On 1/9/13 2:53 PM, Rob Weir wrote:
> On Wed, Jan 2, 2013 at 10:34 AM, Rob Weir <[email protected]> wrote:
>> On Wed, Jan 2, 2013 at 10:26 AM, Samer Mansour <[email protected]> wrote:
>>> Rob,
>>>
>>> How should I send the zip to you? I finished four placements, we can create
>>> more variations on the wording, it's a simple change.
>>>
>>
>> If it is less than 25MB, you can mail it to me directly.
>>
> 
> A quick update for the team:  Samer and I worked offline to get his
> design variations for the download page integrated.  So the content
> experiment is now up and running and we'll know in a couple of weeks
> which version gives the greatest rate of sharing via Facebook, Google+
> and Twitter.
> 
> I'd rather not make public the exact URL's for the variations, since
> visiting those pages directly could distort the statistics being
> gathered, but if you visit http://www.openoffice.org/download you will
> probably be shown one of the variations with the social sharing icons.
> 
> Currently we're seeing a "conversion rate" of around 0.1%, meaning
> 0.1% of visitors to the page share their excitement with their social
> network.  Since we get around 100K downloads/day, that amounts to
> around 100 shares per day.  And if each person has 100 people in their
> social network, that gives us 10,000 impressions per day, which is
> some nice free advertising.  To put that in perspective, 10,000 is
> greater than the size of our Twitter, Facebook and Google+ subscriber
> lists combined.  Plus, the nature of the sharing is that the post will
> come from someone familiar, since it is shared within a social
> network.  So the impact is that much greater.
> 
> The point of this "content experiment" is to find the design that
> gives the highest conversion rate.  This is the first step.  We might
> do additional variations later as well.
> 
> In any case, thanks to Samer for his contributions over the holidays!
> 
> Regards,
> 
> -Rob
> 
> 
>> Thanks!
>>
>> -Rob
>>
>>
>>
>>> Samer
>>>
>>> On Thu, Dec 20, 2012 at 10:59 PM, Samer Mansour <[email protected]> wrote:
>>>
>>>> Rob,
>>>>
>>>> My apologies holiday season is hectic I haven't had a chance to finish off
>>>> the variations. I have an early morning otherwise I would do it now. I will
>>>> aim for Monday to have it done. Next week work will also be very quite so I
>>>> will start building the 'like us' experiment pages as well then.
>>>>
>>>> Sam
>>>>
>>>>
>>>> On Wednesday, December 19, 2012, Rob Weir wrote:
>>>>
>>>>> On Wed, Dec 19, 2012 at 11:18 AM, Samer Mansour <[email protected]>
>>>>> wrote:
>>>>>> Ok I had the impression that it had to all be done before we start the
>>>>>> experiment to give them a good benchmark.
>>>>>
>>>>>
>>>>> Yes, we should have all of the variations for a given experiment
>>>>> before we start that experiment.  I don't think Google let's us add
>>>>> new variations without restarting the experiment.
>>>>>
>>>>>> I have 4 done as of last night, I'm almost done a more elegant 5th one.
>>>>> The
>>>>>> separate section idea, but all the existing backgrounds have arrows
>>>>>> suggesting "download" or "get". I created its own "share" background
>>>>>> http://postimage.org/image/9brst32ex/ which we can clean up more if we
>>>>>> choose to go with it. I changed the Hue and replaced the images with the
>>>>>> vector art of Share I created in Inkscape.
>>>>>>
>>>>>
>>>>> Cool.
>>>>>
>>>>>> I have some errands after work, I might be able to churn out a few more
>>>>>> designs.  I will also name them correctly before handing them off to
>>>>> you.
>>>>>> I just have to figure out paths and changes into a document (or here),
>>>>> so
>>>>>> that its clear what is to be changed/deployed and in what path.
>>>>>>
>>>>>
>>>>> Assume the files will end up in www.openoffice.org/download, alongside
>>>>> the current download/index.html page.
>>>>>
>>>>> -Rob
>>>>>
>>>>>
>>>>>>
>>>>>> On Wed, Dec 19, 2012 at 9:24 AM, Rob Weir <[email protected]> wrote:
>>>>>>>
>>>>>>> On Fri, Dec 14, 2012 at 10:54 PM, Samer Mansour <[email protected]>
>>>>>>> wrote:
>>>>>>>> Oh, I'm sooo excited. I can help with coding the different pages.  We
>>>>>>>> could
>>>>>>>> do the like and share experiments separately so they are not
>>>>> affecting
>>>>>>>> each
>>>>>>>> others results.
>>>>>>>>
>>>>>>>> Lets start with a sharing experiment.
>>>>>>>>
>>>>>>>> We could make the labels "test1-footer" "test2-header"
>>>>> "test3-searchbar"
>>>>>>>> as
>>>>>>>> opposed to "facebook", "google", "twitter". We don't care which
>>>>> network
>>>>>>>> they pick, its the placement/wording/style whether they
>>>>> see/convenient
>>>>>>>> to
>>>>>>>> click it.  The metric would sum the three buttons for each test,
>>>>> under
>>>>>>>> the
>>>>>>>> same label.
>>>>>>>>
>>>>>>>> We can all make friendly bets, like picking which horse will win a
>>>>> race.
>>>>>>>> Once we collectively come up with 9, we'll start a separate thread
>>>>> where
>>>>>>>> people can give their prediction.
>>>>>>>>
>>>>>>>> In the spirit of JFDI, I will make the variations you suggested Rob
>>>>>>>> right
>>>>>>>> now.  I will wait until Tuesday Night for anyone else variation
>>>>>>>> suggestion
>>>>>>>> on this thread.
>>>>>>>>
>>>>>>>
>>>>>>> OK.  I've made some changes to the main website to accommodate the
>>>>>>> experiment.  The issues was that the website logic would automatically
>>>>>>> try to apply the site template to our test pages, wiping them out.
>>>>>>> With my change any page that ends in "-passthru.html" will be
>>>>>>> unprocessed by the site template and will be displayed as-is.  So
>>>>>>> we'll need to name our test bases something like
>>>>>>> "social-variation1-passthru.html", etc.
>>>>>>>
>>>>>>> Let me know when you have some variations ready to test and I can get
>>>>>>> them provisioned onto the website.
>>>>>>>
>>>>>>> Also, we should avoid putting the actual test file names and links
>>>>>>> onto the mailing list.  That would encourage list subscribers to load
>>>>>>> the pages and since our behavior differs from the random visitor's
>>>>>>> behavior this can distort the experiment,

thanks for sharing this, very interesting

Juergen

>>>>>>>
>>>>>>> -Rob
>>>>>>>
>>>>>>>>
>>>>>>>> On Fri, Dec 14, 2012 at 11:12 AM, Rob Weir <[email protected]>
>>>>> wrote:
>>>>>>>>
>>>>>>>>> On Fri, Dec 14, 2012 at 1:24 AM, Samer Mansour <[email protected]>
>>>>>>>>> wrote:
>>>>>>>>>> 1. Progress on "Download Share" - Smaller images and text,
>>>>>>>>>> anti-aliased
>>>>>>>>>> using inkscape.
>>>>>>>>>> http://www.winsor.ca/samer/aootest2.htm
>>>>>>>>>>
>>>>>>>>>> 2. Progress made on possible template placements for "Follow
>>>>>>>>>> OpenOffice"
>>>>>>>>>> http://www.winsor.ca/samer/aootest3.htm - Left Nav Links & Header
>>>>>>>>>> Links
>>>>>>>>>> http://www.winsor.ca/samer/aootest4.htm - Title Nav Links &
>>>>> Footer
>>>>>>>>>> Links
>>>>>>>>>>
>>>>>>>>>> In 2. should we eliminate some placements? Or should we experiment
>>>>>>>>>> all 4
>>>>>>>>> in
>>>>>>>>>> separate files?
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> As I understand it each Content Experiment needs to test one thing,
>>>>>>>>> have one goal that we measure.  For example, we could measure % of
>>>>>>>>> visits that generate a page share via a social network.
>>>>>>>>>
>>>>>>>>> So it looks like we want to have at least two different content
>>>>>>>>> experiments:
>>>>>>>>>
>>>>>>>>> 1) On the download page, to test sharing.
>>>>>>
>>>>
>>>>

Reply via email to