How do I A/B test a short link?
Point one short link at two or more pages and split visitors between them to see which performs better. How A/B testing a link works and how to set it up.
Shortlen Editorial · · 3 min read
The short answer
Put an A/B test on the link: one short link points at two or more destinations, and visitors are split between them — an even split or weights you choose. Returning visitors keep seeing the same version, bots aren't counted, and you get real statistics. When one version wins, you send everyone to it. A/B tests are part of Shortlen's paid plans.
Key points
- An A/B test lets one short link send different visitors to different pages, so you can compare them with the same link in the same campaign.
- You pick an even 50/50 split or set weights; a returning visitor keeps seeing the same version so the test stays fair.
- Bots and link previews are served but not counted, so the result reflects real people.
- You get proper statistics — lift, a p-value and confidence — and can track conversions, not just clicks.
- When there's a winner you send all traffic to it in one step; A/B tests are on the paid plans.
On this page
If you want to know which of two landing pages, offers or headlines works better, you don’t need two separate links and a spreadsheet. An A/B test lets a single short link do the comparison for you, splitting its visitors between the versions and keeping score.
When an A/B test is the right tool
Reach for one when you have the same audience and want to compare two versions of a destination: a new landing page against the current one, two offers, or two versions of a page with different copy. Because it’s one short link, you can drop it into a campaign exactly as you normally would and let the test run underneath.
If instead you want to send different people to different places on purpose — by device or country — that’s a smart link, not an A/B test.
How it works
- The split. Point the link at two or more destinations and choose an even 50/50 split or set weights (say 70/30) when you want to send more traffic to the safer option.
- Sticky assignment. A returning visitor is kept on the version they first saw, so the same person doesn’t flip between variants and blur the result — and nothing is stored on their device to do it.
- Real people only. Bots and link-preview scanners are served but not counted, so the numbers reflect actual visitors.
- Real statistics. You get lift, a p-value and a confidence level, plus an estimate of how many more visitors you need — so you know when a result is solid rather than noise.
- Conversions, not just clicks. You can record a conversion (a sale or sign-up) against the test from your site or server, and compare variants on what happened after the click.
How to set one up
- Open the link and add an A/B test. Give it a name and add your variants, starting from the page you’re comparing against.
- Choose the split. Even, or weighted toward one variant.
- Let it run. Share the one short link as usual; the test collects data in the background.
- Read the result and finish. When there’s a clear winner, send all traffic to it in one step — or end the test and keep the original.
A couple of honest notes
- Give it enough traffic. A test needs a reasonable number of visitors per variant before the statistics mean anything; ending too early can crown the wrong winner.
- Change one thing at a time. If two variants differ in several ways, a win won’t tell you which change caused it.
A/B tests are part of the paid plans, alongside the other routing tools — see the features page, and pricing for which plan includes them.
Questions people ask
What's the difference between an A/B test and a smart link?
A smart link sends visitors to different places on purpose, by a rule you set — iPhone users here, visitors in Germany there. An A/B test sends comparable visitors to different versions at random to find out which one performs better. They solve different problems and can be used separately.
Will the same person see different versions each time?
No. Assignment is sticky: a returning visitor is kept on the version they first saw, without anything being stored on their device, so your comparison isn't muddied by people bouncing between variants.
Can I measure more than clicks?
Yes. You can record conversions — a sale or a sign-up — against the test from your own site or server, so you compare variants on what actually happened after the click, not just how many clicked.
Which plan includes A/B tests?
A/B tests are part of Shortlen's paid plans, alongside smart links. The pricing page shows which plan has them.