View source for Labs/Test Pilot/Test Proposals/Tab Open/Close Study

= Tab Open/Close Study =

* Champion: Limi  (limi@mozilla.com)
* Status: Draft
* Duration: 1 week
* Implementation: Not started yet.  Will be implemented with Test Pilot extension.
* Data to collect: 

== Introduction ==

Tabbed browsing is now the standard interface offered by all major web browsers, and its been argued that the browsing experience becomes difficult to manage when more than a handful of tabs are opened.  

In conjunction with our recent focus on tabs and their possible reinvention (see the [http://design-challenge.mozilla.com/summer09 Summer 2009 Design Challenge: Reinventing Tabs], as well as in numerous blog posts) this study will seek to provide designers and developers additional information and insights to help inform the design of future tabbed interfaces.

== Goals ==
This will be the very first study of tabbed browsing through Test Pilot. Our goal for this study is to understand and design:
* Tab behaviors after people close a tab (approved)
* Ways to manage and navigate among a large number of tabs
* What the distribution of tabs are, wrt. how many people have open, max/min.
* How many people open new windows to group tabs
** How widely spread are sessions?  Ex. Google search and new results in tabs
** How long do split sessions survive? 
* Behavior around tab dragging and docking (approved)
* Tab selection errors 
** T2(tab select) - T1(tab select) < 1.5s

== Instrumentation ==
* Tab open/close hooks are easy
* Tracking "open in new window"(OiNW) and subsequent metrics is not easy.
** Domain overlap
** Lifetime variation in OiNW tabs versus other tabs (expectation is shorter if these are truly speculative opens, preserving the original)


== Details  ==

===== '''Design question 1:''' What would be the best default tab behavior after users close a tab?  =====

In the past, we have observed that people close a tab in different contexts, for example:<br> 

#When people finish using the tab: e.g. users close the tab when finish watching a video. <br> 
#When people organize browser tabs: e.g. users close some less important or less used tabs when there are too many tabs open in a browser. <br> 
#When people temporarily hide a tab: e. g. users close a tab when he detects someone is passing by; he would like to open it again when nobody is watching. <br>

For different contexts, people have different intentions after close a tab, for example:<br> 

#open another interesting tab<br> 
#close another less important or less used tab<br> 
#open whatever tab<br> 
#open the most often used tab<br>

For this test, we would like to to explore what would be the best default browser behavior after users close a tab. We will simply log what users would do after he/she close a tab, and through these log data, we may be able to detect certain patterns that help us design the default browser behavior that meaningful to most users.&nbsp; <u><br></u> 

<u></u> 

<u>Annotations:</u><br> 

*T1: the tab user just closed it<br> 
*T2: the tab shows automatically after user closed T1<br> 
*T3: the tab user selects after T2<br>

<br> 

<u>Metrics to consider: </u><br> 

*User actions after close T1:&nbsp; T3 - T2=? 
*Relations between tabs:&nbsp; relations between T1&amp;T2? relations between T1&amp;T3?&nbsp; relations between T2&amp;T3? (e.g. opened through the same tab? within the same domain? opened at the same time? no relation at all?) 
*Notable charactor for T1,T2 and T3: e.g. most often used? open the longest? previously opened? <br>

<br> 

<u>Generic parameters to log:</u><br> 

*How many tabs are open in total when users close T1? 
*Any add-ons or settings may influence this user behavior?

&nbsp;&nbsp; &nbsp; &nbsp; &nbsp; * so far 273 add-on is related to tab behaviors based on AMO. <br> 

*Considering both mouse action and keyboard action <br>

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; * close the current tab: command+w (Mac); contrl+w (Win) <br> 

===== '''Design question 2:&nbsp;''' What should be the best default placment when users open a new tab?  =====

We have observed that people open a tab for different reasons, for example:<br> 

#Open a new blank tab to type URL or to perform search<br> 
#Open a new tab through links within current tab<br>

We also notice that people doing following actions after they open a new tab:<br> 

#drag the newly opened tab next to the previous tab<br> 
#drag the newly opened tab next to a particular tab <br> 
#drag the newly opened tab to a new window<br>

<br> 

For this test, we would like to explore what we can do to support people open a new tab in a meaningful way.<br> 

<u></u><br> 

<u>Annotations:</u><br> 

Tc: the opening tab before user open Tn (could be Tl or Tr)<br> 

Tn: the newly opened tab<br> 

Tl: the tab left to Tn after Tn is dragged <br> 

Tr: the tab right to Tn after Tn is dragged<br> 

<br> 

<u>Metrics to consider: </u> 

*Type of Tn:&nbsp; with content (opened through a link), or blank?<br> 
*User actions after opens Tn when it is blank:&nbsp; how users load content for the newly open blank tab?&nbsp;e.g. type URL, perform search, through other means? After the loading the content, do user perform drag and drop to group Tn with other tabs?<br> 
*User actions after open Tn when it is with content: activate Tn immediately? Time between open Tn and activate Tn? do user group Tn with other tabs?<br> 
*Relations between tabs: relations between Tc and Tn? relations between Tn and Tl/Tr? '''<br>'''<br>

<u>Generic parameters to log:</u> 

*How many tabs are open when users open and drag Tn?<br> 
*Any add-ons or settings may influence this user behavior?

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; * so far 273 add-on is related to tab behaviors based on AMO. 

*Considering both mouse action and keyboard action

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; * open a new blank tab: command+t (Mac); control+t (Win); right click on a tab, through menu button<br> 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; * open a new tab through a link: command+click (Mac); control+click (Win); right click, setting in Firefox 

<br> 

<u>Basic info for this test:</u> 

*How long the browser is opening for?<br> 
*Tab/window amount per person during the browser opening<br> 
*How many times Firefox becomes unfocused during the time opening? 
*Permanent tab amounts open during the broswer opening. <br>

<br>

= Test Data Analysis =

== Tab-related addons ==

'''FoxTab'''

'''Tab Mix Plus (TMP)'''

'''Tabs Open Relative'''

'''FireGestures'''
*Observation 1:
when switching/closing/undoing close tab, using
fire gestures, it logs that I did this with the keyboard. I don't know
if this is a problem or not, but thought I would put it out there.

*Observation 2:
I have another issue but the same problem with UI type (FireGestures
installed). I'm using TabMixPlus to open new tab just after current
one. My test case:
- I have two opened tabs: A, B
- Switching to A using mouse - test pilot logs: "0, Switch, Keybord"
- Click link on page A, which open a new tab (named "C" for example)
between A and B (so I have: A, C, B)  - "test pilot logs: 2, Open,
Link"
- Switching to B using mouse - test pilot logs: "2, Switch, Keybord"

This may look like I've switched to the just opened tab (C), but this
is not true.

P. S. There are two things which can help in data validation:
1. Button to open "Raw Tab Usage Data" window
2. Auto refresh of this window