Daniel Buchner
unread,Dec 6, 2013, 5:31:27 PM12/6/13You do not have permission to delete messages in this group
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Will Bamberg, David Walsh, Eric Shepherd, Maris Fogels, mdn-drivers, John Karahalis, Holly Habstritt, Ali Spivak, Luke Crouch
To really instrument and test this, we'd need a few measures (IMO) - you
make the call if it is worth it:
1. Setup an A/B test with and without the TOCs
2. Track clicks of TOC links
3. Track clicks of sub-links within TOC'd content sections
4. Track time-on-page before navigation to a sub-link within a section
There's two things to assess once you have the data:
1. Is the TOC even engaged by users that often? If not, the point is
moot and it should go.
2. If the TOC is engaged often, you can then compare the time it takes
for a user to reach a content block and navigate to a new sub-linked
document without the TOC (using scroll and scan), to the time it takes for
users who engage the TOC to navigate to the same sub-linked document. If
the time it takes a user to complete the goal step is reduced noticeably in
the presence of a TOC, there's your user benefit - weigh that in your
decision accordingly.
Thoughts?
- Daniel
On Fri, Dec 6, 2013 at 2:17 PM, Will Bamberg <
wbam...@mozilla.com> wrote:
> I guess the assertion is that an indented ToC is more scannable, so it's
> easier to find the element you're looking for. I'm not sure how to build an
> A/B test for that. Maybe just, how often people click links in the ToC,
> since they find the link they are looking for there rather than scrolling
> through the page or giving up? How fast people click on the ToC after
> loading the page?
>
> Will
>
>
>
> On 2013-12-06 2:08 PM, Luke Crouch wrote:
>
>> I hear many calls to remove them, but if someone can give us a hypothesis
>> to test an expected value of keeping them, I'd be okay with having an A/B
>> test in-place at the time of launch and then quickly make the call
>> post-launch.
>>
>> If no-one cares enough to come up with an hypothesis, test, and value to
>> measure we can go by simple consensus or majority here.
>>
>> -L
>>
>>