Thursday December 4th, 2014 by Heiko Tietze and Thomas Pfeiffer
KGet is a versatile and user-friendly download manager. It has a lot of useful features and was designed to work well with Plasma and other KDE applications. KGet has a long history in KDE, the first version being shipped with the KDE3 series. The maintainer started porting to KF5 a few weeks ago and felt like the time is right to make it look more beautiful and asked the Visual Design Group for support. And as part of this design process we run a survey first on how people make use of this tool, what they like and what they need.
The call for participation was accomplished by a short voting with the goal to invite only potential users and interested people. We asked all except for those who refuse to use any download manager to participate in the survey, and asked the others for a short comment.
Table 1: Do you use KGet? (Total number of voters: 411)
|Yes, regularly or from time to time||41%|
|No, but I'd consider it if it had some features I'm missing||23%|
|No, I'm using a different download manager||12%|
|I don't want to use any download manager||24%|
The first two questions were asked to reveal groups of participants. For instance, there might be user groups with special needs, or the results depend on the way KGet is being used.
Usability starts with the common view on the user, which can be characterized by personas. The best way to create and verify persona without clairvoyance is to ask people whether or not they fit into one of the descriptions. In this study we presented the short label of the five KDE personas with the hypothesis that KGet is used by recreational users (Susan) or tech-savvy ‘optimizers’ (Matt).
The result clearly represents the audience of Planet KDE: highly tech-affine people who love to fiddle around with software. Surprisingly often is Susan represented in this respect. Berna, who represents the less tech-savvy office user, was not reached with our survey. We need other communication channels to contact her.
For the results we analyze Philip, Matt, and Susan in detail with the idea that these user groups have different scopes.
Type of Integration
KGet can be used as a plugin or stand-alone tool, and since we also invited people who do not use it yet we asked about how it is used currently.
The usage types are fairly equally distributed. For the evaluation it is of special interest what benefits would persuade potential users.
Familiarity with features
Next question was about the familiarity with KGet. Participants had to estimate their knowledge regarding a couple of functions on a Likert scale between ‘never heard about’ and ‘use it regularly’. An even-point scale was used to force a choice. The following figure 3 shows the result as a deviation from the expectancy value of 3.5 along with the additional information on how many people state that they never heard about that feature.
Obviously KGet has a problem with the discoverability of some of its features. Metalinks, transfer lists, groups, and destinations are unknown to about 50% of all participants. On the other hand, people know the function (or expect it from a download manager) to allow pausing and to integrate well into the system.
Of course familiarity depends on the user and the way he or she uses KGet. Participants who associate themselves with Susan report to ‘never heard about’ for groups and destinations with about 25%, but metalinks and transfer lists are similarly unknown. And both Matt and Philip have comparably low values like all users with 50-60% never having heard about the features.
A clear (though rather trivial) difference can be found for people who use KGet as stand-alone tool: they are much more familiar with adding a download through its interface. Similarly the unfamiliarity increases for participants who do not use KGet.
Reasons to use KGet
The question about the reason to use KGet was asked with the goal to identify the core features. That means a feature that is not used regularly can still be very relevant. Several options were offered in the survey for multiple choice.
Familiar features are the most relevant ones, like pausing the download, having the tool nicely integrated into Plasma, and the support for major protocols. Least interest is found for the monthly data cap, groups (probably because this function is rather unknown), and accessibility.
The picture is similar for individual personas. Grouping gets a little bit more relevance for Philip (for about ~30% it is an important feature). Provided protocols and easy handling is about 10 to 15% less important for Matt (compared to Susan and Philip).
And in respect to the integration or rather users that use none or other download manager the greatest difference is found for the nice design (important for 24% of non-users vs. 11% of current users). There is no particular feature that would turn the refusers into KGet’ters: providing the default protocols is more relevant to non-users (56 vs. 42%), the capability to pause the download is more important for current users (72 vs. 84%).
Visualization for downloads
To separate the visualization from other requirements (that will be reported in another posting) we asked which information they want to see.
Feedback is always highly appreciated. Less relevant for users are details on time of pause, origin, and destination.
Finally we included some kind of benchmarking test in the study. The ISO norm 9241-110 demands software to meet a couple of non-functional requirements. We ask participants for their rating on those categories using a five point Likert scale. Values below 3 should be treated as down-voting. Since we ran the same test some time ago for Dolphin we are able to compare the results here.
KGet receives throughout a positive rating albeit lower compared to Dolphin. Filtering out the participants who do not use KGet leads to increased convergence but still shows the same general picture.
If you want to analyze the data your self here are the raw data along with the R scripts: 20141201_KGet-Descriptives.tar.gz
Requirements for the future release
After the actual survey another test was started asking users for their needs and preferences. The questionnaire utilized the Kano method and was provided on a different platform. Results can be found in the posting on “Using the Kano method to prioritize requirements“.
The short survey addressed the question how KGet is used and if people are still interested in it today with flat data rates and high speed access to the Internet.
- KDE persona works
Participants are able to characterize themselves as one of the predefined KDE personas with a distribution that makes sense from a methodological and statistical point of view. We expected Susan and Matt as the primary users for KGet but most participants are rather like Philip. This result might be an effect of the selective sample of Planet KDE, Twitter, Reddit, and Google+ where the study was announced.
- Audience is not too large but has some potential to increase
In the voting 41% reported to use KGet at least from time to time. Another 35% could be understood as potential users that consider to use KGet or might switch from other tools.
- Features are rather unknown
Many of the fancy features that KGet offers are neither known by participants nor by the actual users. Of course, if users do not use a feature regularly this does not automatically mean it is not important. Asking for the reason to use KGet did not reveal any particular killer feature though.
One can interpret this results in two ways: all the advanced stuff is not necessary for the average user. She just wants to have control over the download resources. Alternatively it might be treated as a communication/promotion and presentation problem. For example, the UI could be designed around the group feature, making it a core aspect. Users that are forced to use it may get easily convinced of the advantage – even when they do not need it yet.
Learn more about requirements for a future KGet in the results from the parallel study utilizing the Kano method.
- Benchmarking positive with some room for improvement
The assessment of KGet in respect to non-functional requirements shows positive results but as well some room for improvement when compared to Dolphin. On the other hand, Dolphin is an advanced program, widely used and well known. For KGet it could make sense to have less options for individualization, if the program should be simple not only by default.
Putting all together we see two directions for the future development. First, the data support the idea of a simple tool that provides more control over the download and integrates well into Plasma. A nicely designed plasmoid accomplished by a plugin for Firefox and Chrome might solve the tasks for most users. The other way is to enhance the interface for advanced usage. If KGet would be some kind of central interface between the local system and resources on the Internet supporting all respective tasks it might become very useful too.
What do you think?