> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tyrionurl.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Setting Up an A/B Test
> Split your link traffic between multiple destination URLs to find out which landing page converts best.
## What is A/B Testing in TyrionURL?
A/B testing (also called split testing) lets you direct a portion of your short link's traffic to two or more different destination URLs. TyrionURL tracks which variant each visitor was sent to, and — when conversion tracking is configured — which variant drove the most revenue or leads.
Instead of guessing which landing page performs better, you let real traffic decide.
***
## Feature Overview
| Capability | Details |
| ------------------------ | ------------------------------------------------------- |
| **Traffic splitting** | Configure percentage split across variants |
| **Multiple variants** | Support for 2+ destination URLs |
| **Conversion tracking** | Revenue and lead events tracked via API key integration |
| **Built on short links** | A/B tests are added to existing short links |
***
## When to Use A/B Testing
* Comparing two landing page designs for the same product
* Testing a promotional offer vs. the standard price page
* Evaluating different calls-to-action (e.g., "Buy Now" vs. "Get Started Free")
* Measuring performance of two product detail page formats
* Testing localized vs. global landing pages
***
## Setting Up an A/B Test
A/B tests are added to existing short links. Create a short link first (or open one from your dashboard) and then enable A/B testing from the link's settings.
→ See: [Creating Your First Short Link](/getting-started/first-link)
From the link's action menu, click **A/B Test** or open the link settings and toggle **Enable A/B Testing**.
The A/B Test modal opens. You'll see:
* **Variant A** — pre-populated with the link's original destination URL
* **Add Variant** — click to add Variant B, C, etc.
For each variant, enter:
* The destination URL
* The traffic percentage to allocate
Traffic percentages across all variants must sum to 100%. TyrionURL validates this before saving.
Enter an optional completion date for the test. After this date, the test stops splitting traffic.
TODO: Needs Product Verification — confirm whether traffic defaults to Variant A or stops entirely after completion date.
Click **Save**. The A/B test is now active. Every visitor to the short link will be routed to a variant according to the configured split.
***
## How Traffic is Split
TyrionURL uses server-side random assignment on each request to ensure statistically fair distribution. If you configure a 60/40 split:
* Approximately 60% of visitors go to Variant A
* Approximately 40% go to Variant B
Assignment is random per click — not per session or cookie. This means the same visitor could theoretically see different variants on repeated visits.
***
## Viewing A/B Test Results
After your test runs, view results in the link's analytics panel:
| Metric | Description |
| ---------------------- | -------------------------------------------------------------------- |
| **Clicks per variant** | How many clicks each destination received |
| **Conversion events** | Leads or revenue attributed to each variant (requires API key setup) |
| **Conversion rate** | Clicks → conversions ratio per variant |
→ Full details: [Tracking Conversions](/ab-testing/tracking-conversions)
***
## Best Practices
Run the test long enough to reach statistical significance — typically at least 100–200 clicks per variant minimum before drawing conclusions.
Change only one variable per test (destination URL). If you change both the page design and the offer simultaneously, you can't isolate what drove the difference.
Set a completion date to ensure you remember to review results and make a decision.
Set up [conversion tracking](/ab-testing/tracking-conversions) before launching so you capture revenue data from the start.
***
## Common Mistakes
**Ending the test too early** — A few days of data is rarely enough. Early trends often reverse. Wait for sufficient sample size before declaring a winner.
**Testing during atypical periods** — Avoid launching A/B tests during holidays, major sales events, or external disruptions that might skew behavior.
**Not setting up conversion tracking** — Click counts alone don't tell you which variant converted better. Set up the [API key integration](/ab-testing/api-keys) before starting.
***
## Related Pages
Generate keys to enable conversion tracking.
Capture revenue and lead events from partner sites.
View click and performance data.