Insights - FAQs

Frequently Asked Questions about the Insights suite of analytic reports.

What is App Insights?

T-Mobile Advertising Solution/s (T-Ads) App Insight is an tool that allows deep analytics via T-Ad’s proprietary Mobility Data. Mobility Data is T-Mobile’s exclusive data set that matches app ownership and engagement data to T-Mobile's devices, enabling marketers to analyze their mobile app interests and behaviors.

Our measurement products will allow marketers to quantify the impact of their campaigns, run through our audience network, in driving awareness for their brands and impacting outcomes.

General Use Case for Insights:

• Competitive Insights: Discover how your app compares to competitive apps for total active users, new installs, and engagement

• App Content Engagement: Understand if your app content is compelling enough to keep users engaged and increase usage

• Target Alignment: Identify which AppGraph personas over-index for using your app and see if they align with the expected brand target audience.

• Media Tactics Justification: Show clients the effectiveness of your business model/media plans by illustrating how they drove increases in Ownership/Installation/Growth/Retention for their brand. Find stories in the data to support your media decisions

• New Business Opportunities: Use the insights in new business pitches to illustrate to brands how you can help increase their ROI, including identifying areas where they could improve their tactics and how they stack up against their competition

• Audience Segment Creation: Build audiences based on App Insights to re-engage dormant or non-active users. Create Loyalist or Active audiences to market promotional messaging.

What is a Focus Event, and how is it used in Insights reports?

The building block of all Insights engagement data is the capture of, and understanding of, an app's focus event from a device. An app is 'in focus' on a device when that app is open, and it is the only app that can been seen and interacted with on a device's screen at a given point in time... hence the moniker that the engagement event was the focal point of device usage.

All engagement data currently in Insights is tied to the existence of a Focus Event record for a given app on a device. This includes foundational measurements such as dwell time, ownership and the state of being an active or dormant device. Its important to note that there are separate captured events for Install and Uninstall of an app.

Insights additionally utilizes a minimum dwell time threshold on all Focus Events, in which any Focus Event that is less than 2 continuous seconds (< 2 seconds) in duration is dropped from all data aggregations. This ensures that Insights is only presenting intentional focus events in its reports, removing those events that may be accidental or otherwise unintentional.

Why is the RUN button on the report page non-responsive?

In order for the Run button on any report page to allow a report to generate, there must be a selection made for every filter on the report page.

Note that the Date Granularity filter is always preset to "Days", but can be changed to either Weeks or Months as desire. All other filters on the report page must be manually set for the first run of the report.

What is the difference between Ownership and Installation metrics?

The key difference when evaluating results on the Ownership report page versus the Installation report page, is that a device contributing to Ownership report page measurements can have installed the app of interest at any time, but on the the Installation report page, the device contributing to the report page measurements must have installed the app of interest within the bounds of your selected date range.

For Example

• When calculating the number of devices that own App "A" on Dec 12, the devices which count towards this measurement may have installed App "A" at any time up to, and including, Dec 12. So long as the device still owns the app, ie. have not uninstalled App "A" by Dec 12, then that device is counted as owning the app.

• When calculating the number of devices that install App "A" on Dec 12, devices being counted must have downloaded App "A" on Dec 12. Reinstallations will count towards this measurement, so long as an installation action occurred on Dec 12 for App "A", then the device performing that action is counted towards the finished Installation measurement.

• When calculating the number of devices that uninstall App "A" on Dec 12, devices being counted must have uninstalled App "A" on Dec 12. So long as an uninstall action occurred on Dec 12 for App "A", then the device performing that action is counted towards the finished Uninstall measurement.

Use Cases

The Ownership report is better for cases where the total app footprint size is very important to the analysis. The Installation report is better for cases where end user adoption rates are more important to the analysis.

Why doesn't Ownership change commensurate with Installs/Uninstalls?

Ownership numeric values will change (rise or fall) based on three (3) factors:

1. The number of devices installing the app.
2. The number of devices uninstalling the app.
3. The number of reporting devices that newly show app engagement in the Insights system.

While the first two factors in this list are fairly self-explanatory, the third factor can be somewhat confusing. As part of the build out of the Insights product, there are backend systems which continuously monitor for reporting devices that show signs of app ownership. When these systems detect ownership from a device that is not already covered in the ownership footprint and/or covered by a reported installation event, that device is then immediately added to the ownership pool of devices for that app.

Due to the way that this system is recognizing and incorporating new devices all the time, its possible for the change in numeric value of devices owning an app to be mismatched to the reported number of devices installing/uninstalling the app. For Example:

• On 12/8 "App A" shows 1,500,000 devices owning.
• On 12/9 "App A" shows 12,000 devices installing and 2,000 devices uninstalling the app.
• However, on 12/9, "App A" shows 1,511,000 devices owning.

Notice that the 12/9 Ownership count is NOT the result of:
(1,500,000 Devices Owned + 12,000 Devices Installed) - (2,000 Devices Uninstalled) = 1,510,000

The 12/9 Ownership count is ACTUALLY the result of:

1. +12,000 Devices Installing
2. -2,000 Devices Uninstalling
3. +1,000 Devices Newly Discovered as Owning

• (1,500,000 Devices Owned + 12,000 Devices Installed + 1,000 Devices Discovered) - (2,000 Devices Uninstalled) = 1,511,000

NOTE: As the Insight's data tables are established more fully over time, the number of discoverable devices will continue to shrink until only Installs and Uninstalls are contributing factors to Ownership change.

Why am I seeing weird garbled characters in my CSV export?

If you are opening a csv file in Microsoft Excel and are seeing incorrect, or garbled characters, such a é, ü, Å, etc.; this could be because the CSV file is UTF-8 encoded. Insights uses UTF-8 because it is the standard for information exchange in modern applications. However, Excel does not automatically use UTF-8 when opening CSV files. To resolve this issue, please do the following after exporting a CSV file from Insights...

PC

• On a Windows computer, open a new Excel workbook.
• From the Excel Ribbon menu, click the "Data" tab.

For older versions of Excel...

• While in the "Data" tab, click the "From Text" option.
• Select the CSV you file want to import and click "Import".
• Excel will now display the "Text Import Wizard".
• From within the Import Wizard: Select the "Delimited" radio button.
• From within the Import Wizard: In the "File origin" field, select "65001 : Unicode (UTF-8)".
• From within the Import Wizard: Click the "Next >" button.
• From within the Import Wizard: Select the "Comma" checkbox.
• From within the Import Wizard: Click the "Finish" button.
• In the dialog window that appears, click the "OK" button.

For newer Versions of Excel...

• Click the "From Text/CSV" option.
• Select the CSV you file want to import and click "Import".
• In the next box, for "File origin" field, select "65001 : Unicode (UTF-8)".
• Click the "Load" button.

Excel will now display your CSV file, including non-English characters, properly.

MAC

If you are using Excel 2016 or later, then the above "For newer Versions of Excel" will work on Mac.

For older versions of Excel, you can select "File - Import", and then once you select the CSV you wish to import, you will be able to choose the encoding type to use.

Where does the Mobility data come from?

Mobility Data consists of App Focus (T-Mobile android OS only) and App Broadband (T-Mobile network) data and our Insights reporting suite incorporates those two sources while utilizing our Scaling methodology to represent the US mobile device population.

App Focus Data
App Focus Data are the foundational app engagement records that are at the core of the Insights product's metrics. This pool of users consists of ~7-8 million devices engaging apps on any given day. This data set is collected from those devices via a proprietary T-Mobile SDK or Headless application in the T-Mobile network which are post-paid Android OS, and in which the end user of the device has not opted-out of analytics data collection and usage.

App Broadband Data
The broader T-Mobile network facilitates the ability understand basic app engagement from between 40 million and 50 million Android and iOS devices on any given day. This wide breadth of devices which increases our coverage/reach of devices is used in App Insights to form the basis of the App Insights Scaling methodology.

Where does the demographic data come from?

The Age and Gender demographic data that is being pulled in is coming from self-reported user data (App Selector) during the Out Of Box Experience (OOBE) for T-Mobile devices. There is a prompt that will allow the user to identify their Age/Gender. It's all part of our Mobility data that powers App Insights.

How is the data projected?

Our multi-step Scaling methodology which incorporates both our App Focus data and App Broadband data allows us to represent and project the greater US population of mobile devices (inclusive of iOS phones and other carriers). To do so, we need to create distributions (Device Region, Device OS, Device Age/Gender, Device Service Type) of mobile devices to determine the footprint for weighting.

• We map the ~7-8MM Insights Reporting devices (App Focus) to the 40MM-50MM App Broadband devices
• Then a weighted Scalar Panel devices which are distributed and projected to represent approximately 273mm US Mobile Devices

Is there a reporting lag in data availability?

Data in the Insights reports are updated daily but there is a data lag of ~4 days.

• For Daily data, a day is reportable around four (4) days after that date has passed.
• For Weekly data, a week is reportable around four (4) days after the Sunday of that week has passed. Remember that App Insights' weeks are defined as a Monday - Sunday seven (7) day period.
• For Monthly data, a month is reportable around four (4) days after the last day of the month has passed. Remember that App Insights' months are defined as calendar months.

However, there are instances where our source data is delivered behind schedule that can delay processing up to several days (i.e. data delivery for a several days can add to the standard data lag).

Additionally, its it's important to note that data will continue to trickle in over the course of following 30 days after the day, week, or month is made reportable. Due to the continual trickle in of additional app engagement data, no day, week or month is final until 30 days have passed, at which point that day, week or month is locked from further change regardless of any additional data that may trickle in after those 30 days have elapsed.

Please keep in mind that this does mean that if there is a re-summarization effort made on historically locked data, those data may change to allow in the additional app engagement data that has come in after the 30 day lock was originally implemented.

What does the Logarithmic toggle do?

This toggle feature allows a user to change back and forth between logarithmic and linear scales. This feature is available in the Ownership, Installation, and Engagement Breakdown charts.

Logarithmic allows you to toggle the bar chart logarithmic scaling on or off so that you can get the desired presentation look and feel that you want. A good use case for the new Logarithmic toggle is comparing apps of a similar size against each other (Logarithmic toggle off) as the range of values are similar. When comparing large apps against small apps (Logarithmic toggle on) it will help better visualize a wide range of values than plotting the data on a linear scale.

A logarithmic scale (default for breakdown charts) is a way of measuring or comparing the relative size of numbers. It allows values to span a wider range since the differences between small numbers are shown as much as large numbers.

A linear scale is a way of measuring or comparing numbers by using a consistent value increments. It is a straightforward way to compare values, as the increments values are consistent.

How far back does App Insights data in terms of historical data availability?

In accordance with our 2 year data retention policy, our platform will be retained for a maximum period of two years.