Most text fields offer no formatting at all. A comment box, a game name, a spreadsheet cell, an SMS, a profile bio: no toolbar, no bold button, no way to strike a word through. There is still a way to do it, and it works because the line is part of the text rather than a property applied to it. That difference explains both why it works everywhere and why it occasionally causes trouble.
Rich text editors store formatting separately from the words. The word is in one place, the instruction to strike it is in another, and the editor combines them when it draws the page. Strip the formatting layer and the word survives unchanged.
Unicode strikethrough works the opposite way. After each of your letters, a combining mark is inserted: U+0336, COMBINING LONG STROKE OVERLAY. That mark has no width of its own. It draws through the character before it, and the two render as one visible unit.
So there is no formatting layer. The line is in the string, made of characters, travelling with the text wherever it goes.
That is why it survives a copy into a field with no formatting support. The field is not being asked to format anything. It is just storing characters, which is all it ever does.
Four marks cover the common cases and they stack.
Single strikethrough uses U+0336, the long stroke overlay. Double strikethrough adds U+0335, the short stroke overlay, on top of it, so each letter carries two marks.
Underline uses U+0332, COMBINING LOW LINE, which draws below the base character rather than through it. Double underline uses U+0333, COMBINING DOUBLE LOW LINE.
All four come from the Combining Diacritical Marks block at U+0300 to U+036F, the same block that puts an accent on a French vowel. Nothing unusual is happening. These are ordinary characters used for their ordinary purpose.
The strikethrough generator and the double underline generator each produce one of these directly.
This is the property that makes overlay styles more useful than they look.
A substituted alphabet replaces your letters entirely. Type an A and get a different character. A strikethrough style leaves your A as U+0041 and inserts a mark after it. The word underneath is still plain ASCII.
So a system that strips combining marks before comparing can still match the plain word. Search sometimes finds struck-through text where it would never find bold text. Copy it into a spreadsheet and a formula matching on the plain word may still work, depending on how the spreadsheet compares.
It also means the styling is cleanly reversible. Remove every combining mark and the exact original falls out, spacing and punctuation included. The unzalgo tool does that, and it works on strikethrough as well as on heavier effects.
One mark per non-space character means the cost scales with length, which is unusual on this site. Almost every other style adds a fixed amount or nothing at all.
Measured on a 48-character string, single strikethrough added 41 characters. Double strikethrough added 121, because the second mark is added to every character that already carries the first.
That is what breaks a bio. A Facebook bio caps at 101 characters and a single pass of double strikethrough on a short phrase will exceed it. An Instagram bio at 150 and a WhatsApp About at 139 are only slightly safer.
Where it costs nothing worth caring about is a message body. A WhatsApp message allows 65,536 characters. A Discord message allows 2,000. Strike as much as you like in either.
The character counter will tell you the real number before you commit.
Combining marks are supported far more widely than the decorative alphabets, because they are required for ordinary writing in dozens of languages. A system that cannot render a combining mark cannot render Vietnamese.
What varies is positioning. The mark is drawn relative to the base character by the font, and fonts differ in how well they place it. On a well-made font the stroke crosses the letter cleanly at the right height. On a poorly matched one it sits too high, too low, or slightly off to one side, so a struck word looks ragged rather than struck.
Monospace fonts tend to handle it worst, because the mark and the base are both fitted to a fixed cell and there is less room to adjust.
This is a cosmetic failure rather than a functional one. The word is still there and still legible.
Unicode defines a canonical ordering for combining marks, so that two strings with the same marks in a different order compare as equal.
Systems that normalise text apply that ordering, which can rearrange the marks on a character that carries several. For single strikethrough this is harmless, because there is only one mark. For double strikethrough, and for anything heavier, the stored result can differ from what you composed.
Some systems go further and strip combining marks from certain fields outright, usually name fields, because stacked marks break layout for everyone else on the page. That is deliberate validation rather than a bug.
The practical consequence: strikethrough is reliable in message bodies and less reliable in structured fields, which is the same conclusion the length arithmetic reached by a different route.
Better than with zalgo, worse than with plain text.
Your letters are ordinary characters, so a reader that ignores the combining marks announces the correct word. A reader that announces marks by name interrupts after every letter, which on a twenty-letter phrase is twenty interruptions.
There is no way to know which behaviour a given listener will get, and the range between them is wide.
The usual rule applies. If the struck text carries meaning that someone needs, it is worth reconsidering. If it is a joke, a correction shown for effect, or a crossed-out price, the meaning is mostly in the contrast and losing the line costs less.
In a document that supports real formatting, use the real formatting. It is announced correctly, it is searchable, and it costs no characters.
A comment box with no toolbar, where you want to show a correction rather than hide it.
A spreadsheet cell, where the alternative is applying cell formatting that will not survive an export to CSV.
A game or chat nickname, where there is no formatting at all and the strike is part of the look.
An SMS or a messaging app without its own markup, where a struck word is the only way to express a retraction visually.
A plain text file, where by definition no formatting layer exists.
In all of those, the alternative is not rich text. It is nothing. That is when carrying the line inside the string is worth the character cost.
Several places do have their own markup and it is always the better option when available.
WhatsApp renders a tilde on each side of a word as strikethrough, and it applies that to plain characters only, so wrapping already-styled text leaves the tildes visible. The WhatsApp page offers those wrappers as copy formats alongside the unicode styles, precisely because they are not interchangeable.
Discord uses double tildes in message bodies and it has the same restriction.
Word processors, email clients and documentation tools all have real strikethrough that costs nothing and is announced correctly.
So the decision tree is short. If the field has formatting, use it. If it does not, the combining mark is your only option, and it works.
Combining marks stack, so nothing stops you applying a strikethrough and an underline to the same text. The result is a letter carrying two marks, one through and one below.
It renders on most systems and it is worth knowing what it costs. Two marks per character doubles the overhead, so a 48-character phrase carrying both is roughly 90 characters longer than it started.
Ordering is the other catch. Unicode defines a canonical order for marks so that equivalent strings compare as equal, and a system that normalises will reorder them. That has no visual effect when one mark draws above the letter and one below, and it can have one when two marks compete for the same position.
The combination is fine in a message body and unsuitable anywhere with a limit, which is the same conclusion single strikethrough reaches by a smaller margin.
It inserts one combining mark per non-space character, so the cost grows with your text. On a 48-character string, single strikethrough added 41 characters and double added 121.
Sometimes. Your letters stay plain ASCII underneath, so a system that strips combining marks before comparing will match the plain word, unlike with a substituted alphabet.
The font places the combining mark relative to the base character, and fonts differ in how well they do it. Monospace fonts tend to be the worst at it.