A/B Testing For Your Ad Stacks to Improve Demand-Path optimisation
Experiments allow you to set up a head-to-head against two or more tag configs to optimise your revenue
Activate Advanced Experiment Technology
Fusion experiment technology uses 'configs'. These configs can be added to an 'Experiment' and each config gets a percentage of traffic to A/B test and achieve optimal revenue performance.
Configs contains configurable attributes such as;
- Ad Unit
- Ad Stack (SSPs & Adserver)
- Ad Sizes
- Targeting (Geos, Devices + more)
- Behaviours (Ad Refresh + More)
Benefits of Fusion Experiment Technology
Our Fusion experiment technology is a incredible tool for making data-driven decisions, for optimising user experiences, and maximising revenue on your websites.
Multivariate Testing
You can easily perform A/B tests by duplicating configurations and comparing their performance. You could have 10 different configs (variants) running on 10% of the traffic to see which set up earns the most revenue.
Easy Configuration
Rather than requiring Developer resource to set up an experiment with custom code, through Fusion, you can quickly create experiments by duplicating an existing config, and using our UI toggles to turn certain features on or off.
Granular traffic Control
You can control the percentage of traffic that is exposed to each configuration. If one of your config variants looks to be outperforming the rest, you can upweight the percentage of that specific variant without fully committing.
Optional Revenue Impact Assessment Before Committing
You can assess the potential revenue impact of changes depending on the outcome of the experiment without fully committing to them by leaving the new configuration disabled.
Flexible Experiment Scheduling
If you don't like working evenings and weekends, schedule what date and time your experiments go live, and assess the impact of that experiment on your return to work!
Allow your Experiment To Run Indefinitely
We realise that lots of factors can be at play when running experiments such as seasonality, so our experiments don't need an end date. This allows you to put your winning variant from the initial test on 90% of the traffic for example but still leave 10% of the traffic for continual testing in case of any changes.
Lock Experiments To Stop Accidental Changes
Once an experiment is live, it becomes locked, preventing accidental adjustments and ensuring the integrity of the data collected during the testing period... if you wish.