Text Tools·4 min read

What Makes a Good Slug: Rules That Hold Up

Ask what makes a good slug and most answers jump straight to web addresses, but slugs live everywhere identifiers do: filenames, image assets, anchor links, database keys, CMS entry names, S3 buckets. A slug is text flattened into its safest possible form, letters, digits, and one separator, nothing else. Get the flattening rules right and the same string works in a URL path, a shell script, and a config key without quoting or escaping anywhere. This guide covers the mechanics: diacritics, symbols, whitespace, separators, and the failure modes that surface months later.

Open the tool All insights

What makes a good slug, wherever it ends up

Three properties, no exceptions.

Three properties cover it. Predictable: the same source text always produces the same slug, so different systems can regenerate it and agree. Portable: only unreserved characters, so no environment ever needs to escape it, a slug should paste cleanly into a terminal, an address bar, and a YAML value alike. Readable: a human glancing at creme-brulee-recipe knows what it names, which is more than a UUID offers.

Everything a slugifier does, transliteration, stripping, collapsing, lowercasing, serves one of those three properties. Once you see which rule serves which property, the settings stop feeling arbitrary.

Slug vs page title, two different jobs

Presentation changes. Identity doesn't.

The slug vs page title distinction trips up CMS users constantly. A title is presentation: full Unicode, punctuation, styled casing, freely editable next week. A slug is identity: constrained charset, stable, referenced by other systems. Titles can change without breaking anything. Change a slug and every stored reference to it, links, imports, scripts, image sources, has to be told.

The practical consequence: derive the slug from the title once, at creation time, then let the two drift apart. A post retitled for clarity keeps its original slug precisely because the slug's job is to not move.

Special characters in slugs, the actual mechanics

Transliterate, collapse, strip.

Special characters in slugs get handled three distinct ways, and knowing which is which explains every surprising output. Diacritics are transliterated: é becomes e, ü becomes u, so Crème survives as creme rather than vanishing. Whitespace is collapsed: any run of spaces or tabs becomes exactly one separator. Everything else, punctuation, quotes, emoji, symbols with no letter equivalent, is stripped outright in strict mode.

Stripping beats percent-encoding for identifiers. An encoded sequence like %C3%A9 is technically legal in a URL but unreadable, fragile across systems, and outright illegal in many filename contexts. A slug that needed encoding has already failed at its one job.

A slug generator example with genuinely messy input

Six words in, five hyphens out.

Input: Crème Brûlée: Chef's 10 Best Recipes! With the default hyphen separator and Lowercase ticked, the slug generator returns creme-brulee-chefs-10-best-recipes. Walk the transformations: two accented words transliterated, the colon and exclamation mark stripped, the apostrophe deleted so Chef's fuses into chefs, digits kept as-is, and six word boundaries turned into five hyphens.

Switch the separator to an underscore and the identical input yields creme_brulee_chefs_10_best_recipes, ready for a Python module name or a database key where hyphens would read as subtraction.

Mistakes that produce fragile slugs

Cheap now, expensive at migration time.

The slugs that cause trouble later share a few origins:

Renaming slugs casually. Every stored reference, bookmarks, hardcoded paths, import scripts, keeps pointing at the old one. Treat a slug rename like an API change.
Mixing separators across a project. A photo-final asset next to a photo_final_v2 asset guarantees somebody greps for the wrong one. Pick hyphen or underscore once and enforce it.
Keeping case significance. About-Us and about-us are different files on Linux and the same file on a default Mac volume, a mismatch that surfaces only at deploy time.
Trusting hand-typed slugs. Humans introduce doubled hyphens, stray periods, and trailing separators that a generator never would. If it wasn't machine-flattened, verify it.
Slugifying symbol-only input. A title made entirely of punctuation strips down to an empty string, and a blank identifier breaks whatever consumes it. Check the output isn't empty before saving.

Tips for slugs that survive renames and migrations

Generate once, persist forever.

Store the slug rather than recomputing it. Slugify libraries evolve their transliteration tables, and a slug regenerated two years later can differ by one character, which means one broken reference per consumer. Generate at creation, persist the value, and treat the stored copy as truth.

Keep slugs short but whole, dropping words is fine, cutting mid-word isn't. And when a date belongs in the identifier, write it sortable: a 2026-07-notes file lists in order everywhere, while july-notes-26 sorts nowhere.

Slug Generator alongside the other text tools

Flatten, convert, or clean.

Slugification overlaps with a few neighbors. Case Converter changes casing styles, camelCase to snake_case and friends, without stripping characters, so use it when the input is already identifier-safe. Find and Replace can retrofit an existing pile of names, swapping underscores for hyphens across a whole list in one pass. And Trim Whitespace pre-cleans pasted titles whose invisible padding would otherwise become stray separators.

There's also a Slug Generator in the SEO category aimed at page URLs specifically. This one is the general purpose flattener, built for every identifier your filesystem, codebase, and CMS will ever ask you to name.

Try it now

Open Slug Generator

The tool is one click away. No sign up, no upload, no payment.

Open Slug Generator

Write a review