Search Behavior Study: How does the user's search behavior change?

NEW
Unassigned

Status

Mozilla Labs Graveyard
Test Pilot Data Requests
7 years ago
2 years ago

People

(Reporter: Mardak, Unassigned)

Tracking

Details

(Whiteboard: Analysis)

(Reporter)

Description

7 years ago
The main question we're looking to answer is how does the user's search behavior change with AwesomeBar HD installed. To measure this, the study detects when the user visits a search result page from a list of 42 pre-defined search engines.

These events show up with a event name like "search searchbar" with a value "search Google" meaning a search was done on Google from the search bar. There are 6 kinds of searches that are tracked:

search searchbar: search from the search bar
search urlbar: search from the location bar e.g., default search provider or keyword search
search awesomebar: search from the add-on
search home: search from about:home
search reload: user reloaded a search result
search link: search from web content (either linked or in-page search box)

I would assume just counting up the number of "search <type>" entries for each user can give an estimate of search behavior change -- mainly looking at awesomebar search quantity compared to searchbar + urlbar.

The study randomly configures users to be in 1 of 4 buckets, and this state is stored as part of the submission metadata "testpilot".
0: ui shows 14 category text links
1: ui shows 14 category icons
2: ui shows "search the web" link
3: ui shows google icon

Note: the "testpilot" metadata value may be missing if the user chose not to install the add-on, so there's actually 5 buckets of data to investigate. The study also tracks "addon" "disabled", "uninstalled" events if the user chooses to disable or get rid of the add-on, which may result in some "search awesomebar" events while still ending up with a missing "testpilot" metadata value.

Updated

7 years ago
Whiteboard: Analysis
(Assignee)

Updated

2 years ago
Product: Mozilla Labs → Mozilla Labs Graveyard
You need to log in before you can comment on or make changes to this bug.