On 30 July 2014 19:53, Hannes Magnusson <hannes.magnus...@gmail.com> wrote:

> On Wed, Jul 30, 2014 at 6:20 AM, Mike Griffiths <mgriffi...@gmail.com>
> wrote:
> > On 30 July 2014 14:14, Peter Cowburn <petercowb...@gmail.com> wrote:
> >
> >>
> >>
> >>
> >> On 30 July 2014 09:34, Mike Griffiths <mgriffi...@gmail.com> wrote:
> >>
> >>>
> >>> On 30 July 2014 09:08, Peter Cowburn <petercowb...@gmail.com> wrote:
> >>>
> >>>>
> >>>>
> >>>>
> >>>> On 30 July 2014 08:08, Mike Griffiths <mgriffi...@gmail.com> wrote:
> >>>>
> >>>>> Hi all,
> >>>>>
> >>>>> I would like to broach the subject of introducing some sort of
> analytics
> >>>>> package, such as Google Analytics, to the PHP.net website and its
> sister
> >>>>> sites.
> >>>>>
> >>>>> I personally believe this would be of huge benefit to lots of us, as
> >>>>> currently there is little to no way (that I'm aware of) of viewing
> >>>>> useful
> >>>>> statistics and metrics on our visitors.
> >>>>>
> >>>>> Hannes and I spoke about this last night, and he is against the idea
> of
> >>>>> using a third party package to do this, and would rather write
> >>>>> something.
> >>>>>  Personally I think something like Analytics would be far more
> powerful
> >>>>> and
> >>>>> versatile than that, and will take a matter of minutes to get up and
> >>>>> running.  It's also pretty easy to share with anyone who wants to see
> >>>>> it.
> >>>>>
> >>>>> The kind of useful information I'd like to see, which is only what
> >>>>> comes to
> >>>>> mind initially, I'm sure there's far more once we delve deeper into
> it:
> >>>>> - The docs pages with the highest bounce rates, top x pages.  It's
> >>>>> likely
> >>>>> these don't have the right information, or it could be that it has
> the
> >>>>> best
> >>>>> information and that's why people leave quickly.
> >>>>> - Docs pages with the longest visit time.  Ties in with the above
> point,
> >>>>> and could highlight problems - users could be spending time reading
> the
> >>>>> notes to try and find what they need.
> >>>>> - Visibility of our most popular pages
> >>>>> - What are people searching for in search engines to find us?  Can we
> >>>>> identify common user typos and try to handle them better?
> >>>>> - What percentages of our users are on X browser version (this would
> >>>>> have
> >>>>> been valuable when we had our recent design change)
> >>>>> - Is there a news item gathering a lot of attention at the moment?
>  Can
> >>>>> we
> >>>>> identify what makes a news item popular vs something that no one is
> >>>>> particularly interested in?
> >>>>> - How much traffic do we get across all of the mirrors?  Is there a
> >>>>> particular mirror that is more popular than others?
>
>
> What is the endgoal here? Other then gaining statistics for your
> accounting department and sprinkling PKI arguments, what will we do
> with these numbers?
>

Not sure about the tone of that, but the end goal is to help people who
contribute spend their time one what's needed.  I thought knowing where our
traffic is most would help with that.


>
> Its all fun and cool and ego boosting to see huge visitor numbers, and
> it can create good laughs seeing how many use certain browsers or
> whatever.. but so?
>
>
> Seeing the most common search terms and most access doc page are
> indeed very useful and we should definitely bring that back.
> Philip used the search terms in the past to improve php.net/foobar
> redirects in the past.
>

OK, let's do it.


> The most commented-on-docpages are also used by the doc team to see
> what needs to be improved... That number is probably similar to the
> most accessed docpage.
>

How would someone see that data currently?


>
> I don't mind adding some metrics, but lets figureout what exactly we
> are going to use it for first. Thats the only way we can figureout
> which datapoints we need.
>
> Lets identify the problem before providing solutions.
>

Agreed.  I'm as open to suggestions as anyone.


>
>
>
>
> >>> Well I've had an email this morning from someone saying we used to have
> >>> GoSquared, and it was very useful and used often, but unfortunately it
> was
> >>> removed by someone who didn't use it.
> >>>
> >>>
> >> I don't know who emailed you, but it was indeed useful and was used
> >> often... by three people... for a few weeks.
> >>
>
> Is this one person who mailed you protected by attorney/client privilege?
> Thats an oddly anonymous statement.
>

No, but given I was emailed directly, I didn't bother naming.


>
> I removed GoSquare because it had served its purpose during the redesign.
> Well, thats not true. The reason I removed GoSquare was because it was
> so slow I cried myself to sleep.
> I also do not care for being tracked like this.
>

No one mentioned you.  I don't really care who removed it, it would be
useful to have back in my opinion, but no one wants you to cry yourself
asleep again, so let's shelve that.


>
>
> > Things like this probably get little use because they're not known by
> > everyone.  If there was something added into the VCS login, or whatever,
> > that made it a bit more visible then it will likely get more use.  I've
> > lost track of the number of times something has been suggested and we get
> > "Oh there's a tool for that, but no one uses it", or even worse "...it's
> > been removed" - I wonder how much of this is because people aren't aware
> of
> > their existence?
>
>
> People that work on things around here are pretty familiar with the
> availabilities..
>

This seems very closed, which is part of the problem.  Surely if there are
tools they should be made available to everyone who contributes.  Most
contributors aren't able to spend vasts quantities of time on things, so
these sorts of metrics could be invaluable.  I'm not sure a simple link is
asking too much...


>
>
>
>
> >>> I'd be happy if GoSquared was re-instated.  Google Analytics is the
> >>> market leader, so it's the example I used, but I don't mind what it is
> >>> that's used.
> >>>
> >>
> >> My main concern is introducing something that affects every single user
> >> visiting php.net, and on every single visit, for something that one or
> >> two people might look at once a month or less.
> >>
> >> Don't forget that we are already logging lots of information that could
> be
> >> collated and presented somewhere centrally (say, a new page on master)
> with
> >> zero impact on the site.
> >>
> >
> > Like I say - I don't mind how we get the data, so long as we can all get
> > it, in a usable format.  If it's all being collected already but not yet
> > presented, then let's crack on with that.  If not, let's discuss
> > alternatives.
>
>
> Excellent \o/
> To be able to present the data in "usable format" we need to identify
> what the data will be used for and the end goal.


Yes.


> Failing that, we should hire couple of experienced data analysts and
> get someone to staple the KPI reports :)
>
>
>
> -Hannes
>



-- 

Mike Griffiths - Blog <http://mike-griffiths.co.uk>
mgriffi...@gmail.com

Reply via email to