Determining Experiment Results

Once you have an experiment running, you’re able to see it on your Experiments page:

You can see the name, type, status and dates.

Through the status you can see what experiments have been completed, in progress or in setup.

In progress experiments

If you click on an experiment that’s in progress, you’ll see it’s current metrics:

For the experiment in the example above, the results are still inconclusive.

That can either be because the changes are not impactful enough, or not enough time has passed.

At any point, you have the option to either End the experiment, or apply the experiment to your campaign.

You can also “overwrite” the Base campaign with the Trial campaign. That means that any changes will become active on 100% of the traffic.

Completed experiments

For completed experiments, you can see the results in the ‘Experiment Summary’

The screenshot above shows all the details for the base and trial variations.

You can then compare the base and the trial with your chosen metrics.

In this case, it shows that  clicks went up by 147%.

Below that you can see ‘95% confidence interval’, +130%-164%, Statistically significant.

According to Google, the higher the confidence level, the less likely luck was involved, meaning the experiment was statistically significant.

In this example, it indicates that the variation was considerably different and not down to luck. Soyou can be fairly certain that if you make changes to the main campaign, you can replicate these results.