Studies are the core object in the Attribution suite. A Study is a child of a Campaign, and one Campaign can have multiple Studies. Studies can be thought of as an instance of a Study Type with its own parameters.

In most cases, users will only have one Study per Campaign. However, there are scenarios in which it might make sense to have two studies. For example, if you want to Study the same campaign with different attribution windows, this could be presented as two different Studies of the same campaign.

Study Types

There are 3 types of Studies Mira runs

Net Lift Study

Net Lift Studies, sometimes called "Four Prop (Proportion)" Studies, our standard studies. These make use both a pre period and control group for, which leads to 4 different conversion rates:

Post-Only Study

This Study is used when there is no pre period, either by design or due to Conversion Properties not being configured in time. As such, this study makes a comparison between the a control group and a test (exposed) group during the campaign/post period, and produces a single lift.

Exposed Only Study

In this study, there is no control group. The Exposed group is observed in the pre period as a baseline for performance, and lift is computed as the percent change from pre to post that the exposed group experiences.

Study Structure

This section describes how Studies are run

Cuts

The simplest study will have a single set of numbers reported: the control conversion rate, the exposed conversion rate, and the lift between the two. "Cutting" refers to the process of grouping the results in different subgroups and observing the effects.

Cuts can originate from factors on either the exposure side or conversion side of the attribution join. Common exposure cuts include:

• Market of exposure
• Vendor
• Format
• Creative or group of creative
• Device type (iOS and Android)

Exposure side cuts reduce the denominator as well as the numerator of a conversion rate, because they limit the group of people being considered outright. Conversion-side cuts reduce only the numerator, since it is only limiting conversion events. Common conversion cuts include:

• Property
• Goal
• Market of conversion (footfall)

🚧

Limited Console Support

Studies created in the console currently only support Property and Goal/Pixel cuts.

Attribution Window

The attribution window, sometimes called a lookback window or conversion window, is the amount of time a user has to convert after being exposed, in order to be counted as an attributed conversion.

For example, if a user is exposed on 2020-12-01 10:21:00, and the window is 7 days, then they are a converted user only if they have a conversion event between 2020-12-01 10:21:00 and 2020-12-08 10:20:59. If the user doesn't convert until, say, 2020-12-10, then that conversion event would not lead to an attributed conversion.

Study Periods

In the ideal scenario, we observe two groups - the exposed and control - in two different periods: the pre and the post. While there is always a post period, whether or not a study has a pre period is up to the user and depends on when conversion data is available.

The post period is the period of time from the start of the media campaign to the end of the media campaign plus the attribution window. For example, if we have a four-week campaign starting 5/1 and ending 5/27, and the attribution window is 7 days, then the post period spans from 5/1 to 6/3.

Pre Period

The pre period is a period of time prior to the start of media. The pre period lasts for the same amount of time as the attribution window. Usually this is a period of time immediately preceding the start of the campaign, but this doesn't have to be the case.

The point of a pre period is to measure a baseline or ambient conversion rate for either group. Ideally, the control and exposed groups have similar conversion rates during the pre period, since prior to the campaign start the exposed group has not yet been exposed.

Whether a pre period is done is up to the client. However, in order to be eligible for a pre period, a study must have a conversion environment configured before the start of the pre period!

Creating A Study

To create a new study in the Mira Console,

• Navigate to the Campaign you wish to study
• Click on the Studies tab
• Click the + New Study button
• Select the type of study. Most users should select the Four Prop Lift Test
• Provide a study name and description
• Choose an attribution window, in days
• Set the study date range. The Start and End dates will default to that of the Campaign the Study is a child of. If using a pre period, these dates will automatically be set based on the attribution window. However, they can be manually overridden.

Associating Conversion Properties

After a Study has been created, it needs to have Conversion Properties associated with it. To do this,

• Navigate to the specific Study
• Scroll down to the Conversion Properties section
• Click the + Add Properties button
• Select the Properties you want to add one at a time, and click the save button.

Results

After a study is created, your account rep will reach out to you with any questions and confirm that the Study has been correctly configured. The Results Results will become available after the campaign ends.