Your CMS committed two different crimes this week, and both are visible in the address bar. First it took your careful headline, "What We Learned From Testing 47 Coffee Grinders Over Six Months (and What We Would Do Differently)," and squeezed out a ninety-four-character monument to itself: what-we-learned-from-testing-47-coffee-grinders-over-six-months-and-what-we-would-do-differently. Then you noticed that your colleague's most-visited post, the one that still earns traffic every week, lives at post-17.
One slug says too much. The other says nothing at all. Both are worth fixing, because the slug is the one part of your page that travels without you: into search results, into group chats, into plain-text emails, into the browser history of a reader hunting for your article at midnight. Small text, outsized exposure.
The slug is the part of the URL you control
Break a URL apart and most of it was decided before you wrote a word. The protocol, the domain, maybe a folder like /blog/ or /products/: fixed, someone else's call. The slug is the stretch at the end, and it is the one piece a writer actually chooses. cast-iron-skillet-12-inch says what the page is. ?p=1183&cat=07 says the database was in a hurry.
Slugs also appear in more places than the address bar. Search results print them under the title. Messaging apps preview them when someone pastes a bare link. Screen readers speak them aloud, hyphen by hyphen, to a blind visitor deciding whether to click. The standard is stricter than it sounds: a good slug reads clearly out of context, to a stranger, with no headline attached.
Keywords in slugs: real, and small
Since you are wondering about rankings, here is the honest version. Google's engineers have said for years that words in a URL count as a ranking signal, and also that the signal is small. If two pages were otherwise neck and neck, a matching slug might tip the balance, but pages are never otherwise identical, and the URL is never doing the heavy lifting.
The click is where the slug earns its keep. When your result appears in a list of ten, the searcher scans titles first and slugs second, and a slug containing their own words quietly confirms the promise before they commit. The slug is a second headline for scanners, and scanners are the readers who decide whether blog posts get finished at all. Random parameters raise a small doubt that a competitor's clean link does not.
The rules worth remembering
Hyphens, not underscores. Google has said plainly that hyphens are treated as word separators in URLs, and underscores are not treated the same way. To a crawler, best-coffee-grinders is three words; best_coffee_grinders can read as one glued-together token. Old underscore links still work, and fixing them is rarely worth the trouble. For anything new: hyphens.
Lowercase, always. Many servers treat /About and /about as two different addresses, which means one capitalized link creates a duplicate of your own page and splits its signals between the copies. Even where the server is forgiving, the web settled on lowercase long ago, and consistency is what makes a site audit readable.
Three to six words. Long slugs get truncated in search results and cut off in chat windows, and what survives is the beginning, so the meaningful words belong up front. Short slugs are also easier to read aloud, easier to type, and easier to remember an hour later.
| Slug | Verdict | What gives it away |
|---|---|---|
post-17 | Fails | No keyword, no context, no reason to click |
what-we-learned-from-testing-47-coffee-grinders-over-six-months | Too much | Stopwords bloat it; the tail will truncate |
47-coffee-grinders-six-months | Better, still loud | The number up front reads like a listicle |
coffee-grinder-test-results | Right | Clear, keyword-first, honest about the page |
best_coffee_grinders | Wrong separator | Underscores do not split words for crawlers |
Coffee-Grinder-Results | Wrong case | Risks duplicate addresses; breaks convention |
Stopwords: to cut or not to cut
Older SEO advice says strip every "a," "the," and "of," because search engines ignore them anyway. That is true of the crawler and beside the point. The slug's most important reader is the human scanning a chat log or a results page at speed. Apply the same test you would in plain language writing: if cutting the small words keeps the meaning, cut them; if the slug turns into a word jumble, the characters you saved were not worth the confusion. "The Art of the Cold Brew" dies gracefully as cold-brew-guide. It dies ugly as art-cold-brew.
When to change a slug, and when to leave it alone
Changing a published slug with no redirect is vandalism against your own rankings. Every external link, bookmark, and old tweet now points at a 404, and Search Console fills with errors. With a permanent 301 redirect in place, most of the page's standing follows it to the new address. Most, not all: a redirect is a small toll paid at every hop, which is why the destination has to be worth it.
Worth paying: the slug is actively wrong or embarrassing (post-17, misspellings, a date that makes evergreen content look expired), the page has pivoted topics, or you are migrating the whole site and building a redirect map anyway. Not worth paying: cosmetics. how-to-season-cast-iron-skillet will not rank differently as season-cast-iron-skillet, and any change can wobble while crawlers catch up. Update your internal links to the final address while you are there. A redirect is a detour, not a home.
A five-minute slug habit
The whole discipline fits in one paragraph. Write the headline first, because a slug should compress a headline, never replace one. Reduce it to three to six words, meaningful words first. Lowercase, hyphens. Search your own site for near-duplicates before publishing, because coffee-tips-2 is just the post-17 problem rebuilt by hand. Then treat the slug as load-bearing: after publication it changes only with a redirect and a reason.
For fixing a backlog, let tools carry the boring half. A slug generator turns a column of headlines into candidate slugs in bulk. A case converter rescues a legacy folder of TitleCase URLs in one pass. And find and replace is how you repoint the internal links afterward without opening every page by hand.
Frequently Asked Questions
Do keywords in the slug actually help SEO?
A little, and less than most advice implies. Google has described URL keywords as a minor signal for years. The larger benefit is human: a slug containing the searcher's own words helps the click, and the click helps everything downstream.
How many words should a slug have?
Three to six for articles and product pages. That is long enough to be descriptive and short enough to survive truncation in search results and chat previews. Put the meaningful words first.
Should the slug match the headline exactly?
Almost never. Headlines are written for rhythm; slugs are written for recognition. "What We Learned From Testing 47 Coffee Grinders Over Six Months" loses nothing as coffee-grinder-test-results.
Are underscores ever acceptable in URLs?
They work fine. Links with underscores load, and nothing breaks. But Google does not treat them as word separators the way it treats hyphens, and hyphens cost nothing, so use them for anything you create today and leave the old ones alone.
Is it worth changing an old, ugly slug?
If the page matters and the slug is meaningless, misspelled, or dated, yes, with a 301 redirect. If you only want to shave a stopword, no. The toll is real and the gain usually is not.
Say it out loud
Here is the test that never fails. Recommend the page aloud, the way you would to a friend: "textoria dot com, slash coffee grinder test." If you would be embarrassed to say it, do not publish it. If it fits in a sentence, it fits in a URL. Compress your next slug by hand, before the CMS turns it into a ninety-four-character monument.