Convert PDF to Numbers on Mac (2026 native workflow)

You opened the PDF in Preview, selected the table, copy-pasted into Numbers, and got one column of mashed text where the rows should have been. You tried File → Open from Numbers and got a polite refusal — Numbers opens .numbers, .xlsx, .csv, and .tsv, not .pdf. You searched “convert PDF to Numbers” and got a list of web tools that want you to upload your file, then deliver an .xlsx you have to convert to Numbers anyway.

The gap is real. Numbers is the spreadsheet for a Mac-primary user — handoff to Pages, native iCloud sync to iPad and iPhone, sane chart defaults, the Smart Categories pivot view that Excel still can’t quite match. But the file format is a closed bundle, the import paths are narrow, and the world ships PDFs not XLSX.

This guide covers the workflow that actually works in 2026: convert PDFs to Apple Numbers spreadsheets on Mac, on-device on macOS 14.4+, with column types preserved, formulas re-applicable, and the round trip to iPhone and iPad working over iCloud.

Why Numbers (and not just Excel-then-rename)

If you already pay for Microsoft 365, the obvious move is “convert PDF to XLSX, open in Excel, save Numbers later.” For a Mac-primary user who lives in Numbers, the Excel detour is friction:

Number formatting differs. Excel’s currency cells survive a Numbers import, but conditional formatting, slider controls, checkboxes, and Smart Categories don’t. Anything Numbers-specific you set up after import is lost if you ever round-trip back to XLSX.
The chart defaults are different. A Numbers chart from a Numbers-native sheet looks like the Apple keynote charts; a chart in an imported XLSX inherits Excel’s defaults until you redo it.
iCloud sync is reliable for .numbers, less so for XLSX in the Numbers app. A .numbers file edited on Mac, opened on iPad, edited again on iPhone, and reopened on Mac just works. The same XLSX through Numbers occasionally rewrites cell types in ways that surprise you.
Numbers handles freeform layouts that Excel doesn’t. Multiple tables on the same sheet, floating images and shapes between tables, tables that aren’t anchored to A1 — all are first-class in Numbers and translate poorly through XLSX.

For a quarterly report that lives in Numbers across three Apple devices, you want the data to land in Numbers natively, not pass through XLSX on the way.

What “PDF to Numbers” actually means

There are three kinds of PDF a Mac user typically wants in Numbers, and they need different treatment:

A PDF with a table that’s literally a table. Bank statements, sales reports, supplier price lists. Drawn grid lines, rows are rows, columns are columns. The job is to extract the table and land it in a Numbers sheet with the right column types (numbers as numbers, dates as dates, currency as currency).
A PDF with implicit tabular data. Receipts, invoices, contracts with embedded line items, lab reports. The data is structurally a table — line items, totals, key/value metadata — but visually it’s a designed page with boxes, headers, and prose around it.
A scanned PDF. A paper document photographed or scanned. The “PDF” is really an image wrapped in a PDF container; the OS doesn’t see selectable text. OCR is a prerequisite before any extraction.

The first case has a few decent free paths. The second and third are where the workflow matters.

Method 1: ignitai → Numbers (the on-device way)

ignitai treats extraction as a language task: take the PDF (text or scan), describe what you want, get back a structured spreadsheet. On Mac, the Numbers output path is direct.

Open ignitai. Drag the PDF in. A single PDF or a folder of them. macOS 14.4+ on Apple Silicon (M1 onward) runs the on-device model — nothing leaves your Mac. Intel-era Macs fall back to a hosted pipeline with documented zero retention.
Describe what to extract, in plain English. For a typical sales report:

“Return one row per line item with date (ISO 8601), product_sku, product_name, quantity, unit_price, line_total, customer_name, and region. In a separate sheet, return the report metadata: report_date, total_revenue, total_units, and report_period.”

Save the prompt as a preset for next quarter.
Pick Numbers as the output format. ignitai writes a native .numbers file with two tables in one document. Currency formatting auto-applied to amount columns, ISO-8601 dates parsed as Numbers date cells, integer counts as integer cells.
Hit Extract. A typical 10-page sales PDF on an M2 Mac extracts in 5–8 seconds. A folder of fifty PDFs takes about three minutes.
Open the .numbers file. Numbers opens it directly. The two tables sit on one sheet by default; drag the second table to its own sheet if your workflow prefers that.

What you get out is a Numbers document, not a Numbers-flavored XLSX. Smart Categories work, sliders work, conditional formatting saves correctly, and iCloud round-trips to iPhone and iPad without rewriting cell types.

Method 2: Preview → copy → Numbers (the no-install path)

For one clean text-based PDF with one obvious table:

Open the PDF in Preview.
Hold Option to toggle rectangular selection. Drag a rectangle around the table.
Cmd-C to copy.
Open Numbers, place the cursor in cell A1, Cmd-V to paste.
If the result lands as a single column of concatenated rows, undo, and use Edit → Paste and Match Style instead. If that still mashes, the PDF lacks proper table structure and Method 1 is the right path.
Spot-check three things: the row count matches the source, the right-most column landed as numbers (Numbers has “Number” in the column-type indicator), and the date column landed as dates (not text strings that look like dates).

Works on PDFs with proper text + table structure. Fails on every scanned PDF and on most PDFs from older accounting software that draws tables as positioned text boxes rather than tagged tables.

Method 3: tabula-py + a script (the DIY path)

For a Mac-primary developer with a recurring single-source PDF (one supplier, one statement format, stable layout):

brew install tabula-java
pip install tabula-py pandas

Then a script that opens the PDF and writes a CSV that Numbers opens cleanly:

import tabula
tables = tabula.read_pdf("report.pdf", pages="all", lattice=True)
tables[0].to_csv("report.csv", index=False)

Open report.csv in Numbers. File → Save As → format .numbers to convert to a native document.

This is a perfectly fine path for one stable PDF source. It breaks the moment the source is a scan, the layout changes, or the PDF has multiple tables of different shapes you have to stitch back together. For a single recurring report, the script pays for itself once. For a varied batch, the per-PDF maintenance burden eats the savings.

Method 4: Web converters (and why most Mac users shouldn’t)

Smallpdf, iLovePDF, Adobe’s PDF-to-Excel — all work, all output XLSX, all want you to upload. The reasons not to use them for a real Numbers workflow:

The output is XLSX. You then open in Numbers and save as .numbers. The detour is unnecessary, and Numbers’ XLSX importer occasionally re-types columns in surprising ways.
Privacy surface. A bank statement, a salary file, or a supplier price list is exactly the kind of document you don’t want in a shared upload bucket — even with a documented retention policy, your data left the device.
Free-tier gating. One to three files free, then a monthly subscription. The recurring fee is the same magnitude as a one-time on-device app while keeping your data on the third-party server.
No Apple-aware output. None of them know what Numbers wants — Smart Categories, ISO date parsing, currency-symbol recognition. You get an XLSX, and the polish is yours to redo.

For a tutorial PDF with no real data, fine. For your actual Mac workflow, no.

What “preserved formatting” looks like in practice

A clean PDF-to-Numbers result has four properties the import path either gets right or doesn’t:

Column types parsed, not stringified. A 1,234.56 lands as the number 1234.56 in a number cell with thousands-grouping display. A 2026-04-17 lands as a date cell, not a 10-character string. A $45.20 lands as a currency cell with the $ as a unit, not as $45.20 as text. Numbers’ formula bar tells you which: a number cell shows the bare number; a stringified cell shows the literal text.
Currency unit recognized. $, €, £, ¥, kr — Numbers’ currency cell type carries the unit explicitly. An importer that lands $45.20 as text loses the unit; one that lands it as currency carries it through.
Dates as Numbers dates. Numbers stores dates as native date values, not as their text form. A date column lets you sort chronologically, group by month, and feed =YEAR() and =MONTH() formulas. A text column dressed up to look like dates breaks all three.
Multi-table layout preserved. A PDF with header metadata at the top and a line-item table below should land as two tables in the same Numbers sheet, side by side or stacked, with the layout that lets you point a formula in the metadata table at a SUM over the line-item table.

If your converter gets the table out but flattens types to text, you’re inheriting work — every column needs an explicit type-set in Numbers, and the formula you write referencing the date column won’t work until you fix the cells. Method 1’s prompt-driven extraction lets you specify types up front; the other methods leave them inferred.

A worked example: turning a quarterly bank statement into a Numbers ledger

Concretely: you have a 14-page Q1 bank statement PDF and want a Numbers ledger that splits transactions into income, recurring expenses, and one-off purchases for the quarter. The end-to-end flow on Mac:

Drag the statement into ignitai.
Prompt:

“Return one row per transaction with date (ISO 8601), description, amount (positive for credits, negative for debits), running_balance, and category (one of: income, recurring_expense, one_off, transfer). Infer category from description; mark recurring if the description matches another transaction within the same statement period.”
Pick Numbers output.
Open the .numbers file. Two tables: a transactions table and a summary table.
Add a Smart Category in Numbers grouped by category. Numbers shows quarterly totals per category natively, with subtotal rows you can collapse.
The same .numbers file syncs to your iPad over iCloud. You can review the summary on the iPad on the train without re-importing.

The whole ledger lives natively in Numbers. No XLSX detour, no losing the Smart Category on the next round-trip, no manually retyping the date column.

For the broader bank-statement workflow with CSV output, the Mac CSV bank-statement guide covers the QuickBooks / Xero import path. For the iPad-side review, the iPad bank-statement guide is the partner.

Round-tripping Numbers → iPhone and iPad

The Numbers file format syncs over iCloud Drive without any export/import friction. For a Mac-extracted Numbers file:

Save the document into iCloud Drive (any folder).
On iPhone or iPad, open Numbers; the file appears in the Recent list.
Edit on either device; the change syncs back to the Mac within seconds.

That’s the whole appeal of staying Numbers-native. The same .numbers file is the Mac desk view, the iPad couch view, and the iPhone on-the-go view, with no format conversion in the loop. Posts that target the mobile-side workflows — the iPhone receipt scanner guide and the iPad receipt review guide — converge on the same pattern: capture or extract on the right device, sync the Numbers file across all of them.

When Numbers is the wrong target

Honest cases where Numbers isn’t the right output:

Your accountant uses Excel. If the file is going to leave your Mac and land on a Windows desktop, save the final as XLSX (File → Export To → Excel). Use Numbers as the working format, XLSX as the handoff.
You need pivot tables Excel-style. Numbers’ Smart Categories cover most of what pivot tables do, but if you have an accountant building cross-tab pivots with calculated fields, Excel’s pivots are still ahead. Stay in XLSX.
You’re feeding data into a script or a database. Numbers .numbers files are bundles, not flat. Output CSV (or have ignitai write CSV directly) and skip Numbers as an intermediate.
Your data has many sheets that need to behave as a workbook. Numbers handles multi-sheet documents fine, but Excel’s cross-sheet INDIRECT references and named-range scoping are more deterministic. Heavy multi-sheet workbooks belong in Excel.

For Mac-primary single-document personal and small-business work, Numbers is the right native target. For team handoffs to Windows users, fall back to XLSX.

The two reconciliation checks per import

After importing into Numbers, two quick checks separate “I have a file” from “I have a working sheet”:

Column-type spot-check. Click each column header. The column-type indicator should match the data: Number for amounts, Date and Time for dates, Currency for monetary columns. Any column inferred as Text where it should be Number means a formula bar will treat it as a string and your SUM will return zero. Fix in Format → Data Format.
Row-count sanity. The bottom of the table should show a row count that matches what the PDF actually had. PDFs with multi-line cells (a description that wraps onto a second visual line) sometimes import as two rows; the inverse — two source rows merging into one — is rarer but worth catching. Compare against the PDF page count or the visible row count in the source.

Skip these two and the formula error surfaces an hour into the analysis, when you’ve already built three charts on top of stringified numbers.

Bottom line

For a Mac-primary user who wants a Numbers-native sheet from a PDF: install ignitai, drag the PDF in, write the prompt once, pick Numbers as output, hit Extract. For a single clean text PDF with no real data, Preview → copy → Numbers paste works in a pinch. For a recurring single-source PDF a developer can script, tabula-py plus a CSV save is fine. For everything else — scans, mixed templates, multiple tables, anything you’d rather not upload — the on-device path is the shortest distance from PDF on disk to a Numbers document that round-trips cleanly across your Apple devices.

The CSV-output equivalent for accounting handoffs is in the Mac CSV with AI guide; the Mac batch pattern across mixed document types is in the Mac batch guide; the Excel-on-Mac path for invoice workflows is in the Mac invoice guide. The app is the same; only the output format changes.

Get ignitai on the App Store — free download, $19.99/mo unlocks unlimited extractions after the 3-day trial.