{
  "question_id": "Q-0002",
  "slug": "how-to-digitize-paper-receipts",
  "display_title": "What's the best way to digitize paper receipts?",
  "format": "article-v2",
  "applies_to": {
    "countries": [
      "US"
    ],
    "frameworks": [],
    "tax_year": null,
    "platforms": []
  },
  "general_concept": true,
  "summary": "The best way is whichever method, at your volume and setting, yields a copy that shows the whole receipt legibly, is checked against the paper before you rely on it, can be found by its content, and survives the loss of any one device or account. A phone app suits a daily trickle and field capture, a document scanner suits desk batches, and a bulk service suits a large backlog you then check.",
  "body": "## What must a digital copy of a receipt show and be?\n\nSettle the standard before choosing a tool. The IRS's Publication 583 says all requirements that apply to hard copy books and records also apply to electronic storage systems that maintain tax books and records. It says such a system must index, store, preserve, retrieve and reproduce the stored records in legible format, and must provide a complete and accurate record of your data that is accessible to the IRS. How long records must be kept, and whether the paper may then go, are separate questions.\n\nPublication 583 points to Rev. Proc. 97-22 for the details. That revenue procedure covers systems that electronically image hardcopy documents, so your capture tool, storage and index together form one. Section 4.01 of Rev. Proc. 97-22 sets these general requirements:\n\n| Rev. Proc. 97-22 requires | What that means for receipts |\n|---|---|\n| An accurate and complete transfer of the paper, and the ability to index, store, preserve, retrieve and reproduce what is stored | The image shows the whole receipt, and every file can be found and printed |\n| Reasonable controls to ensure the system's integrity, accuracy and reliability | Decide who can add, edit and delete files |\n| Reasonable controls to prevent and detect unauthorized creation, addition, alteration, deletion or deterioration of stored records | Keep a backup the people who capture cannot change, and at each periodic check compare the store with it so an unexplained new, changed or missing file is found |\n| An inspection and quality-assurance program, evidenced by regular evaluations that include periodic checks of stored records | Check every capture, and re-check stored files on a schedule |\n| A retrieval system that includes an indexing system, and the ability to reproduce legible and readable hardcopies | Each file can be located from a record of what it is, and printed readably |\n| A high degree of legibility and readability on screen and in hardcopy, which the reproduction process must maintain | Every letter and numeral can be identified positively and quickly, with none that could be taken for another (a 0 that might be an 8), and groups read as words and complete numbers |\n| Information that supports your books, for example cross-referenced to give an audit trail between the general ledger and the source documents | Each file ties to its entry in the books |\n| Complete descriptions of each system, including all procedures for its use, and of its indexing system, available to the IRS on request | Write down how you capture, store and index |\n| At an examination, or for IRS tests of the system, retrieving and reproducing the records, including hardcopies if requested, and providing the resources needed to locate, retrieve, read and reproduce them | Produce any receipt, and the means to view it, on request |\n| No agreement, such as a contract or license, that would limit or restrict the IRS's access to and use of the system | Read the terms of any app or service that holds the files |\n| Retaining the stored records for as long as the IRS's recordkeeping regulation requires | How long is a separate question |\n| Each system meeting these requirements if you use more than one; records in a system whose hardware and software you stop maintaining are deemed destroyed unless they remain available to the IRS in conformity with the procedure | If receipts sit in more than one system, each must meet every requirement here; keep files that open without any one app |\n| Reasonable data compression or formatting, provided these requirements are still met | Compression that costs legibility fails |\n\nThe legibility and readability definitions are the test every capture must pass. What a receipt must itself show to support an expense is a separate question; the copy's job is to carry everything the paper shows.\n\n## Which capture method fits your volume and setting?\n\nChoose on how many receipts arrive and when, where they are captured, who handles the paper and where the file lands:\n\n| Method | Fits | Watch for |\n|---|---|---|\n| Phone or tablet app | A daily trickle, and capture where the purchase happens | Light, flatness and focus vary with every shot, and files can sit on one device or in one person's account |\n| Document scanner | Batches at a desk, and a backlog you scan in-house | Whether the model feeds narrow, curled and long receipts and saves a standard file to a folder you control; check its own documentation |\n| Bulk conversion service | A backlog too large for in-house time | The paper leaves your hands before capture |\n\nRev. Proc. 97-22 says that using a third party, such as a service bureau, to provide the system does not relieve you of its responsibilities. So before using a service, get a written description of its procedures, read its terms for anything limiting access, count what you send, and have the paper come back to you. If it offers to dispose of the paper, decline at this stage: the check and the separate decision about originals stay with you.\n\nTools also change images. Google's undated help for scanning with the Drive app on Android lists steps to adjust the colors or make the image grayscale, and a Clean tool to erase stains, fingers and more. An edit that removes marks can remove detail, so compare the image with the paper after the last edit. Comparing products is a separate question.\n\n### What changes when people capture in the field?\n\nThree things move out of sight when capture happens away from the desk:\n\n- **Quality.** Give everyone the same shooting rules. Expensify's undated SmartScan troubleshooting page, written for its own app, advises laying the receipt flat on a contrasting surface, capturing the full receipt with all edges, the total, date and merchant name visible, and avoiding shadows and glare.\n- **Destination.** Set every device to save into a folder or account the business owns, never only to a camera roll or a personal account, so a lost phone or a departing employee takes no receipts away.\n- **Central check.** Someone other than the capturer, where staffing allows, opens each new file from the business store, confirms it is whole and legible, and asks for a recapture while the paper still exists. Working alone, do this check yourself at a set time each week. The paper stays with the capturer, or comes in, until its file passes.\n\n## What keeps an image legible and openable later?\n\nLegibility is set at capture and must then survive compression and printing; Rev. Proc. 97-22 allows reasonable compression only while its requirements are met and requires the reproduction process to maintain legibility. Google's undated help for converting files to text in Drive gives, as tips for the best results with its own text recognition, sharp images with even lighting and clear contrasts, documents right-side up, and text at least 10 pixels high. Whatever the tool, if compression or a filter blurs or thins small print, change the setting and capture again.\n\nBecause of the deemed-destroyed rule in the table, keep each receipt as a standard image or PDF file that opens without the capture app, export files from any app that holds them only in its own format, and test by opening a sample on a device that does not have the app.\n\n## How do you check each capture before relying on it?\n\nCheck while the paper is still in your hand; a capture is not finished until step 4 passes. Work through these steps for each receipt, in this order:\n\n1. Flatten the receipt and lay it on a contrasting surface in even light.\n2. Capture the whole receipt, one page per section of a long receipt, and the back if anything is printed or written there.\n3. Crop, rotate or adjust the image if needed.\n4. Open the file from the business store, not the app, and compare it with the paper: all four edges, every line, the vendor, date, items, tax, total and payment details, zoomed until every letter and numeral can be identified positively and quickly, with none that could be read as another.\n5. If anything is missing, cut off or unreadable, capture again, keeping the best attempt until one passes; if none passes, keep that best attempt and handle it as a faded receipt below.\n6. Record its identifying attributes, and correct any extracted data against the image.\n7. Put the paper in a holding tray labelled with the date or batch; whether it may later be discarded is a separate decision.\n\n## Why keep the image behind the data an app extracts?\n\nExtracted data is a reading of the receipt, not a copy of it. Expensify's undated SmartScan troubleshooting page says SmartScan reads receipt images and automatically fills in details like the amount, date and merchant; that it may misread certain receipts, especially handwritten receipts, faded ink or unusual formats; and that it requires a readable date and total amount on the receipt. Check each extracted field against the image, correct it there, and keep the image attached to the data. If only the data survives, an extraction error can never be found.\n\n## How do you make a receipt findable by its content?\n\nTwo things make a store searchable. The first is recognized text: Google's undated help for scanning with the Drive app on Android says it scans receipts to save them as searchable PDFs. That text exists only if your tool makes it: Drive's Android scanner saves as .pdf or .jpg, and its page describes the result as searchable PDFs, so choose .pdf there; for any other tool, check its own documentation, and test by searching the store for a word printed on a receipt you have just captured. Recognized text can be wrong, so search for amounts in the index rather than the text. The second is an index: Rev. Proc. 97-22 defines an indexing system as one that permits the identification and retrieval of stored records for viewing or reproducing, gives as an example a unique number for each stored document with a separate database describing each one, and says the requirement is met if the system is functionally comparable to a reasonable hardcopy filing system.\n\nRecord these attributes for each file, in an index or the file's properties:\n\n- A unique identifier\n- The purchase date, vendor and total\n- The payment method, such as the card used\n- The entry in the books it supports, which gives the audit trail\n- When it was captured, by whom, and that it passed the check\n\nRev. Proc. 97-22 also requires reasonable controls to protect the indexing system against unauthorized creation, addition, alteration, deletion or deterioration of its entries: keep the index in the business store, limit who can change it, and include it in every backup.\n\nThe naming and folder convention is a separate question, as is matching receipts to bank and card transactions.\n\n## How should copies be stored so one failure loses nothing?\n\nThe US-CERT paper Data Backup Options, published by CISA, gives the 3-2-1 rule: keep 3 copies of any important file, 1 primary and 2 backups; keep them on 2 different media types to protect against different types of hazards; and store 1 copy offsite, outside your home or business facility.\n\nThe same paper warns that cloud users may become locked into one provider and surrender most of their control over their own data, so do not let a capture app's cloud hold the only copy. It also says rolling backups can silently propagate corruption or malware from primary files to backups, so keep at least one copy that is not overwritten automatically. Keep the store in an account the business owns, with more than one person able to recover it where staffing allows.\n\nRev. Proc. 97-22 calls records-management practices a business decision within the taxpayer's discretion and lists among them testing to confirm records integrity. Test by opening a few files restored from each backup on a schedule.\n\n## How do you handle long, oversized and faded receipts?\n\nThese forms fail quietly, producing a file that looks complete and is not:\n\n| If the receipt is | Then |\n|---|---|\n| A clean single page | Use the standard check. |\n| Long, multi-page or oversized | Capture it in overlapping sections as pages of one file, top to bottom, so the header and the total are both included and no line falls between sections. Google's undated help for scanning with the Drive app on Android, for example, includes a step to scan another page into the same scan. Check each join against the paper. |\n| Faded or partly unreadable | Capture without filters that lighten or thin the print, and try a scanner if a phone fails. If any part of the print still cannot be identified, the image is not a complete copy: label the file partly illegible, record what can be read, and mark the paper as not converted on a slip or sleeve, never on its print, keeping it apart from captured paper so the separate decision about originals sees it for what it is. Rev. Proc. 97-22 lists retaining hardcopies of records that are illegible or cannot be accurately or completely transferred among its recommended records-management practices. |\n\n## How do you run new paper as a routine and the backlog as a project?\n\nNew paper needs a routine that never ends: a fixed time each day or week, the per-receipt check, one tray for paper awaiting capture and one for captured paper awaiting the separate decision, and periodic checks that stored files still open. At each periodic check, print a few stored receipts, including a long one, at normal size and confirm every letter and numeral can be identified positively and quickly; Rev. Proc. 97-22 requires legibility on screen and in hardcopy. Receipts that arrive by email or download go into the same store with the same attributes, so one search covers every receipt.\n\nA backlog is a bounded project, run in this order:\n\n1. Count the boxes and the years they cover, and choose a scanner or a service on that volume.\n2. Unfold and unstaple, and set long and faded receipts aside for separate handling.\n3. Work in batches, such as one box or month, logging the number of paper receipts and the number of files in each. Capture the back of any receipt with print or handwriting on it, and ask a service to scan both sides.\n4. Check each batch before starting the next: counts match, and each image is compared with its paper for all four edges, every line and anything printed or written on the back; recapture any that fails before the batch is boxed.\n5. Set an end date, and send new paper into the routine from the first day, not into the backlog.\n\nMoving a whole paper system to digital without losing its history is a separate question.\n\n## What does one receipt look like from capture to retrieval?\n\nA technician for a three-person plumbing business buys fittings at a supply house for 186.40 on the business card:\n\n- **Captured.** At the counter he flattens the receipt on a dark folder and photographs it with an app set to save searchable PDFs into the business's shared receipts folder, whose search reads their text.\n- **Verified.** Opening the file from the shared folder and comparing it with the paper, he sees the total line is cut off and retakes it. The second shows all four edges, and no digit of 186.40 could be taken for another when zoomed.\n- **Identified.** That evening the owner opens the file from the shared folder, confirms it passes, and records ID R-0412, the date, the vendor, 186.40, the card, the job and the ledger entry. The app had read the total as 186.48, so she corrects it from the image.\n- **Stored.** The folder is the main copy; a nightly copy goes to an external drive only the owner can change, and each weekly copy goes, dated and never overwritten, to storage away from the premises. The index, kept in the same folder, goes with every copy.\n- **Retrieved.** Months later a customer queries the job's materials. Searching the store for \"fittings\" finds the file through its recognized text, and the R-0412 index entry leads to the ledger entry and the card charge.\n\nThe paper went into that week's tray; what happens to it is a separate decision.",
  "sources": [
    {
      "id": "REF::1",
      "url": "https://www.irs.gov/publications/p583",
      "title": "Publication 583 (12/2024), Starting a Business and Keeping Records",
      "publisher": "Internal Revenue Service",
      "published": "Revised December 2024",
      "retrieved_at": "2026-10-01T00:41:45+00:00",
      "sha256": "da630a3acccd2603e6f36b5127d52a646a606fb97bcced74308b392f08847226",
      "supports": [
        "C1",
        "C2",
        "C3",
        "C4"
      ]
    },
    {
      "id": "REF::2",
      "url": "https://www.irs.gov/pub/irs-irbs/irb97-13.pdf",
      "title": "Internal Revenue Bulletin No. 1997-13: Rev. Proc. 97-22 (books and records; electronic storage; imaging)",
      "publisher": "Internal Revenue Service",
      "published": "Bulletin No. 1997-13, March 31, 1997",
      "retrieved_at": "2026-10-01T00:41:45+00:00",
      "sha256": "67d93ff81e42092198b0dae049db482e5608ebbe45ce6f6728c83d9dda2668d5",
      "supports": [
        "C5",
        "C6",
        "C7",
        "C8",
        "C9",
        "C10",
        "C11",
        "C12",
        "C13",
        "C14",
        "C15",
        "C16",
        "C17",
        "C18",
        "C19",
        "C20",
        "C21",
        "C22",
        "C23",
        "C24",
        "C25",
        "C26",
        "C27",
        "C28",
        "C29",
        "C30",
        "C32",
        "C33",
        "C55",
        "C56",
        "C59",
        "C60"
      ]
    },
    {
      "id": "REF::3",
      "url": "https://www.cisa.gov/sites/default/files/publications/data_backup_options.pdf",
      "title": "Data Backup Options",
      "publisher": "Cybersecurity and Infrastructure Security Agency (produced for US-CERT)",
      "published": "2012 (page footer: (c) 2012 Carnegie Mellon University, produced for US-CERT)",
      "retrieved_at": "2026-10-01T00:41:45+00:00",
      "sha256": "dd87cf959d4a68da37cb5393ad0648fc2ad4eb090534138144b4eb7e15930450",
      "supports": [
        "C34",
        "C35",
        "C36",
        "C37",
        "C38",
        "C39"
      ]
    },
    {
      "id": "REF::4",
      "url": "https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid",
      "title": "Scan documents with Google Drive (Android), Google Drive Help",
      "publisher": "Google",
      "published": "undated",
      "retrieved_at": "2026-10-01T00:48:36+00:00",
      "sha256": "b744a0f0c0a2d7f50b9a8b1be9b2f6b578897ba112366a0c5ed2d6daa240d1e5",
      "supports": [
        "C40",
        "C41",
        "C42",
        "C43",
        "C58"
      ]
    },
    {
      "id": "REF::5",
      "url": "https://support.google.com/drive/answer/176692?hl=en",
      "title": "Convert PDF and photo files to text, Google Drive Help",
      "publisher": "Google",
      "published": "undated",
      "retrieved_at": "2026-10-01T00:41:45+00:00",
      "sha256": "631fa18ba85ee010a737ae8b9dd3ea3fc71060a0cacc0886227938339855fb31",
      "supports": [
        "C44",
        "C45",
        "C46",
        "C57"
      ]
    },
    {
      "id": "REF::6",
      "url": "https://help.expensify.com/articles/new-expensify/reports-and-expenses/Troubleshoot-SmartScan-Issues",
      "title": "Troubleshoot SmartScan Issues",
      "publisher": "Expensify",
      "published": "undated",
      "retrieved_at": "2026-10-01T00:41:45+00:00",
      "sha256": "ce2a06afc7dd5c50ca0dc855ca1497aecb39b0637f3facfa645651df3ac234c5",
      "supports": [
        "C48",
        "C49",
        "C50",
        "C51",
        "C52",
        "C53",
        "C54"
      ]
    }
  ],
  "related": [
    {
      "question_id": "Q-0003",
      "slug": "whether-to-keep-paper-receipts-after-scanning-them",
      "display_title": "Do I need to keep paper receipts after scanning them?"
    },
    {
      "question_id": "Q-0001",
      "slug": "how-a-small-business-should-organize-receipts",
      "display_title": "How should a small business organize receipts?"
    },
    {
      "question_id": "Q-0004",
      "slug": "how-long-a-business-needs-to-keep-receipts-and-records",
      "display_title": "How long does a business need to keep receipts and records?"
    },
    {
      "question_id": "Q-0013",
      "slug": "how-to-name-and-file-digital-receipt-and-document-files",
      "display_title": "How should I name and file digital receipt and document files?"
    },
    {
      "question_id": "Q-2010",
      "slug": "how-to-move-from-a-paper-receipt-system-to-digital-without-losing-history",
      "display_title": "How do I move from a shoebox/paper system to digital without losing history?"
    },
    {
      "question_id": "Q-0016",
      "slug": "how-to-match-receipts-to-bank-and-credit-card-transactions",
      "display_title": "How do I match receipts to bank and credit-card transactions?"
    }
  ],
  "review_class": "consequential",
  "review_class_trigger": "claim_level_review_required",
  "provenance": {
    "author_model": "claude-opus-5-5",
    "reviewer_model": "claude-opus-5-5",
    "review_verdict": "ACCEPT",
    "review_source": "closure",
    "review_verdict_on_sha256": "daed7edb5c5f4d4788f5e430c9fcaffe0a1ea03b5851ab74b848a3927da99483",
    "editorial_disposition": "ACCEPT",
    "corrections": 1,
    "approved_by": null,
    "approved_at": null,
    "article_sha256": "daed7edb5c5f4d4788f5e430c9fcaffe0a1ea03b5851ab74b848a3927da99483",
    "source_map_sha256": "42731f4e2a5c26a71bc90865bac7bd265f4651017428790f1d17a13e57626887",
    "transform_sha256": "f796d895bcd11d397f46152c9ddcea8912b21b9bb78f24d1ae9d238fe45137bf"
  },
  "offer": "ask",
  "offer_id": null,
  "sample_target_id": null,
  "datePublished": "2026-10-01T20:06:59Z",
  "reviewed_at": "2026-10-01T20:06:59Z",
  "content_sha": "2e0f9c8ef2909ade4bc550aecd4272eea37a211815b71e09d6c215be811b4d8c",
  "release": "2.14.0",
  "slug_provenance": "minted at first publication",
  "question_text": "What's the best way to digitize paper receipts?",
  "jsonld_types": [
    "Article"
  ],
  "related_question_ids": [
    "Q-0003",
    "Q-0001",
    "Q-0004",
    "Q-0013",
    "Q-2010",
    "Q-0016"
  ],
  "aliases": [],
  "alias_provenance": []
}