Storing Electronic Transaction Data Is More Than a Folder of PDFs

The short answer

Invoices and receipts exchanged by email or downloaded from a website do not, as a rule, meet the storage requirements simply because the PDF sits in a folder. Storage means two things at once: the data can be retrieved when it is needed, and it cannot be quietly rewritten afterwards. Japan's National Tax Agency sets out, for electronic transaction data, that measures against alteration are taken, that the data can be displayed on screen and printed, and that it can be searched by transaction date and other dates, transaction amount, and counterparty. The search requirement, however, does not apply to everyone: depending on the size of the business and on whether the data can be handed over during an inspection, it may not be needed (see below).

Why "just put it in a folder" falls short

What gets asked for during an audit is not the existence of a file but the ability to produce a particular transaction's data on the spot. A folder holding a thousand PDFs does not meet that bar if finding "last August, the 110,000 yen payment to A Trading" takes half an hour.

The second issue is rewriting. A PDF can be overwritten. Leave it in a folder anyone can write to and there is no way to establish later whether an amount was edited. The tax agency's guidance asks for one of the following: receive data that already carries a time stamp, add a time stamp to data you receive, exchange and store the data in a system that keeps a history of corrections and deletions, or set internal procedural rules against alteration, operate them and keep them on hand. Choosing the rules route means writing those rules down and actually following them. The agency publishes sample rules on its website.

Which documents this covers

Paper and electronic arrivals are treated differently. What arrives as an email attachment or a download is itself the record that has to be kept. Printing it, filing the paper and deleting the data is not an available option now. The agency's Q&A explains that the former option of keeping printouts instead was abolished, and that the data itself has to be kept.

There is a grace measure for businesses that cannot yet meet the standard rules. According to the agency, if the head of the competent tax office accepts that there was a reasonable cause for not storing the data under the standard rules, and the business can respond during an inspection both to a request to download the data and to a request to present or submit printouts of it, simply keeping the data is enough. No advance notification is needed, and the agency lists staff shortages and lack of funds for systems among the reasons it accepts. Even then, the data itself must not be deleted.

Scanning invoices that arrived on paper is a separate set of requirements. This article stays with electronic transaction data and does not cover scanned paper storage.

What "searchable" means in practice

The search items named are transaction date and other dates, transaction amount, and counterparty. On top of searching by those three, the agency's guidance requires one of two things: either range searches on date or amount plus searches combining two or more items, or the ability to respond when a tax official asks to download the data during an inspection.

Without dedicated software, there are broadly two ways to make records searchable by the three items.

Option 1: put the three items in the filename

This is the baseline when no dedicated software is involved: names like "20260817_ATrading_110000.pdf" and "20260820_YamadaKomuten_48500.pdf", with the date, counterparty and amount always in that order. The agency's Q&A gives an example in the same form and describes entering all three items in a consistent order. This approach assumes you can respond to a download request during an inspection.

Write dates as eight digits so they sort correctly, use one consistent spelling per counterparty, and write amounts without separators. If "A Trading" and "A Trading Co." both appear, searching stops working, so fix counterparty names against a single list.

Option 2: keep an index and link it to the files

Build a sheet with columns for date, counterparty, amount and filename, and keep it aligned with the folder. The agency's Q&A says that if the spreadsheet's own functions let you search by range and by combinations of items, the search requirement can be considered met. This lets filenames stay short and lets you add columns such as the nature of the transaction or the person responsible. The weakness is drift between the index and the folder, so add a monthly check that the counts still match.

When the search requirement does not apply

The search requirement does not fall on every business in the same way. According to the agency, if you can respond to a download request during an inspection and one of the following applies, you do not need to meet the search requirement:

• sales in the base period (normally two years earlier: the year before last for sole proprietors, the business year before last for companies) were 50 million yen or less
• you can present or submit printouts of the data organised by transaction date and other dates and by counterparty

The agency's Q&A explains how to judge whether sales exceed 50 million yen, including how this differs from taxable sales for consumption tax. Check the agency's guidance to see whether you qualify. Even where the search requirement does not apply, the measures against alteration and the ability to display and print the data still do.

Five rules to write down

Decide these five at the outset. They describe how people behave, not what software does.

1. Location: which folder, and whether it is split by department or by month
2. Naming: the filename format, and the agreed list of counterparty spellings
3. Who files it: the recipient, or accounting in a batch
4. By when: how many days after receipt the file must be in place
5. Corrections: what happens to the original when a corrected invoice is reissued

The fifth is the one most companies skip. Overwriting with the reissued invoice erases the earlier record. Keeping both as separate files and marking in the index which one is effective leaves you able to explain the sequence later.

What this looks like in practice

Take a wholesaler receiving about eighty invoices a month by email, with one person in accounting. Each time an email arrives, they save the attachment and rename it. At thirty seconds each that is forty minutes a month. As volume grows the task slides down the list, until a hundred files are processed in one sitting at month end — and batching is exactly when naming mistakes multiply.

The part that can be reduced is the naming. If the date, amount and counterparty can be read out of the PDF and offered as suggestions, the person only reviews and corrects. That is the job of AI OCR, and OCR Reader is intended for it, though it is planned for release and not yet available. The limits of that reading are covered in a separate article. Where several invoices arrive combined in one PDF, splitting comes first.

Limits

Retention periods, which transactions are in scope, and how treatment varies with circumstances are all subject to change. This article does not assert them. Confirm the current requirements in the National Tax Agency's own guidance, and check how they apply to your situation with your tax accountant.

Backups are a separate topic from the requirements, but they matter in practice. If the only copy lives on one PC, a hardware failure takes everything. Treat meeting the storage requirements and not losing the data as two different problems.

Summary

• A PDF in a folder is neither retrievable nor protected from rewriting.
• The search items are transaction date and other dates, amount, and counterparty, although the requirement falls away for businesses with sales of 50 million yen or less, among others, that can respond to a download request.
• Either encode those in filenames or maintain an index linked to the files.
• Without one consistent spelling per counterparty, nothing is really searchable.
• Check retention periods and categories in the National Tax Agency's guidance.