Experiences in Logging Everyday App Use Marek Bell 1 , Matthew Chalmers 1 , Lucas Fontaine 1 , Matthew Higgs 2 , Alistair Morrison 1 , John Rooksby 1 , Mattias Rost 1 , Scott Sherwood 1 1 School of Computing Science, University of Glasgow 2 Department of Statistical Science, UCL. 1 {firstname.lastname}@glasgow.ac.uk 2 M.Higgs@ucl.ac.uk ABSTRACT This paper discusses our experiences in logging app use on computers, mobile phones and tablets. We have created logging software to record app launches on iOS, Android and Mac OS X devices, and have used this in a study with 13 students over a period of one month. This paper discusses the practicalities of logging across multiple devices, the forms of enquiry suitable for log analysis, and the ethics of logging. We also discuss future work in which we will scale the study up to thousands of users. Categories and Subject Descriptors H.5.m [Miscellaneous]: Information Interfaces and Presentation (e.g. HCI) Miscellaneous. General Terms Measurement, Experimentation, Human Factors. Keywords Mobile devices, Logging, Apps. 1. INTRODUCTION Mobile devices (including laptops, phones and tablets) are central to day-to-day life for many people. Many HCI studies have studied usage of individual apps (e.g. [4]), but such individual logging does not say a great deal about the holistic use of a device. Services such as Flurry (www.flurry.com) have been able to go some way to building pictures of overall device use by virtue of their logging software being built into many thousands of individual apps, but only have data on those apps that incorporate the Flurry service. More recently, several quantitative studies have been published [1][2][3] that create a fuller picture by examining all apps launched on mobile devices. For example, Böhmer et al. [1] have logged and analysed app launches on over four thousand Android devices, and Do et al. [2] have analysed logs provided by Nokia of 111 of their users. In this paper we discuss our own experiences in holistically logging app use on mobile devices. Our work extends previous quantitative studies by logging app use across several operating systems. Ours is the first study, as far as we are aware, to log the use of all apps launched on iOS devices. We have also logged launches on Mac OS X and Android devices. This enables wider coverage, but also enables us to examine how people use devices in tandem (for example how someone switches between an iPad, an iPhone and a MacBook). Our study can also be differentiated from prior work by our use of qualitative interviews in conjunction with quantitative analysis, and our concern with the user experience and the ethics of logging. This paper reports on the development and use of a suite of logging applications known collectively as AppTracker. The paper discusses a study of thirteen students over a one-month period. This study is a precursor to a large-scale trial we will run with many thousands of users, and therefore this paper makes no attempt to generalise about populations. Instead it addresses issues associated with the practicalities of logging, directions for analysis, and ethics. 2. APPTRACKER We have developed logging applications for three platforms: iOS (i.e. iPod, iPad, iPhone), Android, and OS X. AppTracker logs the time when apps are opened and closed, and when devices are locked and unlocked. We also log the connectivity of the device and when it is being charged. We do not log anything that is done within individual apps, or network activity – only when apps are opened and closed. We have taken a conscious decision not to log websites visited on the device or location information. Previous work by Böhmer et al. [1] and Do et al. [2] has shown such information to be useful, but we were uneasy about the ethical implications of this (and would ourselves not be comfortable revealing such information). The logs are stored on a database. To implement AppTracker, we had to resolve a number of technical and conceptual challenges. These included: • Backgrounding is not readily supported in iOS. We therefore had to devise an approach that would enable a logger to run unobtrusively over the long term. To our knowledge ours is the first log based analysis of iOS devices. • Logging inevitably has an effect on the performance of the devices. We therefore had to minimise battery consumption and data transmission. • Working across platforms required us to address how a consistent log can be generated from diverse devices. • Clearly, there are also ethical considerations. We minimised ethical concerns by choosing not to log location or any content. We treated ethics as an ongoing topic during the research (for example by discussing privacy issues with participants during our interviews). 3. PILOT STUDY For the study we recruited 13 students at the University of Glasgow. We asked them to run AppTracker on one or more device for a minimum of 30 days, to complete a questionnaire and to sit for a semi-structured interview with a researcher. This was an exploratory study designed as a first foray for AppTracker in Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Digital Economy ‘13, Nov 4–6, 2013, Salford, UK. Copyright 2013 ACM 1-58113-000-0/00/0010 …$15.00.