Some keyword analysis for the website

[Rob Weir <ro...@apache.org>]

I was interested in how our website was performing against some common
search queries.  In particular I was thinking of users searching the
web in three general categories;

1) Users who wanted to do something immediately related to OpenOffice,
like download it, report a bug, solve a problem.

2) Users who were investigating.  They have heard about something
called 'OpenOffice" but wanted to learn more.

3) Users who never heard of OpenOffice but are looking for something a
free or open source office suite.

In general, with English-language queries, we did very well.  Some
areas did poorly, but in general we did very good.

Results are here in the wiki:

https://cwiki.apache.org/confluence/display/OOOUSERS/Keyword+Performance

Why do we care?  Because more visitors equals more users.  And more
happy users means more recommendations.  And more users means more
potential small contributions to the project, like bug reports.  And
some small portion of those contributors may get even more involved.

I see it as a pyramid, where the base is users, and a subset of them
become casual contributors, and over time subset of them become
committers.    One way to grow the project we grow the base of the
pyramid.

** I could use help, if anyone can spare an hour **

These results are for English, per the search results that Google
gives to Americans.  I'd like to check the results in a few other
popular languages, like German, French, Spanish, Italian, Japanese,
Russian, Dutch, etc.  Since most of our website is in English, it is
quite possible that some of these queries might do very poorly.

So if anyone can help, please add a new column to the tables and enter
the results you get in Google.

Based on what we find, we can take actions, such as:

-- using some alternate keywords on some webpages

-- updating a prominent, but out of date page

-- adding a new page to the website to address a specific common query

- translate a good English page to create an NL equivalent.

Thanks!

-Rob

Re: Some keyword analysis for the website

[Rob Weir <ro...@apache.org>]

On Thu, Jul 5, 2012 at 5:20 PM, Louis Suárez-Potts <lu...@gmail.com> wrote:
>
> On 2012-07-05, at 16:00 , Rob Weir wrote:
>
>>
>> One other thing that has evolved since the early 2000's.  The website
>> itself is considered to be very important according to Google, with a
>> "PageRank" of 8.  So if we get the right content on the website, it is
>> magnified by this high PageRank.
>>
>> -Rob
>
>
> Thus do we involve ourselves in SEO?
>

You could call it that, certainly.   Or you could call it "web site
usability".

More people come to the website via google searches than come from
typing in "www.openoffice.org" directly, or follow links to us from
other websites.  In fact over half of our traffic comes from search.
So a large part of making the website work well for users is to make
it work well for users who come via search queries.

> The only problem we used to have with OOo was that, for purely fiendish reasons, the last thing we'd want would come first—paid-for links on the right, but also top results that often had little to do with what we wanted people to find.
>

Were these unwanted pages on the OOo website?  Or external sites?

> Around 2004 or so a donor gave us 50K for Google AdWords. I spent a lot of time creating these, putting them up, monitoring, etc. The result, as I reported then was, again, hardly noticeable.
>

Let's see where things stand today.  From my English language test, it
looks very good.  We're placing #1 on most of the relevant queries.
We're #2 on a few, but that is still great.   The ones I'd be more
concerned with are the ones where we are not even on the first page.
So a most urgent case would be where the English query is #1 and the
translated query is not in the top 10.

> What sold among our primary audience then (note: "then")—and these were Windows users more or less unfamiliar with open source and just wanting a free alternative to MS Office—was word of mouth news that this a) existed and b) worked.
>

No doubt this is still influential.

> What caused spikes in downloads, I noted, was news reportage in larger (read more) venues. For that reason, I kept a kind of spiky tone to press reports, etc.: got in the news. That came to a halt with corporate evolutions, but while I was more independent, it worked wonders, and put OOo in the news and resulted in more downloads more effectively than tweaking search terms.
>

This is not an either/or question.  We should get good news stories
and ensure the website is optimized.

> One thing now to consider: where do our visitors come from? Say, arbitrarily speaking, from Brazil. In that case, with fisl13 around the corner—and I'm presenting there on AOO(i), among other things—It would be good to start a campaign on AOO in Brazil (for Brazil by Brazil kind of thing)?
>

Of website visitors, by country, Brazil is #16.  (Top 5 are US,
Germany, Italy, Japan, France).

Within Brazil, the top visits come from:

1.	Sao Paulo
2.	Rio de Janeiro
3.	Rio Grande do Sul
4.	Minas Gerais
5.	Parana
6.	Santa Catarina
7.	Federal District
8.	Bahia
9.	Ceara
10.	Goias


But I agree that it would be good to get AOO known as a brand there.
They previously knew BrOffice more, right?

-Rob

> louis
>
> PS fisl13 is huge. I'll be giving a presentation on AOO and ODF and the necessity of these tools of free production for the present and future.

Re: Some keyword analysis for the website

[Louis Suárez-Potts <lu...@gmail.com>]

On 2012-07-05, at 16:00 , Rob Weir wrote:

> 
> One other thing that has evolved since the early 2000's.  The website
> itself is considered to be very important according to Google, with a
> "PageRank" of 8.  So if we get the right content on the website, it is
> magnified by this high PageRank.
> 
> -Rob


Thus do we involve ourselves in SEO?

The only problem we used to have with OOo was that, for purely fiendish reasons, the last thing we'd want would come first—paid-for links on the right, but also top results that often had little to do with what we wanted people to find.

Around 2004 or so a donor gave us 50K for Google AdWords. I spent a lot of time creating these, putting them up, monitoring, etc. The result, as I reported then was, again, hardly noticeable. 

What sold among our primary audience then (note: "then")—and these were Windows users more or less unfamiliar with open source and just wanting a free alternative to MS Office—was word of mouth news that this a) existed and b) worked. 

What caused spikes in downloads, I noted, was news reportage in larger (read more) venues. For that reason, I kept a kind of spiky tone to press reports, etc.: got in the news. That came to a halt with corporate evolutions, but while I was more independent, it worked wonders, and put OOo in the news and resulted in more downloads more effectively than tweaking search terms.

One thing now to consider: where do our visitors come from? Say, arbitrarily speaking, from Brazil. In that case, with fisl13 around the corner—and I'm presenting there on AOO(i), among other things—It would be good to start a campaign on AOO in Brazil (for Brazil by Brazil kind of thing)?

louis

PS fisl13 is huge. I'll be giving a presentation on AOO and ODF and the necessity of these tools of free production for the present and future. 

Re: Some keyword analysis for the website

[Rob Weir <ro...@apache.org>]

On Thu, Jul 5, 2012 at 3:17 PM, Louis Suárez-Potts <lu...@gmail.com> wrote:
> Hi,
>
> On 2012-07-05, at 15:02 , Rob Weir wrote:
>
>> I was interested in how our website was performing against some common
>> search queries.  In particular I was thinking of users searching the
>> web in three general categories;
>>
>
> Interesting. I did this thing back in the early 2000s and altered the keywords to be more popular. But it had minimal effect—then. The story *then* was that browsers had grown more sophisticated and simply altering keywords (or creating an endless stream of popular keywords in invisible ink) had, after 2001, zero effect.
>

Cool.  Feel free to add additional queries that you think are important.


> But times seemed to have changed?
>

Where your webpage ranks for a given user query depends on two things:

1) The relevancy of your page to the query

2) The importance of the page

Google is constantly tinkering with their algorithms, but the above
two factors remain.   Google is much smarter now about rejecting
"keyword stuffing" techniques like you used to see back then.  We
don't want to do that.  We shouldn't try to fake a relevant page.

But let's suppose that we test and find out that we place #1 for the
query "OpenOffice download"  (translated) except for Russian.  Then
that suggest that we need to add or update a Russian NL page and make
sure it has a link to download page.   So it is more about ensuring
that our website has the right relevant content.  And one sign that we
don't have that relevant content is when what we think is a relevant
query does not return the page we thought it should.

Example:  the search for "openoffice hang", which is a common query
for someone trying to debug a technical problem, returns as the
highest page some instructions for doing a "hanging indent".   This
should not be hard to fix.

We've had a huge website migration, as well as a new release.  I've
found broken stuff just by poking around.  I think it would be worth
doing a more methodical investigation to make sure common user queries
are finding us when they should.

One other thing that has evolved since the early 2000's.  The website
itself is considered to be very important according to Google, with a
"PageRank" of 8.  So if we get the right content on the website, it is
magnified by this high PageRank.

-Rob


> Louis
>

Re: Some keyword analysis for the website

[Louis Suárez-Potts <lu...@gmail.com>]

Hi,

On 2012-07-05, at 15:02 , Rob Weir wrote:

> I was interested in how our website was performing against some common
> search queries.  In particular I was thinking of users searching the
> web in three general categories;
> 

Interesting. I did this thing back in the early 2000s and altered the keywords to be more popular. But it had minimal effect—then. The story *then* was that browsers had grown more sophisticated and simply altering keywords (or creating an endless stream of popular keywords in invisible ink) had, after 2001, zero effect.

But times seemed to have changed?

Louis

Re: Some keyword analysis for the website

[Rob Weir <ro...@apache.org>]

On Sun, Jul 15, 2012 at 6:39 PM, Andrea Pescetti <pe...@apache.org> wrote:
> On 05/07/2012 Rob Weir wrote:
>>
>> https://cwiki.apache.org/confluence/display/OOOUSERS/Keyword+Performance
>>
>> These results are for English, per the search results that Google
>> gives to Americans.  I'd like to check the results in a few other
>> popular languages
>
>
> I just added Italian results, compiled by volunteers. We perform rather
> well, but in some cases the first result, while still belonging to
> openoffice.org, is not the best one.
>

Thanks, Andrea!

I've added a column to the wiki page, listing how many search queries
Google receives for each.  I think this will help is prioritize.  Some
get millions of hits, others only a few hundred.

-Rob

> Regards,
>   Andrea.

Re: Some keyword analysis for the website

[Andrea Pescetti <pe...@apache.org>]

On 05/07/2012 Rob Weir wrote:
> https://cwiki.apache.org/confluence/display/OOOUSERS/Keyword+Performance
> These results are for English, per the search results that Google
> gives to Americans.  I'd like to check the results in a few other
> popular languages

I just added Italian results, compiled by volunteers. We perform rather 
well, but in some cases the first result, while still belonging to 
openoffice.org, is not the best one.

Regards,
   Andrea.