Understanding the context of use – where and why a consumer is engaging in a particular activity– is the holy grail of audience measurement. Data points such as demographic segments (age, gender, profession, household income), temporal aspects (time of day, season, events), geolocation, and other behavioral data give us better insights into the specifics of the context of usage of a particular digital product or service.
At Verto, we are constantly trying to improve on this capability for our clients, focusing specifically on the context of use of digital services. Digital consumers today engage with multiple apps and websites during a single session on their digital devices – smartphones, tablets, or PCs. Switching or multitasking between different devices, apps, and websites adds a significant dimension to the overall context of use. Context can therefore depend upon:
- Which apps/sites are used together in the same context?
- Which apps/sites are used before/after a target app/site?
When a user engages with a device, they typically multitask between various apps and websites. But when they take a break from using the device and come back later, their engagement behavior with the device can be viewed as a “soft reset.” Within these repeated engagements with a device, we can search for groups of device sessions that have similar characteristics. These groups of similarly characterized engagement sessions are the different “contexts of use” of digital services. Thus, to construct the framework for analyzing this aspect of context, we need to define what constitutes a device session.
A device session is a series of app sessions carried out on a single device with no longer than a specified time threshold between the end of one app session and start of the next.
The idea is to define the threshold so that a device session would effectively allow us to identify and measure a sequence of user activities that occur in a continuous fashion, without the user leaving or closing the device.
The following two charts provide snapshots of app sessions on mobile (smartphone) and PC over a single day (a 24-hour period) for 15 randomly-selected unique devices of each device type, taken from our consumer panel data. The charts show different rows for different devices and the different colors in each row represent the different app sessions for the device. We can measure and analyze web sessions from their corresponding browser app sessions.
Unsurprisingly, we see that the app session durations on PC tend to be longer than those on mobile. We can also see that there are “bursts” of device use followed by gaps of inactivity during which the device is not used. These bursts of temporally close app sessions are what we consider to be device sessions. Meanwhile, the gaps of inactivity are of interest to us because we want to define a minimum gap length that is likely to result in a different context from the next set of device activity. We aim to identify a timeout that groups app sessions together but does not inappropriately cover long periods of device inactivity.
To select the best timeout threshold, we conducted measurements with various timeout values, namely 30 seconds, 60 seconds, 120 seconds (2 minutes), 300 seconds (5 minutes), and 600 seconds (10 minutes).
The chart below visualizes the decrease in the number of device sessions as a function of the timeout length on PC and on mobile. Based on the chart below, the number of device sessions as a function of time out decreases rapidly with 30, 60, and 120 second cutoffs, and starts to level off after 300 seconds.
Similarly, the chart below provides a visual representation of how the gaps between the different sessions are distributed. We can see the cumulative distribution function (CDF) of all the gaps between app sessions for mobile devices and PC, which shows us that approximately 70% of the gaps between different sessions are less than 300 seconds (five minutes) long.
These data points indicate that if we use a 5 minute timeout, we can largely address the shorter gaps that occur in between app sessions. A 5 minute time out value also gives us an average of over 4 app sessions per device session on PC and mobile (1.7 unique apps per device session on PC and 2.5 unique apps per session on mobile), which we believe offers a sufficient richness of data to provide our customers with insights into multitasking and co-occurrence of digital service use. We also observed that the typical consumer engages with an average of 180 device sessions per month on PC and an average of 500 device sessions per month on mobile. As previously noted, PCs encourage longer average session durations: the average device session duration on PC was 53 minutes, while the average device session duration on mobile was just 10 minutes.
Using this framework, now we can dive deeper into questions around the context of use of app and websites, such as:
- What is the number / share of device sessions where a certain app/site is used?
- What is the share (of time, sessions, or other measures) of a certain app/site out of the related device sessions?