Convert tax return PDF to Excel on Mac (2026 guide)

The loan file is complete except for the part that matters: the borrower is self-employed, and the underwriter’s income analysis wants two years of tax returns broken into numbers the cash-flow worksheet can use. What arrived is two PDFs — a 2024 return at 38 pages and a 2025 return at 41 — each a stack of Form 1040, Schedule 1, Schedule C, Schedule E, Form 4562, and state forms. The worksheet wants Schedule C line 31, the depreciation from line 13, the business miles from Part IV, the Schedule E rents by property. The numbers are all printed. The workflow, at most desks, is still a highlighter and re-typing into Excel.

Everyone who works downstream of a finished tax return meets this. The mortgage underwriter or loan processor spreading self-employed income into a Fannie Mae Form 1084-style cash-flow analysis. The financial advisor onboarding a client, building the planning model from three years of returns the client exported from their filing software. The family-law attorney turning five years of returns produced in discovery into an income exhibit. The CPA inheriting a client whose prior preparer sent locked PDFs and nothing else. The business owner who wants a year-over-year view of the Schedule C without re-opening tax software. The returns hold the data; the data is trapped in a form layout.

This guide is the Mac workflow: convert tax return PDFs to Excel on Mac — read on-device, only the text sent to ignitai’s servers — with one row per line item, the tax year on every row, and SSNs kept out of the output.

In this guide10 sections

Why tax returns defeat table extractors

A tax return is a form, not a table, and grid-based extractors need tables:

1. There is no grid. A 1040 page is labels, line numbers, checkboxes, and amounts arranged in a two-column form layout. A grid detector finds either nothing or a scramble of label fragments and stray digits. The structure you want — line number, label, amount — exists in the reading of the form, not in its geometry.
2. Every return is a different bundle. One taxpayer files a bare 1040; the next attaches Schedules 1 through 3, C, E, SE, Form 4562, and 8995. A layout template built for one return misses half of the next one. The extraction has to identify each form by its printed header, not by page number.
3. The layouts change almost every year. The IRS reshuffled the 1040 in 2018 and has moved lines between the 1040 and its numbered schedules repeatedly since. A parser tuned to 2024’s line positions misreads 2022’s — and a multi-year comparison is precisely the job that spans layouts.
4. Half of what you receive is scanned. Clients photograph paper returns; discovery productions are scans of scans, sometimes stamped “CLIENT COPY” diagonally across the numbers. Layout parsers that need a clean text layer return nothing.

The fix is an extractor that reads the form the way a person does — this is line 31, that’s its label, that’s its amount, this watermark isn’t data — and takes its instructions in plain English.

Method 1: ignitai on Mac (the read-on-device way)

ignitai extracts by description rather than by layout template, so one prompt covers a 2022 TurboTax print, a 2025 preparer copy, and the scanned 2019 return alike:

1. Drag the PDFs in — or use the file picker, or right-click → Open With from Finder. Born-digital software prints and scanned paper returns go through the same flow; the scanned-PDF walkthrough covers capture quality for the paper ones.

2. Describe the output shape in plain English. A prompt is required — the Convert button stays disabled without one. Start from the Form starter or write your own:

   “One row per line item: tax_year, form (1040, Schedule 1, Schedule C, Schedule E, etc.), line_number, label, amount. Tax year from the form header. Skip blank lines, instruction text, and signature blocks. Omit SSNs. Bare amounts, minus sign for negatives, no thousands separators.”

   The tax_year and form columns are what make the flat sheet navigable — filter to form = Schedule C and the business income analysis is right there. Save the prompt as a preset; it syncs to your other devices via iCloud.

3. Pick Excel, hit Convert. Each return is read on your Mac — never uploaded — and only the recognized text goes to ignitai’s servers to build the sheet. On macOS 26, Apple’s native on-device document reading is sharper on dense form pages; the same flow works back to macOS 14.4. A typical return is usually ready in under a minute.

4. Open the result in Excel or Numbers. One worksheet, one row per line item — bare amounts, consistent labels — and the amounts are real numbers in the Excel file, so the pivots run the moment it opens. In ignitai, rows are numbered and the original page is a click away, which makes tying a line back to the return quick; edits happen in the spreadsheet.

All the years, one batch. Select the 2023, 2024, and 2025 returns together and run them as a single batch with the one prompt — the batch-conversion walkthrough covers the mechanics. They consolidate into one sheet, and because the prompt pulls the tax year from each form’s header, a pivot with line items as rows and years as columns is the year-over-year comparison. Filenames never leave your Mac, so identity always comes from the printed page — for multi-client work, add “taxpayer name from the form header” as a column, or run one client at a time to keep the workbooks separate. Photographed paper returns run as their own batch: a drop mixing PDFs and images imports only the PDFs.

One table per run — pick the grain. One row per line item is the analysis grain. The depreciation detail behind Form 4562, or a state return’s own schedules, are different shapes — a second run with a second prompt, not extra sheets in the same workbook. Two tables means two runs.

Method 2: Excel for Mac’s Get Data (no PDF connector)

Excel for Mac ships Power Query under Data → Get Data, but the Mac connector list — Text/CSV, XLSX, XML, JSON, SharePoint, SQL Server, OData — has no From PDF option; that connector is Excel-for-Windows-only. Even on Windows it’s the wrong tool for this document: the From PDF connector detects tables, and a 1040 has none — it returns fragments of form text, not line items. Against a scanned return it returns nothing.

Method 3: a script (a parser per layout year)

For a born-digital return, pdfplumber can read the text layer — and then the real work starts: mapping text positions to line numbers, per form, per revision year. The IRS’s habit of moving lines between the 1040 and its schedules means the parser needs a layout table for each tax year it will ever see, and the scanned returns stay out of reach without bolting on OCR. A lender processing thousands of returns through a commercial parsing API is a different economics; for an advisor or attorney handling returns by the folder, the maintenance never amortizes.

Method 4: web converters (the never-upload line)

Generic web PDF-to-Excel tools fail the form-vs-table problem like everyone else — and on this document the quality question is beside the point. A tax return carries the filer’s SSN, the spouse’s, every dependent’s, the full income picture, and — in the refund section of the 1040 — a bank routing and account number. Uploading that to a free converter you don’t control is the confidentiality decision professional-conduct rules exist to prevent. With ignitai the flow is narrower, so state it exactly before you run client returns: the PDF stays on your Mac, the recognized text (SSNs included) goes to ignitai’s servers, where the AI builds the sheet, the finished file is kept in your ignitai account until you delete it, and “omit SSNs” keeps them out of the workbook you forward, not out of the text that is sent. Check that against your engagement letter and client-consent rules before using it on client data.

On iPhone and iPad

Returns don’t always arrive at the desk:

• iPhone — the client brings the paper return to the meeting: capture the pages you need with the in-app camera scan, apply the saved preset, and the line items are rows before the meeting ends. A Live Activity shows conversion progress from the lock screen. A PDF in Mail goes in through the share sheet the same way.
• iPad — reviewing a discovery production away from the desk: save the PDFs to Files, multi-select from the picker or share sheet (there’s no external drag-and-drop into the app on iPad), convert with the synced preset, and iCloud puts the XLSX on the Mac where the exhibit gets built.

One subscription covers iPhone, iPad, Mac, and Vision Pro, so the preset travels with the engagement.

The tie-out: check the return’s own math

A tax return is unusually easy to verify because it cross-references itself. Three checks in the extracted sheet:

• Components sum to totals. The income lines on the 1040 plus Schedule 1’s additional-income total should reproduce the total-income line; Schedule C’s expense lines should reproduce its total-expenses line. A SUMIFS over the extracted rows against the printed total catches a misread digit immediately — and tells you which form to inspect.
• Flow-through lines match. Schedule C line 31 should reappear on Schedule 1; Schedule E’s result should flow the same way. If both cells in the sheet agree, both reads were right.
• Year columns are complete. In the multi-year pivot, a line that’s blank in one year is either a real change (the business had no depreciation that year) or a page that didn’t make the scan. Thirty seconds against the source PDF settles which.

Then the sheet feeds the job that started this: the cash-flow analysis with its add-backs (depreciation from Schedule C line 13, the depletion and home-office adjustments), the planning model’s effective-rate trend, the support-calculation exhibit, the year-over-year Schedule C the owner wanted. When the engagement also involves W-2 wage history or a folder of 1099s, those have their own Mac flows — and the underwriter pairing returns with bank statements can row those up the same way.

When not to use this

• You can get the data instead of the print. Your own filing software exports its data; a preparer can often re-send a return with the preparer summary or as an organizer export. If a structured source exists, take it.
• An IRS transcript answers the question. For simple verification — did they file, what was the AGI — a free IRS return transcript may be enough. But transcripts are summary-level, lag the filing by weeks or months, and third parties need a signed 4506-C and processing time. When you need line-level detail now, from the PDFs in hand, extraction is the path.
• You need what isn’t printed. The extracted sheet holds what the return shows — not the workpapers, not the receipts behind Schedule C. When the question is “prove this deduction,” you still need the source documents.

Bottom line

For tax returns that need to become numbers on a Mac: gather the years, drag the stack into ignitai, write the one-paragraph prompt (one row per line item, tax year and form on every row, skip the blanks, omit SSNs), pick Excel, Convert, then check the return’s own cross-references before the analysis starts. Build a per-layout-year parser only if you’re processing returns at institutional volume; keep returns away from web converters entirely. The engagement’s hours belong in the income analysis, the plan, or the exhibit — not the re-typing — and the on-device read means the most sensitive document your client owns never left your Mac to get there.

Get ignitai on the App Store — free download; your first document (up to 4 pages) converts free with Sign in with Apple, and one subscription covers iPhone, iPad, Mac, and Vision Pro.

FAQ

How do I convert a tax return PDF to Excel on a Mac?

Drag the return into ignitai, describe the rows in plain English — tax year from the form header, form or schedule name, line number, label, amount, skip blank lines and omit SSNs — pick Excel, and hit Convert. The PDF is read on your Mac, and only its text goes to ignitai’s servers; the result is one sheet with one row per line item.

Can I build a multi-year comparison from several returns?

Yes — select all the years and run them as one batch with one prompt. Include the tax year from each form’s header as a column; the years consolidate into a single sheet, and a pivot with line items as rows and tax years as columns is the comparison. Filenames never leave your Mac, so the year has to come from the printed page — which every return carries.

Is this safe? A tax return has my SSN on it.

The PDF itself is read on your Mac and never uploaded; only the recognized text is sent to ignitai’s servers to build the spreadsheet. Write “omit SSNs” in the prompt so they never land in the workbook you’ll be emailing around — and keep returns away from free web converters entirely, because uploading the whole PDF hands over SSNs, dependents’ SSNs, and the bank account printed in the refund section.