× More Expensive

The Qualified Invoice System's transitional deduction rates for purchases from unregistered suppliers tighten on October 1, 2026:

For a company processing 200 invoices per month with 15% of suppliers unregistered (a realistic number for firms with SME and sole-proprietor vendors), the October 2026 drop from 80% to 50% deductibility raises the stakes. Getting the registration number wrong — or missing the fact that a supplier is unregistered and applying the wrong deductibility rate — now costs 2.5× more per invoice in lost input tax credit. Batch extraction doesn't make compliance decisions, but it eliminates the manual step where transcription errors introduce risk.

Frequently Asked Questions

Can I batch-process both qualified invoices (適格請求書) and simplified qualified invoices (適格簡易請求書) together?

Yes. Simplified qualified invoices — used by retailers, restaurants, and taxi operators selling to unspecified customers — omit the recipient name field. When you batch-process a mix of full and simplified invoices, the tool extracts whatever fields are present on each document. If "Recipient Name" is one of your defined columns, simplified invoices simply leave that cell blank in the output spreadsheet. No separate batch, no separate column configuration needed.

What if a supplier's invoice uses Japanese era dates (令和X年Y月Z日) instead of Gregorian?

The extraction output preserves the original date format from the invoice. If your downstream workflow requires Gregorian dates (e.g., for ERP import), the tool's post-processing layer can convert era dates automatically during export. You define the column as "Invoice Date" and the AI extracts whatever date format appears — 令和8年5月28日 becomes 2026-05-28 on export with format conversion enabled.

How long does batch-processing 200 invoices take?

Processing time scales with total page count, not file count. A batch of 200 typical single-page qualified invoices completes in well under 10 minutes. Multi-page invoices — a consolidated monthly statement from a major supplier with 5 pages of line items, for example — add proportionally to the processing time. All files in a batch share the same column definitions, so you define your qualified invoice fields once and they're applied across every document.

Does the tool verify T-numbers against the NTA registry?

No. The tool extracts the T+13-digit registration number as it appears on the invoice — it does not connect to the NTA's Qualified Invoice Issuer Publication Site to verify whether the number is active and valid. What it eliminates is the manual typing of those 14-character codes. You have the extracted T-numbers in a spreadsheet column, ready for verification. The verification step remains separate, but it becomes a review task (sample-check a column of numbers) rather than a data entry task (type 200 codes into a browser lookup tool).

Does batch extraction satisfy Electronic Books Preservation Act (電子帳簿保存法) storage requirements?

Batch extraction produces a structured spreadsheet that satisfies the searchability requirements — date, amount, and counterparty search with range specification and combination search. However, the law still requires retention of the original documents (the PDFs, scans, or photos) for the 7-year period. The extracted spreadsheet is your searchable index; the original files are your legal record. Both must be preserved.

Can I import the output directly into freee, MoneyForward, or Yayoi?

Yes, through their CSV import functions. MoneyForward Cloud supports journal entry import (仕訳インポート) where you map the spreadsheet columns to journal entry fields — including invoice classification (インボイス区分: qualified, 80% deduction, etc.). freee supports Excel-based data import for transaction entry. The extraction output gives you the structured data these import functions require. The import mapping step is a one-time setup per accounting platform; once configured, monthly batch extraction → CSV export → import becomes a repeatable workflow.

Extract 200 First, Calculate Once

The Qualified Invoice System's real operational impact isn't the three new fields on each invoice — it's what those fields force you to aggregate before your consumption tax return software can begin its work. Accumulation calculation (積上げ計算) doesn't accept totals. It needs individual consumption tax amounts from individual invoices, extracted as-is from whatever format each supplier chose to use.

The October 2026 transition from 80% to 50% input tax credit deductibility makes this aggregation more consequential, not less. The difference between extracting T-numbers correctly across a batch of 200 invoices versus mistyping a handful is real money — in lost input credit, not just processing cost.