Short links rot — and roughly twice as fast as normal links do. We followed 134,515 links that an archive had caught working sometime between 2015 and 2024, and by 2026, 28.5% of the shortened ones were dead, versus 13.5% of the plain links posted right beside them. About 6 of those points are the shortener's own fault — deleted codes, dead services, lapsed domains — and that's the part a link on a domain you own never has.
Link rot is the quiet way the internet forgets itself: a URL that worked yesterday just… doesn't. We wanted to know something specific and slightly petty — do short links rot faster than normal ones? So we rounded up those 134,515 links, every one of them caught working by an archive on a known date, and followed them all again this month. Grab a coffee; the numbers are genuinely wild.
Do short links expire? Here's the clean comparison
The lazy version of this study says "62% of 2019 short links are dead!" and gets torn apart instantly, because so are loads of 2019 links — that's just time. To measure the shortener's effect you need a fair fight. So we compared each short link to the plain URLs posted in the very same Hacker News comment — same author, same topic, same day — differing only in whether someone ran it through a shortener.
A shortened link was 2.11× as likely to be dead as the plain URL sitting right next to it.
And here's the kicker: those control links skew toward well-kept technical docs, which should make shorteners look better than they'd do against the average link. The gap held up anyway.
Why do short links die? Two very different reasons
This is where we resist the clickbait. When a short link fails, it fails in one of two places, and only one of them is the shortener's fault. Either the destination is gone (ordinary rot, nobody could stop it) or the shortener layer itself died — the code got deleted, the whole service shut down, the domain lapsed. We split every link:
So most link rot would've happened regardless — a shortener can't resurrect a deleted page. But that extra 6.2 points only exists because a middleman sat in the redirect, and it vanishes the moment the redirect runs on infrastructure you own. Whenever you read "own your domain," this little bar is the entire reason. It's also why links you make on in.bio — on a domain you control, that never expires — dodge that whole category.
Which URL shorteners are most reliable?
Not all shorteners rot equally — and the pattern is almost embarrassingly logical. Sort them by who stands behind the domain and the survival rates line up like dominoes:
It's all about who's paying the bill. youtu.be is underwritten by Google. A bit.ly link is underwritten by a company that can reprice, get bought, or change its terms. tr.im was underwritten by one person's hobby budget — and every single tr.im link is now gone. We give the full body-count in the shortener graveyard.
How fast do short links rot as they age?
Comparing years is a statistical trap — the mix of domains changes annually, so a raw "2016 vs 2019" table measures your sample, not time. To do it honestly you hold the domains fixed. Here are the only two with a solid sample in all ten years, and they could not behave more differently:
A bit.ly link is roughly seven times likelier to be dead at eight years old than at three, climbing toward a ~40% plateau — links that survive their first few years tend to keep surviving. youtu.be, meanwhile, sat at a flat 0.0% every single year. (Fair asterisk: we confirmed those resolve, not that every video still exists.)
The weird one: 8.6% of working links secretly moved
Here's the finding we're proudest of, because as far as we can tell nobody's measured it. Pew Research and Harvard's Perma project studied whether links die. But our archived receipts also store where each code originally pointed — so we could ask a spookier question: do links that still work still go where they used to? For 8.6% of them, the answer is no.
A dead link is honest — it fails, you move on. A repointed one keeps quietly collecting clicks from everyone who ever trusted it.
Most are innocent corporate moves (allfacebook.com → adweek.com), but "innocent rebrand" and "someone bought the expired destination" look identical from the outside — and whoever clicks can't tell which they got. We dig into it in the full write-up. In the meantime, our link expander shows where any short link really lands before you click, and the link checker flags a dodgy destination.
How to make short links that don't rot
You can't do much about the 22% — if a destination page gets deleted, no shortener saves it. But the shortener's own 6 points? Entirely avoidable. Four habits:
- Use a domain you own. A link on
go.yourbrand.comlives as long as your domain does — not as long as a third party's business plan. - Pick a shortener that doesn't expire your links. This is the whole pitch: on in.bio, a link you create with a free account stays live indefinitely — it won't get pruned, sold off, or shut down under you.
- Keep the destination alive. Six points are the shortener's fault; twenty-two are the page's. Check yours now and then.
- Don't chain shorteners. Every extra hop is one more service that can fail.
Want links that never join this graveyard? Create them on in.bio — our links never expire. Every short link stays live for good, and with a free account it lives on a domain you control (not one a company can shut down), you can re-point it any time, and you get real, bot-filtered click analytics on every one. If you're comparing options, we lined them up in the best URL shorteners, and there's more on redirect choices in link cloaking.
How we actually measured this (the fun part)
Here's the trick that makes these numbers trustworthy. We didn't grab random links and check if they work — of course lots of old links are broken. Instead, every link in the study is one a public archive caught in the act of working: a crawler logged a real 301/302 redirect from it on a specific day. So each link comes with a receipt that says "this worked on this date." We just went back and knocked on the door again.
We pulled those receipts from three independent archives — the Wayback Machine's CDX index, the Common Crawl URL index, and the Hacker News API — then followed every link ourselves, no browser, logging each hop. A link only counts as dead if the shortener is gone or the destination hard-fails. Timeouts and rate limits? Thrown out as "don't know," never counted as dead. Every headline number carries its sample size (n) and 95% confidence interval.
The honest fine print: our sample leans toward domains with enough archived links to measure (bit.ly, goo.gl, youtu.be and friends), tech and documentation URLs are over-represented, and we checked whether a link resolves — not whether the page behind a working link still has its stuff. We say this out loud because if we don't, a reviewer will say it for us.
Steal this study (please)
Download the data: link-rot-study-2026.csv — the aggregate tables behind every chart here (dead rates, sample sizes, 95% confidence intervals), free to re-analyze under CC BY 4.0. Everything is free to cite with a link back. Suggested citation:
InBio (2026). We followed 134,515 short links across 9 years. 134,515 archived links, cohorts 2015–2024, measured Aug 2026. https://in.bio/blog/link-rot-studyWe re-run it every year — the trend line is the real asset, and the numbers only get more damning as the cohorts age.