I asked my first question about a month ago and checked in today
because it was expiring and I hadn't had any email re: a response. 
Unfortunately, I hadn't realized that comments don't trigger an email
... so had missed the fact that a follow-up question had been asked. 
Oops!  As a novice, I have no idea how to extend the time on my
original question (which has now closed) but I still really want a
response so I'm reposting - with an answer to that long-ago query - in
the hopes that I can get an answer.  [Apologies if I should have done
this differently, but I'm still figuring out the system.]

Thanks!

---------------------------------------------------------------
Here's the original question (see
http://answers.google.com/answers/threadview?id=770791)

I work for an organization that is trying to assess the reach of
reports that we publish.  We do this in many ways - media monitoring
and counting downloads from our web site are just two examples.  But
we would also like to understand how widely our reports are
referenced/cited on others' web pages (and be able to compare this to
similar reports by others).  All of the reports (ours and the
comparators) can be downloaded for free on the web as either PDF or
HTML files.

At first, we thought that this would be as simple as running a web
search using a major search engine for the title of the report and the
organization name and counting how many times it comes up.  But this
appears to produce quite inconsistent results for different
organizations or even ever-so-slightly different search terms.  (Note:
searching for links to the reports doesn't work since they tend to be
posted on a few web sites or with a few URLs).

Rather than scratching our heads for a very long time, I thought that
we should see if someone else has already figured this out.  Can
anyone help?  I would love a reference to a document/web page
describing effective tracking strategies or just advice on how to
accomplish this.

----------------------------------------------------------------

In answer to the subsequent question that was asked:

Here's an example of a document that I would be interested in
tracking: http://secure.cihi.ca/cihiweb/products/WaitTimesReport_06_e.pdf

We ran a search with the title (in quotes) and the organization name. 
This gives you an estimated # of pages with the reference with
duplicates suppressed.  However, running the same search just a couple
of hours apart produces quite different estimated #s of pages.  (More
that I would have thought could be explained by changes in actual
references alone).  Bottom line, there's undoubtedly a better way of
doing this.  I'm not necessarily looking for help with running this
specific search, although I would never turn that down.  Ideally, I'd
like a link or reference to a document that describes robust
ways of counting web links/mentions.

Thanks!

---

Request for Question Clarification by pafalafa-ga on 04 Nov 2006 09:17 PST

hlppls-ga,


Thanks for your persistence in reposting your question.

Unfortunately, the GA email notifications aren't working quite they
way they should, so it's necessary to check back with your question
every now and then to see what's been posted.

I would have said, ordinarily, that the answer to your question is
quite straightforward.  Just conduct a Google search like this:



[ "waiting for health care in canada"  "what we know and what we" ]


which returns about 56 'hits' on your report.


I have repeated this search on different computers, and over the span
of a few hours, and the results stay consistent.  I always get a count
along the lines of:


Results 1 - 56 of about 345 


You may want to cut and paste the above terms (sans brackets) and
conduct the search yourself, to see what results you get.

Without knowing exactly what you searched on in the past, I can't
explain the variability you've seen.

However, there are numerous reasons for constructing the search as I
did (rather than, say, as one long quote on the full title of the
report).  Even very minor changes in how a search is constructed can
have a large impact on the number of results.

I'd be glad to explain my search strategy in detail as an answer to
your question.  That way, you'd be able to search on any report of
interest in a consistent fashion, and with (hopefully) reliable
results.

Also, it's possible to set up an automated search of web and/or news
sources, so that everytime a new mention of your report appears,
you'll receive an email alert.  I can include that in an answer as
well, if this type of alert service is of interest.


Let me know if this seems a reasonable approach to answering your question.

pafalafa-ga

---

Clarification of Question by hlppls-ga on 04 Nov 2006 20:25 PST

Thanks very much for this speedy response.  I just tried the search
you suggested and got a similar response, although when I clicked
through the pages there weren't 345 items (even when I clicked on the
show duplicates option).  Perhaps this reflects the "about" caveat in
the count?

In any case, yes, I would very much appreciate the detailed search
strategy and tips on setting up the automated referral.

Many thanks.

hlppls-ga,


Thanks for getting back to me.  I was glad to hear the search I
constructed returned pretty much the same results for you as it did
for me.  That's a good sign.


As for the search itself, let's look it over in a bit more detail:



[ "waiting for health care in canada"  "what we know and what we" ]



Basically, what I've done here is to break a long title into two
discrete phrases, and omitted a potentially problematic term.


You could, of course, simply search on the full title in quotes:


[ "Waiting for Health Care in Canada: What We Know and What we Don't Know" ]


and this will return many of the same results, and give you a good
quick overview of what's out there.


However, search engines sometimes have problems with lengthy phrases,
probably having to do with the fact that long phrases often include a
line break that complicates recognition. For that reason, I recommend
breaking out a short, readily-recognized phrase for searching.


However, a search for just [ "Waiting for Health Care in Canada" ]
returns some spurious results, from other reports with a similar
title, or simply from casual use of the same phrase.  So adding a
second term to the search helps pinpoint your report.

I added [ "What We Know and What we" ] because I've noticed in the
past that some search engines seem to have problems with apostrophe'd
terms, especially in the context of exact phrases (ie, when quotes are
used).  To avoid any possible problems with this, I dropped the
"...don't know" part of the title.


If you follow those guidelines, you should be able to put together a
reliable search on any title of interest.

One more trick to be aware of...You can, if you want, eliminate
results from your own website, if you only want to count hits from
other mentions of your report.  This Google search:


[ "waiting for health care in canada"  "what we know and what we" -cihi.ca ]


returns 41 (rather than 54) hits.  The last term instructs Google to
ignore results from webpages ending in cihi.ca.


===============


I also mentioned the automated search-and-notification feature, which
is a great little tool.  To set them up, simply head to Google Alerts:


://www.google.com/alerts


and enter the search terms you are looking for in the search box
(you'll have to set up a separate Alert for each search).

Next, choose the type and frequency of search from the pull down
menus.  You can search the web, or news sources, or blogs, or groups,
or all of the above.

The alert will send you an email every time new material appears with
your search terms.

You'll probably get an initial email with all the results currently
out there, i.e., a link to all 54 hits to your Waiting for Health Care
report.  But after the initial alert, you only get new alerts as new
material appears.


I hope I've provided some useful tools, and a clear explanation of how
to put them to good use.  But if there's anything else I can do for
you, just ask, and I'm at your service.

All the best,

paf

---

Request for Answer Clarification by hlppls-ga on 10 Nov 2006 20:10 PST

Many thanks for all your help.  I've been trying out the tips already!

(By the way, sorry for not officially rating this answer highly but
that function doesn't seem to be working properly from my computer.)

---

Clarification of Answer by pafalafa-ga on 10 Nov 2006 20:13 PST

Thanks for letting me know...hope everything goes smoothly, but as
always, if there's something else I can do for you, just let me know.

paf

hlppls-ga,
"..as a novice". It's not your fault. The email notification system is
not working at the moment. You have to keep looking at your question
to see if it has been answered or commented on.
Hopefully one of my fellow researchers will be able to help you on this question.