Convert PDF Table to Excel and Keep Every Column

To convert a PDF table to Excel with every column intact, first find the page range that holds only that table, then convert just those pages. Afterward, check three things: that a table running over several pages joined into one block, that stacked headers became a single row, and that no column went missing.

Finding the page range for one PDF table

Converting one table out of a long PDF starts with narrowing the input. Work out the first and last page of the table, then convert only those pages. A converter that sees all two hundred pages will return every table, footnote and chart caption it can find, and you will spend the time deleting.

Read the page numbers the way the PDF counts them

A PDF has its own page count, starting at 1 on the first sheet. The number printed in the corner of the page is often different, because a report may use roman numerals for the front matter or restart numbering after a cover. Always use the position in the viewer, the "page 14 of 212" figure, when you tell a tool which pages to read. Printed numbers lead to off-by-ten mistakes that look like a broken converter.

Find both ends of the table

Scroll to the page where the table's header row first appears. Then scroll forward until the last row ends, which is usually a total line, a blank gap or a new heading. Write down both numbers. If a table continues across a page break, the header row often repeats at the top of the next page, so a repeated header is a good sign you are still inside the same table.

Cut the pages out before converting

Two routes work for most people:

1. Print to PDF from your viewer and choose a page range. The new file holds only those pages, and any converter will treat it as a small document.
2. Use the page-range field of the converter or of Excel's own PDF import, if it has one.

Cutting the file first is the safer habit. It also keeps the other thirty tables in the report from landing in your workbook, where they cost you a sheet each and a decision about which one you wanted.

A quick test that you have the right range

Before converting, read the first row and the last row of your range aloud against the PDF. If the first row of your sheet is a column heading and the last row is the total you expected, the range is right. If you find a stray heading from the next section at the bottom, trim the range by a page.

FastBankConvert is built around bank statements rather than general reports, and it is still getting ready to open. If your table is a transaction list from a statement, you can join the list on the get-started page. For a table inside any other kind of document, the steps in this post still apply to whatever converter you pick.

Tables that span several PDF pages

A table that spans several PDF pages is one table to you and several tables to a converter. Each page is read on its own, so the output often arrives as a block per page, separated by blank rows, repeated headers or page furniture. Your job is to join those blocks into one continuous range.

What a split table looks like in the output

Look for these signs after conversion:

• The header row appears again partway down the sheet, once per page.
• Blank rows sit between blocks at the page breaks.
• A page number or running title ("Annual Report 2025 | 14") lands in the first column.
• A row that straddled a page break is cut in two, with half the values on each side.

The last one is the hardest to see. A row broken by a page boundary has a label in one block and the numbers at the top of the next, so a quick scroll shows nothing wrong until a total fails to match.

Join the blocks in a fixed order

The order below keeps you from deleting something you need.

1. Delete every repeated header row except the first. Sort or filter on the first column for the header text and remove the matches.
2. Delete the page furniture rows: page numbers, running titles, "continued" notes.
3. Delete fully blank rows, using a filter on a column that always has a value.
4. Repair any row split across a break by moving the values up to the label row.
5. Confirm the row count against the PDF.

Check the row count against the PDF

Count the rows in the source by reading the first and last page and any line numbers the table prints. Many tables number their rows, or carry a count in the title. If the PDF says 214 entries and your sheet holds 212 data rows, two rows were lost or merged. Find them before you analyse anything, because the missing ones are rarely a random sample. They tend to be the rows at the top or bottom of a page.

If your table is a bank statement, the same idea works with money in place of a count: the opening balance plus the movements has to land on the closing balance. The post on what survives when a PDF becomes a spreadsheet lists the other things that change in conversion.

Do two-row PDF table headers survive conversion?

Two-row headers do not survive conversion intact. A header with a group label above several column labels, such as "2025" over "Q1", "Q2" and "Q3", is usually read as two separate rows, and the group label lands in only the first of the columns it covered. The fix is to flatten the two rows into one.

Why the group label gets lost

In the PDF, the group label is drawn once, centred over its columns. Converters assign text to columns by position, so the label belongs to whichever column its left edge sits over. The other columns in the group get an empty cell. In the sheet you see "2025" in one cell, three blanks to its right, and the quarter labels on the row below.

Filters, pivot tables, lookups and Power Query all want one header row with a unique name per column. Two rows break each of them.

A worked example: three pages, two header rows

This example uses made-up figures. A report prints a table over three pages. The header has two rows and repeats on each page.

Converted output on page 1 (page 2 and page 3 repeat the same two header rows):

The stacked header says two things at once: the year over Q1 and Q2 and the year over the 2026 column. The empty cell over Q2 means "2025 still applies". The Region label sits on the upper row while Product sits on the lower row, though both are plain column names.

The single flattened header row it should become:

To get there:

1. Fill the group label across its columns: copy "2025" into the empty cell above Q2.
2. Join the two rows per column into one label, with a space between: =TRIM(C1&" "&C2).
3. For columns whose label sits on only one row (Region, Product), the join leaves just that label.
4. Paste the results as values over the first header row, then delete the second header row and the two repeated copies from pages 2 and 3.

After this, every column has a unique name and the table can be sorted, filtered and loaded into a pivot table.

When the header text is wrapped

A long column label such as "Net amount after discount" may be printed on three lines in a narrow column. Converters sometimes return each printed line as its own row, so one header becomes three. Treat it the same way: join the pieces for each column into one label before the first data row.

Did every PDF table column come across?

A missing column is easy to miss, because a sheet with fewer columns still looks like a table. Check by comparing the column count in the PDF with the count in the sheet, then by comparing one full row of values on both sides. Columns go missing in three ways: merged, shifted or dropped.

Merged columns

Two narrow columns close together can be read as one when the gap between them is smaller than the converter expects. You see "Widgets 1,200" in one cell, or two numbers separated by a space. Text to Columns splits these when the space between the two values is the same on every row.

Shifted columns

If one page has an extra or missing cell, the values on that page sit one column to the left or right of the rest. The header says Q2 and the cell holds a product name. Scan each page block, and look at the first data row after each repeated header.

Dropped columns

A column on the far right, outside the page margin as the converter measured it, may not come across at all. This happens with wide tables printed in landscape inside a portrait page. Compare the last header in the PDF with the last header in the sheet.

A short check you can run in two minutes

1. Count header cells in the PDF and in the sheet.
2. Pick the row in the middle of the second page. Read it across in the PDF, then across in the sheet.
3. Sum one numeric column in Excel and compare it with the printed total.
4. Look for any column where text and numbers are mixed. That is the sign of a shift.

If your table prints a total row, the total check is the strongest of the four, since a lost column or a lost row will both change it.

Footnote markers stuck to PDF table numbers

A footnote marker stuck to a number turns a figure into text. The value "1,200*" or "1,200 (a)" is not a number to Excel, so sums skip it and sorting puts it in the wrong place. Remove the marker and convert the cell back to a number.

Why it happens

Reports mark special figures with an asterisk, a dagger or a superscript letter or digit. In the PDF the marker is a separate character placed close to the number. In the extracted text it is part of the same word, because nothing separates them. A superscript 2 is the worst case: "1,2002" can read as a larger number, not as 1,200 with a note.

Find the affected cells

Filter the column by "Text filters" and then by "contains" for the common markers: *, †, (a), (1). A faster way to find all of them at once is a helper column that tests whether the cell is a real number: =ISNUMBER(C2). Every FALSE in a column that should be numbers is a candidate.

Remove the marker

For an asterisk, Find and Replace with * needs care, because Excel treats a lone asterisk as a wildcard that matches everything. Type ~* to find a literal asterisk. Replace with nothing, then convert the column to numbers. For lettered notes, a formula can strip a trailing note: =VALUE(LEFT(C2,FIND(" (",C2&" (")-1)) returns the number before a space and an opening bracket.

Keep the notes. Before you strip them, copy the marker and its row into a notes column, since the footnote may change how you read the figure ("* provisional"). Deleting the marker without recording it loses information the report gave you.

A superscript digit needs the PDF

When the marker is a superscript digit, you cannot tell from the sheet alone where the number ends. Open the PDF at that row, read the printed value, and compare. If the note digit joined the number, the cell will be about ten times larger than its neighbours, which a quick sort on the column makes obvious.

After the table is clean

Once the table has one header row, one block and numeric cells, you can load it with a repeatable query instead of repeating the cleanup each month. The guide to importing a PDF into Excel with Power Query covers that route.

If you would rather understand why the converter split your table the way it did, how table detection finds rows and columns in a PDF explains it, and ruled and borderless layouts behave differently, as the section on ruled tables versus whitespace tables shows. For a table short enough to move by hand, copying a PDF table into Excel may be quicker than any conversion.