Link terminology
A/B link testing
A/B link testing distributes qualifying visitors among two or more destination URLs using configured traffic weights.
Also called: split URL test, weighted link routing, control variant
Definition in plain language
A/B link testing distributes qualifying visitors among two or more destination URLs using configured traffic weights.
- 1Request
Define a control and at least one valid variant destination.
- 2Evaluate
Assign weights that total 100 percent.
- 3Decide
Choose sticky consistent assignment or per-click weighted behavior.
- 4Report
Compare allocation and downstream outcomes only after sufficient representative traffic.
Why a/b link testing matters
It lets teams compare landing-page alternatives behind one shareable link. The routing system can measure destination allocation and click behavior, but choosing a business winner normally requires conversion data from the destination.
The useful question is not merely whether a request happened. It is what kind of request it was, which rule handled it, and what conclusion the resulting data can support. Clear terminology prevents teams from treating every technical event as a person, every redirect as engagement, or every click as a conversion.
How it works step by step
A short-link system evaluates a request in a defined order. The exact sequence matters because an earlier status, access, or routing decision can prevent later rules from running.
For this concept, the practical workflow is:
- Define a control and at least one valid variant destination.
- Assign weights that total 100 percent.
- Choose sticky consistent assignment or per-click weighted behavior.
- Compare allocation and downstream outcomes only after sufficient representative traffic.
How to interpret the result
Read the result in the context of the configured link and the evidence available at request time. Browser signals, network location, cookies, referrers, and user agents are useful but imperfect.
Compare like with like: use the same date range, traffic filters, link ownership, and campaign definition. A change in classification or routing can change a report even when audience demand has not changed.
Deterministic example
The values below are example data, shown so every total can be checked from the visible rows.
Traffic allocation compared with Conversion measurement
How DrutoLink handles this
DrutoLink supports weighted destination variants and reports clicks and estimated unique visitors by variant. Consistent-hash mode keeps a recognized visitor on the same variant; weighted mode can choose independently.
Bots are not normally enrolled as human experiment traffic. Explicit geo or device destinations can bypass the default experiment. Native conversion and revenue attribution are not currently available.
Limitations to keep in mind
- Click distribution alone does not identify the better conversion outcome.
- Small samples can produce noisy percentages even with correct weights.
- Changing weights or destinations during a test changes the experiment being measured.
Frequently asked questions
Do A/B link tests measure conversions?
DrutoLink currently measures click and estimated visitor allocation, not native purchases, signups, or revenue.
Do weights need to total 100 percent?
Yes. A valid allocation needs a complete distribution across usable destinations.
Will the same visitor always see the same variant?
Consistent-hash mode aims for sticky assignment; per-click weighted behavior can vary.