User Profiles

[Paul W. Frields]

On Thu, Nov 19, 2009 at 04:03:52PM -0500, Máirín Duffy wrote:
> On Thu, 2009-11-19 at 15:06 -0500, Máirín Duffy wrote:
> > Since this message is already really long, I'm going to cut this here.
> > Next I am going to make a post about our options on moving forward with
> > a user research plan.
> 
> So first, to figure out which user research methods to use and the
> research deliverables we'll want to produce, I think we need to think
> about to what ends we would like to employ these personas. Here's my
> stab at it:
> 
> - Personas will help us make informed design and policy decisions about
> the default configuration of software in Fedora itself. We need to
> consider these persona's needs and situations when making decisions
> about default application behavior in Fedora, and even the look & feel /
> appeal and style of the default artwork.
> 
> - Personas will help us determine what tasks our target audience wants
> to accomplish with Fedora. This will help us figure out good default
> package selections for Fedora, and also to figure out, for a given task,
> which application is best suited to get the job done.
> 
> - Personas will help us determine the timbre of our messaging in both
> marketing materials and our website in order to attract the very target
> audience we are hoping to gain.
> 
> - Personas will help us streamline the main flows of the Fedora project
> - making it easier to download and install Fedora itself, making it
> easier to join Fedora as a contributor, and making it easy to get help
> with Fedora. 
> 
> - Release-engineering-wise, they can help us determine which updates are
> appropriate to be released when and the appropriate severity for them.
> E.g., it may be determined some sets of packages need to be under
> stricter guidelines than others based on the usage patterns we discover
> in our target audience.
> 
> - Can you think of any other uses?

There might be implications for the test cases on which we concentrate
for Fedora, and the content and prioritization of test days, given
what I see below.  Besides those ideas, this is a great summary of the
goals I'd had in mind.

> I think the kinds of deliverables we are going to want to focus the
> research towards producing then could be along these lines:
> 
> A) A task list. This is a simple list of the things people are using
> Fedora to do. To help in prioritization, indications of frequency and
> how widely practiced the tasks are across the user set should be given.
> Tasks should be goal-centric, not application-centric. E.g., 'Use
> firefox' is not a task. 'Create a logo design for my business card to
> send to the printers' is a task. The task should have a clear goal. 
> 
> B) Most frequently-used applications list (should include rich client /
> desktop apps AND web applications.) Mugshot used to have a tool that
> collected this data, and I think it's in gnome-shell now. Is there any
> way we could allow users to opt-in to this sort of application usage
> data collection?
> 
> C) A list of peripherals used with Fedora and what task workflows they
> are a part of. (Downloading photos off of a digital camera to post on
> your recipe blog and email the link to grandpa, Plugging in a scanner to
> digitize old family photos and store them on a consumer NAS, filming a
> music video for a school project grabbing the video off the camera
> editing and uploading to youtube then emailing the teacher with the
> link, connecting the computer to a TV and watching the movie with the
> whole family, etc etc etc.) Some notion of how frequently the tasks are
> performed / how many users perform them will be important to prioritize
> efforts to streamline these workflows and produce guides / other content
> for the website and marketing materials to show users how to accomplish
> them with Fedora.
> 
> D) Applications commonly installed post-Fedora install. Looking at the
> applications a user installs on top of Fedora, and the type of
> configuration they do on a machine beyond the defaults I think would
> help inform us where the default configuration / package set falls
> short, and/or how packages and configurations might be chunked together
> to make it easier to find them and get them installed.
> 
> E) And of course, most importantly, a set of user personas extrapolated
> from all the data collected during research.
> 
> Here are some user research methods I think we could employ to gather
> this kind of data. Some of them come from my set of IDEO method cards
> [1]:
> 
> 1) An opt-in application usage collection mechanism as described in B
> above.
> 
> 2) Maybe towards C and B, a listing of the Fedora-related bugzilla
> components in terms of frequency folks are filing bugs against them. The
> more bugs, perhaps the more visible problems are in those applications
> thus they might need to be prioritized more.

That makes sense and would probably be pretty easy to gather.

> 3) Surveys & Questionnaires - these are easy to do, although it would be
> nice if we had a survey system in our infrastructure to conduct these. 

There is a polling mechanism packaged (I believe) for Zikula, the CMS
that the Fedora Marketing team is working on at FUDCon Toronto 2009.
Also, I believe Ian Weller or someone involved in the Marketing team
had looked into packaging LimeSurvey, which is also 100% FOSS and thus
eligible for Infrastructure.

> 4) Personal Inventory - interview a handful of target audience members
> and ask them what objects related to their computer and computing
> lifestyle are most important to them and why. Produce a catalog of the
> items per interviewee. This will help us come up with a task list (A)
> and understand how Fedora might fit into our target audience's lives.

How do we go about selecting these people?

> 5) Be Your Customer - with the task list (A) we come up with, members of
> the Fedora project should walk through / enact performing those tasks on
> their own to understand all the issues that arise. Out of this a
> document listing out issues that need to be addressed in order to make
> the tasks easier to perform could be written and bugs filed as
> appropriate.

This is brilliant -- occasionally in rare moments of downtime I work
through a procedure to see what it takes, and use that to file bugs
myself, but hadn't made the contextual connection with the user
profiling.  In large part, this is also how the Desktop SIG worked on
their "Fit and Finish" days to provide additional shine for Fedora 12.
  
> 6) Scenarios - write up a character-rich story involving a made up user,
> who is clearly a member of the target audience, interacting with Fedora
> to provide context for how Fedora is used.
> 
> 7) Behavioral Archeology - look for evidence of activity based on
> organization / placement of things. We could ask for people in our
> target audience to send in screenshots of their desktop and take note of
> what changes they made to their desktop.
> 
> 8) Cultural probes - make 'camera journal kits' including a camera,
> notebook, and instructions and send it out to members of the target
> audience asking them to keep a photo journal of their experience with
> their computer over the course of some set time period. One way of doing
> this to help the participants remember to do it is to send them some
> signal at frequent intervals (phone, text message, email) and ask them
> to write down what they're doing on their computer as soon as they get
> it. 
> 
> 9) Fly on the wall - sit with members of the target audience for a few
> hours and simply observe how they interact with their computers without
> interfering with their activities. 

This is akin to a "sit bird-dog with a notepad" method?

> 10) Think-Aloud Protocol study - Similar to fly on the wall but more
> interactive - observe a member of the target audience interacting with
> the computer and take on a mentor/apprentice role with them - with you
> as the apprentice - and ask them to explain what they are doing as they
> do it. 

Seems like this would be tough to do without role-reversing!  But it
would be fascinating to watch someone with the proper discipline do
it.  I could imagine a third person acting as coach to help people
learn how to do this right.

> 11) Extreme User Interviews - pick a set of target audience members who
> are completely unfamiliar with Fedora and ask them to give it a try,
> writing up their experiences. 

The SO test?

-- 
Paul W. Frields                                http://paul.frields.org/
  gpg fingerprint: 3DA6 A0AC 6D58 FEC4 0233  5906 ACDB C937 BD11 3717
  http://redhat.com/   -  -  -  -   http://pfrields.fedorapeople.org/
  irc.freenode.net: stickster @ #fedora-docs, #fedora-devel, #fredlug