Missing letters after PDF to Excel usually trace back to the fonts in the PDF. Many fonts draw letter pairs such as fi, fl and ff as one joined shape, a ligature, and if the file does not say which letters that shape stands for, the converter drops it. The fix is a Find and Replace for each broken pair.
Why do letters go missing after PDF to Excel?
Letters go missing because a PDF stores shapes, called glyphs, and a converter has to work out which characters each glyph means. Most fonts carry a map from glyph to character. When a glyph has no entry in that map, the converter has no letter to write, so it writes nothing or a placeholder.
On screen the PDF looks perfect, because drawing a glyph needs no character at all. The problem only shows when software tries to read the text: copying, searching, or converting to Excel. That is why a statement can read "Office supplies" in your PDF viewer and "Oce supplies" in the spreadsheet.
Three patterns cover most cases:
- The same pairs vanish every time. "fi", "fl", "ff", "ffi" and "ffl" disappear while other letters survive. The cause is ligature glyphs.
- Whole words turn into the wrong characters. A merchant name comes out as symbols or shifted letters. The cause is a font with a custom or symbol encoding.
- Letters vanish at random. Single characters drop from different words with no pattern. On a scanned page, the cause is usually optical character recognition, not fonts.
Each pattern has a different repair, so identify which one you have before changing anything.
Ligature glyphs and the fi, fl and ff letter pairs
A ligature is a single glyph that joins two or three letters, such as the f and i in "fi", so the dot of the i does not collide with the hook of the f. Typesetting software substitutes them automatically. When the PDF's text map for those glyphs is missing or wrong, the converter loses exactly those letters.
Unicode, the standard that defines how characters are numbered, does include a few ligature characters, but only for compatibility. The Unicode Consortium's FAQ on ligatures and presentation forms says ordinary typographic ligatures should be represented by letter sequences and handled by fonts, and that use of the encoded ligature characters is discouraged. In other words, the text behind a joined "fi" shape should be a plain f followed by a plain i.
A converter can therefore meet three situations:
| What the PDF holds | What lands in Excel | Example (made up) |
|---|---|---|
| Glyph mapped to the letters f and i | Correct text | "Office supplies" |
| Glyph mapped to the single ligature character | Looks right, but does not match a search for "fi" | "Office supplies" |
| Glyph with no mapping | Letters missing, or a box or question mark | "Oce supplies" or "O?ce supplies" |
The middle row is the sneaky one. The word looks correct in the cell, but a lookup, a filter for "Office" or a categorization rule will not match it, because the cell holds one special character where your search has two or three letters.
Words that ligatures hit on statements
Financial PDFs are full of the affected pairs. "Office", "official", "certificate", "financial", "benefit", "profit", "traffic" and "coffee" all contain one. A merchant list with these words missing letters is a strong sign of ligatures rather than a damaged file.
Symbol fonts and custom encodings that swap letters
Some PDFs embed a font subset with its own private numbering, so glyph 1 is whatever letter the document used first, glyph 2 the next, and so on. Without a map back to real characters, a converter that trusts the numbering writes the wrong letters. The word keeps its length but turns into nonsense.
This differs from the ligature case in two ways. The damage covers whole words or whole fields, not just two-letter pairs. And the substitution is consistent within one file: if "a" became "D" once, it became "D" everywhere in that font.
Symbol and barcode fonts cause a related problem. A logo, a check mark or a currency sign drawn with a symbol font can come across as a letter, such as a stray letter where the PDF shows a tick. These stray characters usually sit in their own column or at the start of a description.
Repairing a fully re-encoded font with formulas is possible, because the swap is consistent, but it is slow. Build a two-column table of wrong and right characters from a few known words, then apply it. If more than a handful of words are affected, getting a better source file is faster, as the last section explains.
Finding words with missing letters across a converted sheet
Damaged words are easiest to find by searching for the gaps they leave. A word that lost "fi" often leaves a telltale fragment, such as "Oce", "bene t" or "certi cate". Searching for known fragments, special characters and unexpected spaces finds most of them in a few minutes.
Useful searches, using Find (Ctrl+F) with Look in set to Values and Within set to Workbook:
- Special ligature characters. Search for "fi", "fl", "ff", "ffi" and "ffl" pasted from this page. Each hit is a word that looks fine but will not match a lookup.
- Placeholder marks. Search for "?" and for the box character the converter used. Tick Match entire cell contents off.
- Odd spaces inside words. Some converters replace a missing ligature with a space. Search for common fragments such as " cate", "ne t" or "O ce".
To count damaged cells in a column without searching one by one, use a helper formula. This one flags a cell in B2 that holds any of the five ligature characters:
=SUMPRODUCT(--ISNUMBER(FIND({"fi","fl","ff","ffi","ffl"},B2)))>0
Fill it down and filter for TRUE.
Searching every page sheet at once
If the converter put each PDF page on its own sheet, a fix on one sheet leaves the others broken. Setting Within to Workbook in Find and Replace covers every sheet. Combining the pages first is often cleaner; putting all PDF pages onto one Excel sheet shows how, and then one pass fixes the lot.
Repairing missing letters with SUBSTITUTE and Find and Replace
Repair the special ligature characters first, because they are certain: each one stands for known letters. Then fix the missing-letter fragments, which need judgment. Find and Replace handles both on the sheet itself, and SUBSTITUTE does the same in a helper column when you want to keep the original text.
For the special characters, run five Replace All passes, one per character:
| Find | Replace with |
|---|---|
| fi | fi |
| fl | fl |
| ff | ff |
| ffi | ffi |
| ffl | ffl |
As a single formula in a helper column, with the description in B2:
=SUBSTITUTE(SUBSTITUTE(SUBSTITUTE(SUBSTITUTE(SUBSTITUTE(B2,"ffi","ffi"),"ffl","ffl"),"fi","fi"),"fl","fl"),"ff","ff")
The three-letter forms go first so that "ffi" is not half-replaced by a two-letter rule.
Fixing fragments where letters were dropped
Dropped letters cannot be found by character, so work from a list of words. Sort the description column, scan it for broken words, and write each broken form and its fix in a small two-column table. Then replace them one by one. Use whole-word fragments, such as "Oce " with a trailing space, to avoid changing other words that happen to contain the same letters.
Be careful with short fragments. Replacing "bene t" with "benefit" is safe. Replacing "ce" with "fice" would wreck every word containing "ce". When in doubt, replace only with Match entire cell contents ticked, one full description at a time.
Inventory and product lists
Product and stock reports suffer most, because item names repeat hundreds of times and lookups depend on exact text. If you are converting one, moving an inventory report from PDF to Excel covers keeping product codes and names matchable.
Is the PDF or the converter to blame for missing letters?
A quick test shows where the letters are lost. Open the PDF in a viewer, copy a word with a missing pair and paste it into a plain text editor. If the pasted word is also missing letters, the PDF's text map is the problem. If it pastes correctly, the converter is losing letters the file does contain.
What each result means:
- Copy loses letters too. Every tool that reads the text will hit the same gap. A different converter will not help unless it falls back to optical character recognition for that text. Repair in Excel, or get a better source file.
- Copy works, conversion does not. The converter is reading the file differently from your viewer. Try another tool, or another export setting such as a different output format.
- Search in the viewer fails for the word. If Ctrl+F in the PDF viewer cannot find "office" on a page that shows it, the text layer is damaged in the file itself.
A file that cannot be read cleanly may cause other trouble later. If the converted workbook will not open at all, fixing a converted Excel file that is not opening covers that separate problem. A file that grew unusually large after conversion is another sign of odd font data; a converted Excel file that is too large explains where the extra size comes from.
When missing letters mean converted text cannot be trusted
Stop repairing when the damage is no longer limited to known letter pairs. If amounts, dates or account numbers show missing or swapped characters, the converted text is not safe to use, because a dropped digit changes money without any visible error. Words can be guessed back. Numbers cannot.
Signs it is time to change the source rather than fix the sheet:
- Digits are missing or swapped in any amount, date or reference column.
- More than a few percent of descriptions are damaged, so a word list would take longer than retyping.
- The same word is broken in different ways on different pages, which means the fonts differ per page.
- The running balance does not add up after conversion.
In those cases, download the statement again from the bank's website, since a fresh copy may use a better-embedded font. Failing that, convert a printed and rescanned copy so the text is read from the image instead of the broken font map. Recognition brings its own digit errors, so check every amount against the PDF.
FastBankConvert reads the layout of the statement and, if the balances do not add up after conversion, tells you before you download, which is a useful sign that digits were misread. It is getting ready to open; Get started puts you on the list.