Experiments allow you to make data-driven decisions to better improve your products and overall customer experience. However, not all experimentation results are equal. Even when you have results that seem like clear wins, and losses, there can be underlying data that proves to be otherwise. So, is there a way to use the data gained across all your experiments and enable a continuous experimentation lifecycle?
Dimensional Analysis helps you and your team dissect experimentation results at a deeper level, giving you the insights needed to make better informed future hypotheses and experimentation iterations.
For example: Let’s say your team is running an A/B test on a new checkout flow. After running the experiment, you notice your page load time was severely impacted but might not know why that’s happening. With Dimensional Analysis, you’re now able to break down the data to understand if there is a specific platform or device causing the downgrade. In this case, we notice that the page load time on iOS is what’s driving the negative impact. With this information, you are then able to further investigate and iterate on a more optimized flow for iOS.
Dimensional Analysis isn’t just for Data Scientists, but also valuable for Product Managers and Engineers.
“Through a very simple setup we were able to see our metrics broken down by country, language or device viewport size. This gave us useful insight that would have previously required setting up additional metrics. As a result, we now have an even better understanding of the effects of our changes.”
– Michal Filip, Principal Software Engineer, Monster
*With this feature users will be able to: *
- Quickly triage results: Easily use our Impact Snapshot metric chart to visualize the impact for your metrics.
- Gain deeper insights: Investigate unusual spikes in your experiment results at a dimensional level to better understand what action to take next.
- *Iterate on what experiments to run next: * Get an accurate understanding of what worked or didn’t during your experiments, and use that to build your next hypotheses or follow-up experiments.
For customers using our “Monitoring & Experimentation” product suite, getting started with Dimensional Analysis is easy. Simply send us your events along with your event properties and we’ll use that to create your dimensions. Within Split, you’ll be able to configure five dimensions with five property values each, completely custom to your business needs. Once configured, Split will automatically calculate key metrics associated with the dimensions defined for every experiment – it’s that simple.
Need more information? Be sure to check out our additional resources to help you along your experimentation journey: