[ocaml-infra] ocaml.org on 1st page of google query for ocaml.

Sat Jan 19 15:09:04 GMT 2013

2013/1/19 Amir Chaudhry <amc79 at cam.ac.uk>

>
> On 19 Jan 2013, at 09:43, Sylvain Le Gall <sylvain at le-gall.net> wrote:
>
> > 2013/1/18 Amir Chaudhry <amc79 at cam.ac.uk>
> >> I'd be wary of drawing too many conclusion from this just yet.  There's
> no testing going on regarding intent so those could be anything from casual
> drive-by visitors to hard-core users.  All we can say is that 1/3rd of
> visitors to the ocaml.org site are using Windows.
> >
> > Well, guessing intent from stats is kind of hard. I mean, you CANNOT do
> that even after a long period of time. All these are just clues, you must
> make guess on the stats.
> >
> > I would be deeply surprised that we got a lot of "drive-by visitors"...
> You cannot really end up on the front page of ocaml.org just as you will
> end-up on a Facebook page. The ocaml.org website doesn't really have
> catch all words/sentences.
>
> Perhaps I shouldn't call them 'drive-by', so how about 'first-time,
> unique'? I didn't mean to suggest these are people who got there by
> accident, but that they were simply 'having a quick look'.  ~60% of the
> words that people put into google before arriving at the site are unknown
> to us ("Keyword not provided") so we don't necessarily know how people find
> their way there.
>
> Aside: I did some searching about this and found a blog post that was
> enlightening:
> http://www.practicalecommerce.com/articles/3798-Google-Analytics-Overcoming-Not-Provided-Keywords
>
>
OK, on "first time, unique" (better than drive-by in my mind). Although, I
would consider them as the target of of ocaml.org (i.e. a way to publicize
more OCaml). So having a lot of "first time, unique" is a good thing.

> >> Also bear in mind that stats like this might be skewed by population
> size.  It doesn't surprise me at all that there are a lot of US visitors.
>  Having said that, I don't know what Google Analytics does in the
> background to mitigate this (if at all).
> >
> > What do you want to mitigate ? If there are a lot of US visitors, you
> won't divide the number of US visitors…
>
> I don't want to mitigate.  I wasn't sure if we're seeing the raw numbers
> or the numbers after some behind-the-scenes google manipulations (which I
> doubt in this case). See also search trends [1], which indicates that over
> the last 90 days, France is doing the most searching for 'OCaml'.  I've no
> idea how google works this out and whether they account for overall search
> volume somehow.
>
> [1] http://www.google.com/trends/explore#q=ocaml
>
>
Why Google would change any numbers?

Anyway, I can tell you that some stuff are filtered out:
- browser without JavaScript
- robots

But, you expect this people to be filtered out.

Concerning France, I 100% agree on the fact that there is a problem here. I
don't have any clear explanation/hyptothesis on this topic.

> >
> >> Anecdote: One of the things a startup friend told me is that when your
> browser visits start matching the ratio of overall browser market share,
> then you can consider yourself as breaking out of the 'echo-chamber'.  This
> used to mean going from Firefox-heavy traffic to IE-heavy traffic (Chrome
> may have changed things since then).
> >
> > That probably stands for general purpose web sites -- which is not the
> case of ocaml.org.
>
> Why shouldn't a programming language website (ultimately) have a broad set
> of visitors? You've already pointed out the large number of Windows
> visitors so I don't think it's a stretch to consider this a general purpose
> website (in time).  Those figures are related to the OSes people browse
> from, which are not necessarily where they do their OCaml hacking.
>
> I'd be curious to know what the figures look like for Haskell.org (I've
> sent an email so will let folks know if/when I hear back).
>

;-)

That a nice wish. Let see what are the haskell number.

>
> > I would say that what is important is the number of abandonment
> ("sorties" in french) from the root page. In the case of "ocaml.org", 34%
> abandon from /, 21% goes to install, 14% goes to taste.html... I guess it
> shows that we have a lot of people interested enough to discover how to
> install OCaml... I think there is a clear intent to discover the OCaml
> language when going to ocaml.org.
>
> Curious to know where you're looking for these figures.  I can see
> different pages that state 34% exit rate from root and also 54% exit rate
> from root.  I don't see the ones that indicate e.g "21% go to install" so
> where should I be looking?
>
>
Here:
https://www.google.com/analytics/web/?hl=fr&pli=1#report/content-pages/a22552764w54925729p55893860/%3F_r.tabId%3Dnavigationsummary/

Contenu -> Toutes les pages -> Recapitulatif de navigation

There is an even better way to see it, but there is a problem that should
be fixed about the account configuration:
https://www.google.com/analytics/web/?hl=fr&pli=1#report/inpage/a22552764w54925729p55893860/

> Journeys from the root page are useful but be wary of how we interpret
> them.  e.g if it's a great page and 90% of the visitors find exactly the
> information they wanted and _leave_ then we have an exit rate of 90%.  If
> the page is terrible and 90% of people are frustrated and leave we _still_
> have an exit rate of 90%.  It's a silly example but I hope it makes the
> point.
>
> We should also expect people to land directly on pages from search results
> (not every journey starts from root).  "OCaml Install" as a search term has
> 3 INRIA pages and ocaml.org/install as 4th.  "ocaml tutorial" sends me to
> mirror.ocaml.org (top result), with ocaml.org/tutorial as 3rd.  As more
> content is tweaked and back-links created, I expect the relevant ocaml.orgpages to rank higher.  My point is that landing pages are just as important
> as how people navigate the site and (imho) the search terms are more
> indicative of what people were looking for.
>
>
I am not sure to understand what you expect but I am pretty sure you won't
find the answer to questions like "exit rate is high because people find
the info" or "exit rate is high because people don't like the page" in the
stats. You can run a poll, but you'll have another bias in the population
that reply.

So whatever, a high rank in Google is by itself an achievement and the
number of people that visit the site is high. This is good and I think the
ocaml.org team is achieving its goal.

> Best wishes,
> Amir
>
>
> >
> > On 18 Jan 2013, at 17:37, Sylvain Le Gall <sylvain at le-gall.net> wrote:
> >
> > > Just had a quick look at the stats, and there are some surprises....
> > >
> > > Top OS: Windows (33%), then Linux and Mac...
> > > Top social referrer: Twitter, then Stack Overflow and G+.
> > > Country: US (27%), France (12%)
> > >
> > > Here are the fact I am surprised about:
> > > - people seems to think that the OCaml community is UNIX centric...
> seems to be not that true (at least 1/3 are not using UNIX based system)
> > > - I was expecting to see reddit in the top social referrer and G+ far
> below.... this is not the case
> > > - I was expecting to see at least France in top position here.
> > >
> > > Overall I am surprised by this stats. I think we should take another
> look in 6 months.
> > >
> > >
> > >
> > > 2013/1/18 Sylvain Le Gall <sylvain at le-gall.net>
> > > BTW, I am planning to setup analytics as well to forge.(SOON TO BE).
> ocaml.org and planet.ocaml.org, maybe it makes sense to have all this
> stats in one place (i.e. the analytics account where ocaml.org is already
> hosted).
> > >
> > >
> > > 2013/1/18 Thomas Gazagnaire <thomas.gazagnaire at gmail.com>
> > > > Comes up 2nd for me, right after the wikipedia page (3rd if you
> count the Jane Street add at the top).
> > > > 4th-7th are INRIA sites, 8th is Jane Street's OCaml page, 9th is the
> OCaml Labs homepage and 10th is Planet OCaml.
> > > >
> > > > Also, google.com redirects me to google.co.uk (seems I can't choose
> anymore).
> > > >
> > > > Traffic stats would be interesting but even more so would be how
> people currently find their way to ocaml.org and which pages they land
> on.  I'm curious to know the ratio of direct visits (typing ocaml.orginto the browser) vs  people who follow links.  Of those that follow links,
> who are the referrers.
> > >
> > > We have set-up google analytics for ocaml.org, I'm happy to give
> access to the stats to anyone interested (I don't know if it's easy to
> export the data).
> > >
> > > Thomas
> > >
> > >
> > > >
> > > > ac
> > > >
> > > > On 18 Jan 2013, at 15:26, Sylvain Le Gall <sylvain at le-gall.net>
> wrote:
> > > >
> > > >> Hi,
> > > >>
> > > >> Just want to drop a mail to say that ocaml.org is now on first page
> > > >> http://www.google.com/#q=ocaml
> > > >>
> > > >> It is the 4th entry after caml.inria.fr and wikipedia.
> > > >>
> > > >> This is VERY good (never achieved this with the forge). It is the
> same for bing.
> > > >>
> > > >> Ashish do you have some Google Analytics setup to track the number
> of visitors ? I would be very interested to see the traffic.
> > > >>
> > > >> Regards
> > > >> Sylvain
> > > >> _______________________________________________
> > > >> Infrastructure mailing list
> > > >> Infrastructure at lists.ocaml.org
> > > >> http://lists.ocaml.org/listinfo/infrastructure
> > > >
> > > > _______________________________________________
> > > > Infrastructure mailing list
> > > > Infrastructure at lists.ocaml.org
> > > > http://lists.ocaml.org/listinfo/infrastructure
> > >
> > >
> > >
> >
> > _______________________________________________
> > Infrastructure mailing list
> > Infrastructure at lists.ocaml.org
> > http://lists.ocaml.org/listinfo/infrastructure
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ocaml.org/pipermail/infrastructure/attachments/20130119/8993bffb/attachment.html>