Use predictive analysis to prioritize performance improvements

The what-if analysis widget is a data science feature that uses predictive analytics to tell you the impact of performance changes on conversions, revenue, and other business metrics.

With version 3.0 of what-if analysis, you have more flexibility in how you interact with the interface. You can choose which timing metric to analyze, which filters to apply, and which dimension to break down the analysis. You’ll be able to see how shifts in a sub population will contribute to the overall revenue increase.


The summary provides real-time performance results based on the metrics and the time period that you select. The Reality tab is the default view that shows your current performance results along with projected improvements. The Possibility tab shows your projected performance results.

Current performance

This example of the reality view shows the number of visits (1.6 million) recorded and the number of beacons gathered (30.7 million) within a four-day time period. The chart shows the median load time (2.47 seconds) and the median conversion rate (8.72%) during this time period. All of the recorded metrics are listed on the right. The x-axis represents the session load times in milliseconds. Projections on improving your total revenue are calculated for you.

Projected performance

Based on the projections given in the reality view example, this possibility example shows the results of the projected performance improvements. The chart represents how changing the median load time by 419 ms during this time period can improve the median conversion rate by 1.02 points, which in turn improves the daily revenue by $1.6 million dollars. The projected metrics are listed on the right.

How are projections calculated?

Each time a what-if analysis is generated, mPulse calculates the optimal load time to get the highest conversion rate with the least amount of change in load time.

Load Time & Metrics

This view shows an individual chart for each of the business metrics that you want to optimize for.

The metrics are projected based on the real user traffic on your application and reflect predicted behavior based on a given page speed. The circles are your goal metrics. These are the points at which mPulse predicts the best improvement in a metric relative to the improvement of the page speed. The horizontal lines are how mPulse predicts those metrics will change. Your current business metric and load time position are shown under Current. The session load time (page speed) is on the x-axis.

To adjust a metric, enter the value in the field provided under Projected or drag the slider up or down. To adjust the page speed, click the speed until it reaches the desired value. Once you set your goal speed, the other metrics are updated, and you’ll see the new projection in the summary’s possibility view.


The breakdown view shows performance distributions based on the selected Group By filter. In the example below the Group By filter was set to “Country”. The top ten attributes of the dimension are shown in this view, and each can be modified independently. Business metrics and overall site speed are projected based on the individual modifications. Speed can be adjusted by moving the page load time slider under each chart; when released the chart and metrics will update to present the new results.


Use the filter bar to select your app, timers, metrics, dimensions, and date range to capture performance results, as shown in this example. The Group By filter will enable the Breakdown section of the report, revenue and conversion metrics should be set here if available.
This widget is available from the widget type directory. Just Create a custom dashboard, drag and drop the What-If v3.0 icon on the dashboard, then use the filters to select the time period and metrics for your analysis as shown in this example.